In this action, nine associations of construction contractors challenge on equal protection grounds a City of Philadelphia ordinance creating preferences in City contracting for businesses owned by racial and ethnic minorities, women, and handicapped persons. The district court granted summary judgment to the Contractors, holding they had standing to bring this lawsuit and invalidating the Ordinance in all respects. Contractors Association v. City of Philadelphia, 735 F. Supp. 1274 (E.D.Pa.1990). In an earlier opinion, we affirmed the district court's ruling on standing but vacated summary judgment on the merits because the City had outstanding discovery requests. Contractors Association v. City of Philadelphia, 945 F.2d 1260 (3d Cir.1991). On remand after discovery, the district court again entered summary judgment for the Contractors. We will affirm in part, vacate in part, and reverse in part.

In 1982, the Philadelphia City Council enacted an ordinance to increase participation in City contracts by minority-owned and women-owned businesses. Phila.Code Sec. 17-500. In its present form,1 the Ordinance establishes "goals" for the participation of "disadvantaged business enterprises." Sec. 17-503. "Disadvantaged business enterprises" (DBEs) are defined as those enterprises at least 51 percent owned by "socially and economically disadvantaged individuals," defined in turn as:those 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.

Sec. 17-501(11). The Ordinance further provides that racial minorities and women are rebuttably presumed to be socially and economically disadvantaged individuals, Sec. 17-501(11)(a), but that a business which has received more than $5 million in City contracts, even if owned by such an individual, is rebuttably presumed not to be a DBE, Sec. 17-501(10).

The Ordinance sets goals for participation of DBEs in city contracts: 15 percent for minority-owned businesses, 10 percent for women-owned businesses, and 2 percent for businesses owned by handicapped persons. Sec. 17-503(1). The Ordinance applies to all City contracts, which are divided into three types--vending, construction, and personal and professional services. Sec. 17-501(6). The percentage goals relate to the total dollar amounts of City contracts and are calculated separately for each category of contracts and each City agency. Sec. 17-503(1).

To implement the program, the Ordinance established a Minority Business Enterprise Council and authorized it to promulgate regulations to ensure the goals are met by city agencies in awarding prime contracts and by private contractors in awarding subcontracts. Sec. 17-504(2)(e), (f), (i). The Ordinance specifies that, in developing regulations, the Council must consider: including DBEs on solicitation lists, assuring DBEs are solicited whenever they are potential contractors, structuring contract requirements to permit maximum participation by DBEs, and "investigating and making recommendations concerning the use of the Sheltered Market process, under which contracts would be set-aside so that only DBEs could bid for them." Sec. 17-504(2)(f). The regulations provide that "the contractor's efforts to meet [the] goals shall be considered an element of responsiveness to the bid," and require each contractor to submit a "Schedule for Participation" of DBEs in the contract at issue or to request a waiver if the contractor is unable to meet the goals after a good faith effort. Regulations Sec. 6.1.

The Ordinance also directed the Council to (1) develop a certification procedure for DBEs to prevent fraudulent or "front" DBEs from abusing the program, Sec. 17-504(2)(a); (2) grant exemptions for individual contracts or classes of contracts where there is "an insufficient number of DBEs ... to ensure adequate competition and an expectation of reasonable prices on bids," Sec. 17-505(1); (3) grant waivers to contractors who are unable to meet the percentage goals after a good faith effort, as determined by the Council, Sec. 17-505(3); (4) "recommend contractual language which provides that compliance with DBE participation requirements is material to the City contract," Sec. 17-504(2)(h); and (5) develop and recommend remedies, including but not limited to, termination of the contract in the event a contractor fails to comply with the program, Sec. 17-506(a).

On April 14, 1989, nine contractors associations brought suit in the Eastern District of Pennsylvania against the City of Philadelphia and two city officials, challenging the Ordinance as a facial violation of the Equal Protection Clause of the Fourteenth Amendment.2 United Minority Enterprise Associates, Inc. (UMEA) intervened as a defendant. After the City moved for judgment on the pleadings contending the Contractors lacked standing, the Contractors moved for summary judgment on the merits.

The district court granted the Contractors' motion. It ruled the Contractors had standing, based on affidavits of individual association members alleging they had been denied contracts for failure to meet the DBE goals despite being low bidders. 735 F. Supp. at 1283 & n. 3. Turning to the merits of the Contractors' equal protection claim, the district court held that City of Richmond v. J.A. Croson Co., 488 U.S. 469, 109 S. Ct. 706, 102 L. Ed. 2d 854 (1989), required it to apply the strict scrutiny standard to review the sections of the Ordinance creating a preference for minority-owned businesses. Under that standard, a law will be invalidated if it is not "narrowly tailored" to a "compelling government interest." Wygant v. Jackson Board of Educ., 476 U.S. 267, 274, 106 S. Ct. 1842, 1847, 90 L. Ed. 2d 260 (1986). Applying Croson, the district court struck down the Ordinance because the City had failed to adduce sufficiently specific evidence of past racial discrimination against minority construction contractors in Philadelphia to establish a "compelling government interest." 735 F. Supp. at 1295-98. The court also held the Ordinance was not "narrowly tailored," emphasizing the City had not considered using race-neutral means to increase minority participation in City contracting and had failed to articulate a rationale for choosing 15 percent as the goal for minority participation. Id. at 1298-99. The court held the Ordinance's preferences for businesses owned by women and handicapped persons were similarly invalid under the less rigorous intermediate scrutiny and rational basis standards of review. Id. at 1299-1309.

On appeal, we affirmed the district court's ruling on standing but vacated its judgment on the merits as premature because the Contractors had not responded to certain discovery requests at the time the court ruled. 945 F.2d 1260 (3d Cir.1991). We remanded so discovery could be completed and explicitly reserved judgment on the merits. Id. at 1268. On remand, all parties moved for summary judgment, and the district court reaffirmed its prior decision, holding discovery had not produced sufficient evidence of discrimination in the Philadelphia construction industry against businesses owned by racial minorities, women, and handicapped persons to withstand summary judgment. The City and UMEA appeal.3

This appeal presents three sets of questions: whether and to what extent the Contractors have standing to challenge the Ordinance, which standards of equal protection review govern the different sections of the Ordinance, and whether these standards justify invalidation of the Ordinance in whole or in part. We will address these issues in turn.

Our standing inquiry has two parts: whether the Contractors have standing to challenge the Ordinance at all, and if so, whether they have standing to challenge all or just part of the Ordinance. Prior to this appeal, the parties contested only the first of these issues. In our earlier opinion, we affirmed the district court's holding that four of the nine associations had standing to challenge the Ordinance because the affidavits submitted by their members alleged injury with sufficient particularity, and because the associations satisfied the standards for associational standing. 945 F.2d at 1264-66.

The Supreme Court has since confirmed that construction contractors have standing to challenge a minority preference ordinance upon a showing they are "able and ready to bid on contracts [subject to the ordinance] and that a discriminatory policy prevents [them] from doing so on an equal basis." Northeastern Fla. Chapter of the Associated Gen. Contractors of America v. City of Jacksonville, --- U.S. ----, ----, 113 S. Ct. 2297, 2303, 124 L. Ed. 2d 586 (1993). Because the affidavits submitted to the district court establish the Contractors were able and ready to bid on construction contracts but could not do so for failure to meet the DBE percentage requirements, they have standing to challenge the sections of the Ordinance covering construction contracts.4

We next consider whether the Contractors have standing to challenge the entire Ordinance or only the provisions relating to construction contracts. Because standing seeks to ensure a party has a "personal stake in the controversy," Harris v. McRae, 448 U.S. 297, 320, 100 S. Ct. 2671, 2690, 65 L. Ed. 2d 784 (1980), courts typically only allow a party to raise his own rights rather than the rights of others, Broadrick v. Oklahoma, 413 U.S. 601, 613, 93 S. Ct. 2908, 2916, 37 L. Ed. 2d 830 (1973). This doctrine sensibly enables a court to "avoid[ ] ... adjudication of rights which those not before the court may not wish to assert, and assur[es] that the most effective advocate of the rights at issue is present to champion them," Duke Power Co. v. Carolina Environ. Study Group, 438 U.S. 59, 80, 98 S. Ct. 2620, 2634, 57 L. Ed. 2d 595 (1978). Allowing parties only to litigate their own rights is especially important in constitutional actions because the Supreme Court cautions "never to anticipate a question of constitutional law in advance of the necessity of deciding it [and] never to formulate a rule of constitutional law broader than is required by the precise facts to which it is to be applied." United States v. Raines, 362 U.S. 17, 21, 80 S. Ct. 519, 522, 4 L. Ed. 2d 524 (1960) (quoting Liverpool, New York & Philadelphia S.S. Co. v. Commrs. of Emigration, 113 U.S. 33, 39, 5 S. Ct. 352, 355, 28 L. Ed. 899 (1885)).

Courts considering constitutional challenges to statutes often analyze standing problems in terms of the severability doctrine. Under this principle, when a court determines the legislature intended the challenged sections of a statute to operate independently of the unchallenged sections and finds these sections can so operate, it will consider only the challenged sections, leaving the remainder of the statute intact. SeeUnited States v. Raines, 362 U.S. at 23, 80 S.Ct. at 523-24; 13 Charles A. Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice and Procedure Sec. 3531.9, at 585 (2d ed. 1984). Severing statutes to limit standing promotes the twin goals of avoiding unnecessary constitutional adjudication and sharpening the presentation of the issues. Singleton v. Wulff, 428 U.S. 106, 113-14, 96 S. Ct. 2868, 2873-74, 49 L. Ed. 2d 826 (1976).5

The severability doctrine governs whether the Contractors have standing to challenge the entire Ordinance, or just those provisions of the Ordinance affecting the construction industry. As we have noted, the Contractors only have a personal interest in obtaining construction contracts because these are the only types of contracts they are "ready and able" to bid on within the meaning of Northeastern Florida, --- U.S. at ----, 113 S. Ct. at 2303. The Contractors do not dispute this fact but contend they should be allowed to challenge the Ordinance in its entirety, because the construction provisions are inseverable from the remainder of the statute.

Because "[s]everability of a local ordinance is a question of state law," City of Lakewood v. Plain Dealer Pub. Co., 486 U.S. 750, 772, 108 S. Ct. 2138, 2152, 100 L. Ed. 2d 771 (1988), we look to Pennsylvania law in determining whether the Ordinance is severable. As evidenced by its general severability provision, 1 Pa.Cons.Stat.Ann. Sec. 1925 (Supp.1991), Pennsylvania law favors severability. Commonwealth Dept. of Educ. v. First School, 471 Pa. 471, 478, 370 A.2d 702, 705 (1977).6 Additionally, there is a presumption in favor of severability where, as here, the Ordinance contains a specific severability provision. Sec. 17-508.7

Equipped with these principles, we must decide whether the Ordinance's provisions on different types of contracts "are distinct and not so interwoven as to be inseparable." Saulsbury v. Bethlehem Steel Co., 413 Pa. 316, 320, 196 A.2d 664, 666 (1964). The Saulsbury standard turns largely on functional concerns. Thus, the Pennsylvania Supreme Court severed a statute authorizing financial aid to sectarian and nonsectarian private schools because the statute was "readily capable of being administered along a sectarian-nonsectarian dichotomy." First School, 471 Pa. at 479, 370 A.2d at 706. Accordingly, the court held aid could continue to be provided to nonsectarian schools despite the fact that aid to sectarian schools violated the Establishment Clause. Id. We applied similar reasoning in severing portions of the Pennsylvania Abortion Control Act in Planned Parenthood v. Casey, 978 F.2d 74, 78 (3d Cir.1992), finding "[t]he basic statutory scheme remains intact and can operate independently of the unconstitutional sections."

In urging the Ordinance is not severable, the Contractors rely on Wyoming v. Oklahoma, --- U.S. ----, 112 S. Ct. 789, 117 L. Ed. 2d 1 (1992). There, the Supreme Court considered a Commerce Clause challenge to an Oklahoma statute requiring "all entities providing electric power for sale to the consumer in Oklahoma and generating said power from coal-fired plants located in Oklahoma [to] burn a mixture of coal that contains a minimum of ten percent Oklahoma mined coal...." --- U.S. at ---- n. 1, 112 S. Ct. at 793 n. 1. The state of Oklahoma urged the Court to construe the statute as constitutional when applied to a state-owned utility under the market participant exception to the Commerce Clause, a ruling that would require severing application of the statute to private utilities. Rejecting this contention, the Supreme Court held:

there are no parts or separate provisions in the invalid [section] of the Act [because] it applies to "all entities providing electric power for sale to the consumer in Oklahoma" and commands them to purchase 10% Oklahoma-mined coal. Nothing remains to be saved once that provision is stricken. Accordingly, the Act must stand or fall as a whole.

--- U.S. at ----, 112 S. Ct. at 803. The Contractors argue Wyoming dictates a finding of inseverability in this case because, like the statute there, the Ordinance applies to "all city contracts."

In assessing this contention, we look to the language of the Ordinance. The Ordinance provides it applies to: "all types of city contracts," Sec. 17-502(2), and defines "Types of City Contracts" to include: "all city contracts, whether competitively bid or negotiated, according to the following classes: (a) Vending, to include material, equipment, services and supplies; (b) Construction; and (c) Personal and professional services," Sec. 17-501(6).

These provisions reveal the Ordinance differs critically from the Oklahoma statute at issue in Wyoming. Here, the challenged and unchallenged provisions appear in different subsections, containing the "separate provisions" the Supreme Court found lacking in the Oklahoma statute. Also, the severability provision in the Ordinance is broader than that in the Oklahoma statute. The provision there only authorized severance of "any part or provision" of the statute found void, while the Ordinance directs severance of "any section, subsection, clause, sentence or phrase." Sec. 17-508. Because the provisions dealing with non-construction contracts constitute "subsections" of the Ordinance, Sec. 17-508 authorizes severing these provisions.

Moreover, the Ordinance makes clear that severance of the construction provisions would not prevent the non-construction provisions from continuing to operate. It expressly provides that the percentage goals for minority-owned and women-owned businesses, and businesses owned by handicapped persons, shall be "calculated by examining independently each type of City contract for each agency which lets such contracts." Sec. 17-503. This language demonstrates severance is feasible and indicates City Council envisioned the Ordinance would be administered separately for each type of contract. Therefore, we hold the provisions of the Ordinance dealing with construction contracts are severable from the remainder of the Ordinance. Because the Contractors only have a personal stake in the construction contract provisions, we must limit our review to these provisions unless we find the Contractors have standing to assert the rights of other businesses affected by the sections of the Ordinance dealing with contracts for vending and services.

In United States v. Raines, the Supreme Court delineated certain situations where a party has "third-party standing" to assert the rights of others. 362 U.S. at 22-23, 80 S. Ct. at 523-24. One such circumstance absent here occurs where the statute is not severable and therefore a party's challenge to the statute necessarily involves assertion of the rights of others affected by the statute. Id.Raines also stated third-party standing is appropriate where a third party faces some obstacle in asserting his right, id., an exception also unavailable here. Because we decline to address the constitutionality of the provisions of the Ordinance applicable to non-construction contracts, our ruling will create no precedent that would impede future contractors from challenging the Ordinance in a subsequent action nor otherwise impair their legal rights.8

Denial of third-party standing is especially appropriate here, where all parties have proceeded under the shared assumption that this litigation concerns only the construction provisions of the Ordinance. Neither in the district court nor in this Court has any party focused on application of the Ordinance to other types of contracts. Indeed, we remanded the case so defendants could take discovery "that relates to a pattern of discriminatory practices or instances of discrimination in the Philadelphia-area construction industry." 945 F.2d at 1267. Accordingly, no evidence has been presented on other areas of City contracting. To consider application of the Ordinance to these contracts without the benefit of the adversary process would require us to proceed without "data relevant and adequate to an informed judgment." New York v. Ferber, 458 U.S. 747, 768, 102 S. Ct. 3348, 3360, 73 L. Ed. 2d 1113 (1982). For these reasons, we hold the Contractors have standing only to challenge the provisions of the Ordinance relating to construction contracts. Accordingly, we will vacate that portion of the district court's order invalidating the remainder of the Ordinance.

The Contractors challenge the preferences given by the Ordinance to businesses owned and operated by minorities, women, and handicapped persons. In analyzing these classifications separately, we first consider which standard of equal protection review applies to each classification.

Choice of the appropriate standard of review turns on the nature of the classification. Because under equal protection analysis classifications based on race, ethnicity, or gender are inherently suspect, they merit closer judicial attention. City of Cleburne v. Cleburne Living Center, Inc., 473 U.S. 432, 440, 105 S. Ct. 3249, 3254, 87 L. Ed. 2d 313 (1985). Accordingly, we must determine whether the Ordinance contains race- or gender-based classifications. The Ordinance's classification scheme is spelled out in its definition of "socially and economically disadvantaged persons," described as:

those individuals who have either 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. Sec. 17-501(11).

The district court interpreted this definition to apply only to minorities, women, and handicapped persons and viewed the definition's economic criteria as in addition to rather than in lieu of race, ethnicity, gender, and handicap. Therefore, it applied strict scrutiny to the racial preference under Croson and intermediate scrutiny to the gender preference under Mississippi University for Women v. Hogan, 458 U.S. 718, 724, 102 S. Ct. 3331, 3336, 73 L. Ed. 2d 1090 (1982).9

Disputing this analysis, the City contends the two categories, "social disadvantage" (race, gender, or handicap), and "economic disadvantage" (diminished capital and credit opportunities), are disjunctive, so that a contractor who can satisfy either category qualifies as "socially and economically disadvantaged." Accordingly, the City continues, we should apply rational basis review to the Ordinance rather than strict or intermediate scrutiny because race and gender are only two criteria for determining whether a contractor is a DBE but are not prerequisites to that determination.

We agree with the district court that the definition of "socially and economically disadvantaged individuals" includes only individuals who are both victims of prejudice based on status and economically deprived. This is the only reasonable interpretation and is dictated by the conjunctive phrase "socially and economically disadvantaged," as well as the use of the word "and" connecting the descriptions of each category of disadvantage. Sec. 17-501(11) ("individuals subject to ... prejudice because of their identity as a member of a group ... and whose ability to compete in the free enterprise system has been impaired ...") (emphasis added). Additionally, the last clause of the definition describes economically disadvantaged individuals as those "whose ability to compete in the free enterprise system has been impaired ... as compared to others ... who are not socially disadvantaged." This clause demonstrates the drafters wished to rectify only economic disadvantage that results from social disadvantage, i.e., prejudice based on race, ethnicity, gender, or handicapped status. The plain language of the Ordinance forecloses the City's argument that a white male contractor could qualify for preferential treatment solely on the basis of economic disadvantage.

The City also relies on the affidavit of Curtis Jones, Executive Director of the Minority Business Enterprise Council, who asserts that "any disadvantaged member of plaintiff organizations ... [is] eligible for certification as a DBE without regard to their race, gender or physical capability. The [Council] considers economic and social factors in addition to race and gender in determining whether an applicant should be certified as a DBE." App. 123. But Jones' affidavit, which neither explains its interpretation nor has any legal force by itself, cannot overcome the plain language of the Ordinance. The affidavit is also inconsistent with the Council's regulations, which incorporate the Ordinance's definition verbatim. Regulations Sec. 1.3(S).

Additionally, the City seeks to distinguish Croson on the ground that the Philadelphia City Council, unlike the Richmond City Council, does not have a Black majority. This distinction comes from United States v. Carolene Products Co., 304 U.S. 144, 152-53 n. 4, 58 S. Ct. 778, 783-84 n. 4, 82 L. Ed. 1234 (1938), where the Court first suggested heightened scrutiny under the Equal Protection Clause might be appropriate for laws that prejudiced "discrete and insular minorities" because those groups would consistently be outvoted in legislatures and therefore could turn only to the judiciary for help. Under this theory, affirmative action programs typically do not merit heightened scrutiny because they represent examples of a white majority burdening itself. Croson, 488 U.S. at 496-97, 109 S.Ct. at 723-24 (citing John Hart Ely, The Constitutionality of Reverse Racial Discrimination, 41 U.Chi.L.Rev. 723, 739 n. 58 (1974)). Logically, such programs are suspect where they are enacted by a "predominantly Black legislature." Seeid. at 497, 109 S. Ct. at 724.

But the Croson Court expressly did not rely on the Carolene Products theory. In its discussion, the Court stated: "[e]ven were we to accept a reading of the guarantee of equal protection under which the level of scrutiny varies according to the ability of different groups to defend their interests in the representative process, heightened scrutiny would still be appropriate in the circumstances of this case...." 488 U.S. at 495, 109 S.Ct. at 722; see alsoO'Donnell Constr. Co. v. District of Columbia, 963 F.2d 420, 424 (D.C.Cir.1992) ("The outcome in Croson, expressed in the portions of the opinion joined by a majority, did not turn on who possessed political power in Richmond"). Therefore, we do not believe the racial makeup of the Philadelphia City Council changes the standard of review in this case.

Even were the Philadelphia City Council comprised of a Black majority, we would not rely on this fact in choosing a standard of review. Application of this theory in a reverse discrimination case, if logical, is ahistorical; it renders somewhat hollow the promise of racial progress through political power. More generally, the Carolene Products theory puts a court in the awkward position of nullifying legislative outcomes based on judges' own assumptions about the political process. And as one commentator has noted, the process of selecting groups as discrete and insular minorities necessarily requires normative judgments about political outcomes, i.e., those groups identified as minorities are those whose interests the court believes are inadequately protected by existing policies. Terrance Sandalow, The Distrust of Politics, 56 N.Y.U.L.Rev. 446, 466-67 (1981). Such judgments improperly interfere with the legislative will. For these reasons, we conclude the district court properly applied strict scrutiny to the racial preference in the Ordinance.

We next consider the proper standard of review for the Ordinance's gender preference. In Hogan, the Court held a gender-based classification favoring women merited intermediate scrutiny. 458 U.S. at 728, 102 S.Ct. at 3338. The Ordinance is such a program. Several federal courts have applied intermediate scrutiny to similar gender preferences contained in state and municipal affirmative action contracting programs. Coral Constr. Co. v. King County, 941 F.2d 910, 930 (9th Cir.1991), cert. denied, --- U.S. ----, 112 S. Ct. 875, 116 L. Ed. 2d 780 (1992); Michigan Road Builders Ass'n, Inc. v. Milliken, 834 F.2d 583, 595 (6th Cir.1987), aff'd mem., 489 U.S. 1061, 109 S. Ct. 1333, 103 L. Ed. 2d 804 (1989); Associated General Contractors of Cal. v. City and County of San Francisco, 813 F.2d 922, 942 (9th Cir.1987); Main Line Paving Co. v. Board of Educ., 725 F. Supp. 1349, 1362 (E.D.Pa.1989). Application of intermediate scrutiny to the Ordinance's gender preference also follows logically from Croson, which held municipal affirmative action programs benefiting racial minorities merit the same standard of review as that given other race-based classifications. For these reasons, we reject, as did the district court, those cases applying strict scrutiny to gender-based classifications. Cone Corp. v. Hillsborough County, 908 F.2d 908 (11th Cir.), cert. denied, 498 U.S. 983, 111 S. Ct. 516, 112 L. Ed. 2d 528 (1990); Conlin v. Blanchard, 890 F.2d 811, 816 (6th Cir.1989); American Subcontractors Ass'n v. City of Atlanta, 259 Ga. 14, 376 S.E.2d 662 (1989). As the district court noted, 735 F. Supp. at 1302, these cases neither explained their choice of the strict scrutiny standard nor cited any language in Croson supporting that choice. Accordingly, we agree with the district court's choice of intermediate scrutiny to review the Ordinance's gender preference.

The district court reviewed the preference for handicapped business owners under the rational basis test. 735 F. Supp. at 1307. That standard validates the classification if it is "rationally related to a legitimate governmental purpose." Cleburne, 473 U.S. at 445, 105 S.Ct. at 3257. In selecting this standard, the district court relied on Cleburne, where the Supreme Court rejected an effort to make mentally retarded persons a quasi-suspect class as a potentially limitless exercise which could apply to all groups with "immutable disabilities." Id.

The Contractors contend the district court erred in not applying some form of heightened scrutiny to the preference for handicapped individuals. Acknowledging Cleburne requires us to apply rational basis review, the Contractors maintain that decision is "obviously contrary to the sense of American society as a whole" given the Americans with Disabilities Act. In that Act, Congress expressly found that "individuals with disabilities are a discrete and insular minority, who have been faced with restrictions and limitations, subjected to a history of purposeful unequal treatment, and relegated to a position of political powerlessness in our society...." 42 U.S.C. Sec. 12101 (1990). The Contractors maintain the ADA overruled Cleburne and requires us to apply heightened scrutiny to the preference for handicapped business owners.

We disagree. The Contractors offer no evidence that the ADA overruled Cleburne, and the limited case law is to the contrary. More v. Farrier, 984 F.2d 269, 271 n. 4 (8th Cir.1993) (ADA does not "alter the standard for constitutional equal protection claims"). Moreover, we believe application of heightened scrutiny to the preference for handicapped business owners would run counter to the ADA, which Congress enacted to reduce discrimination against handicapped persons. 42 U.S.C. Sec. 12101(a)(5)-(9) (Supp.1990). Therefore, we hold the district court properly chose the rational basis standard in reviewing the Ordinance's preference for handicapped persons.

We now consider the district court's determination to invalidate the Ordinance in all respects.

Because strict scrutiny applies to the Ordinance's racial and ethnic preferences, we may only uphold them if they are "narrowly tailored" to a "compelling government interest." Wygant, 476 U.S. at 274, 106 S.Ct. at 1847. In Croson, the Supreme Court made clear that combatting racial discrimination is a "compelling government interest." 488 U.S. at 492, 509, 109 S. Ct. at 721, 730. It also held a city can enact such a preference to remedy past or present discrimination where it has actively discriminated in its award of contracts or has been a " 'passive participant' in a system of racial exclusion practiced by elements of the local construction industry." Id. at 492, 109 S.Ct. at 721.

But in view of the Court's concern about race-based classifications, Croson requires the City to "identif[y] the discrimination with the particularity required by the Fourteenth Amendment." Id. In striking down the Richmond ordinance requiring prime construction contractors to subcontract 30 percent of the dollar amount of each contract to minority businesses, the Court emphasized the City's failure to show specific evidence of past discrimination in the Richmond construction industry. 488 U.S. at 498, 109 S.Ct. at 723. It found evidence of prior national discrimination geographically overbroad because "the probative value of these findings for demonstrating the existence of discrimination in Richmond is extremely limited." Id. at 504, 109 S.Ct. at 727. The Court was similarly unimpressed with data showing the percentage of prime contracts awarded to minority businesses (.67%) was disproportionately low given the percentage of Richmond's population that were minorities (50%), and that very few minorities were members of local construction trade associations. 488 U.S. at 499, 109 S.Ct. at 724. In the Court's view, the "relevant statistical pool" was not the minority population but the number of qualified minority contractors. It stressed the city did not know the number of qualified minority businesses in the area and had offered no evidence of the percentage of contract dollars minorities received as subcontractors. Id. at 502, 109 S.Ct. at 726.

Ruling the Philadelphia Ordinance's racial preference failed to overcome strict scrutiny, the district court concluded the Ordinance "possesses four of the five characteristics fatal to the constitutionality of the Richmond Plan," 735 F. Supp. at 1298. As there, the district court reasoned, the City relied on national statistics, a comparison between prime contract awards and the percentage of minorities in Philadelphia's population, the Ordinance's declaration it was remedial, and "conclusory" testimony of witnesses regarding discrimination in the Philadelphia construction industry. Id. at 1295-98.10

a. Anecdotal Evidence of Racial Discrimination

The City and UMEA contend the district court understated the evidence of prior discrimination available to the Philadelphia City Council when it enacted the 1982 ordinance. The City Council Finance Committee received testimony from at least fourteen minority contractors who recounted personal experiences with racial discrimination. In certain instances, these contractors lost out despite being low bidders. This anecdotal evidence significantly outweighs that presented in Croson, where the Richmond City Council heard "no direct evidence of race discrimination on the part of the city in letting contracts or any evidence that the city's prime contractors had discriminated against minority-owned subcontractors." 488 U.S. at 480, 109 S.Ct. at 715.

Although the district court acknowledged the minority contractors' testimony was relevant under Croson, it discounted this evidence because "other evidence of the type deemed impermissible by the Supreme Court ... unsupported general testimony, impermissible statistics and information on the national set-aside program, ... overwhelmingly formed the basis for the enactment of the set-aside ... and therefore taint[ed] the minds of city councilmembers." 735 F. Supp. at 1296.

The district court's approach appears inconsistent with the accepted proposition that "at the summary judgment stage the [trial] judge's function is not himself to weigh the evidence and determine the truth of the matter but to determine whether there is a genuine issue for trial." Anderson v. Liberty Lobby, Inc. 477 U.S. 242, 249, 106 S. Ct. 2505, 2510, 91 L. Ed. 2d 202 (1986). Moreover, the court offered only a few factual examples of this impermissible evidence but did not attempt to quantify or otherwise explain why the impermissible evidence was dominant and why Council members necessarily attached more weight to this evidence than to the permissible evidence.

Yet given Croson's emphasis on statistical evidence, even had the district court credited the City's anecdotal evidence, we do not believe this amount of anecdotal evidence is sufficient to satisfy strict scrutiny. SeeCoral Constr., 941 F.2d at 919 ("anecdotal evidence ... rarely, if ever, can ... show a systemic pattern of discrimination necessary for the adoption of an affirmative action plan."). Although anecdotal evidence alone may, in an exceptional case, be so dominant or pervasive that it passes muster under Croson, it is insufficient here. But because the combination of "anecdotal and statistical evidence is potent," Coral Constr., 941 F.2d at 919, we turn to the statistical evidence proffered in support of the Ordinance.

b. Statistical Evidence of Racial Discrimination

There are two categories of statistical evidence here, evidence undisputedly considered by City Council before it enacted the Ordinance in 1982 (the "pre-enactment" evidence), and evidence developed by the City on remand (the "post-enactment" evidence).

i. Pre-Enactment Statistical Evidence

The principal pre-enactment statistical evidence appears in the 1982 Report of the City Council Finance Committee and recites that minority contractors were awarded only .09 percent of City contract dollars during the preceding three years, 1979 through 1981, although businesses owned by Blacks and Hispanics accounted for 6.4 percent of all businesses licensed to operate in Philadelphia.11 These statistics do not satisfy Croson because they do not indicate what proportion of the 6.4 percent of minority-owned businesses were available or qualified to perform City construction contracts. Under Croson, available minority-owned businesses comprise the "relevant statistical pool." 488 U.S. at 501, 109 S.Ct. at 726.12 Therefore, the data in the Finance Committee Report do not provide a sufficient evidentiary basis for the Ordinance.

ii. Post-Enactment Statistical Evidence

The "post-enactment" evidence consists of a study conducted by economic consultant Andrew Brimmer to demonstrate the disproportionately low share of public and private construction contracts awarded to minority-owned businesses in Philadelphia. Brimmer's study provides the "relevant statistical pool" needed to satisfy Croson--the percentage of minority businesses engaged in the Philadelphia construction industry.13

As a threshold matter, we are not certain the Brimmer study is post-enactment evidence. The study uses statistics for the three years immediately preceding enactment of the Ordinance. As the Contractors concede, "the data used by Brimmer respecting MBE participation as prime contractors to the City prior to 1982 was before City Council when it enacted Sec. 17-500. Brimmer's simple mathematical conclusions respecting that data obviously were apparent to City Council at that time." Contractors Br. at 28. But because we cannot be certain the Brimmer study constituted pre-enactment evidence, we consider whether it was admissible as post-enactment evidence. The district court considered the evidence, but believed it insufficient to change the result. Accordingly, it did not comment on its pre- or post-enactment character.

Several courts have held post-enactment evidence is admissible in determining whether an Ordinance satisfies Croson. Coral Constr., 941 F.2d at 921; Harrison & Burrowes Bridge Constr. v. Cuomo, 981 F.2d 50, 60 (2d Cir.1992); Concrete Works of Colorado, Inc. v. City and County of Denver, 823 F. Supp. 821, 836-37 (D.Colo.1993). As the Court of Appeals for the Ninth Circuit explained in Coral Constr., were post-enactment evidence inadmissible, "a municipality having [some] evidence would face the dilemma of deciding whether to wait the months necessary for further development of the record, risking constitutional culpability [to Blacks] due to its inaction, or to act and to risk liability [to Whites] for acting prematurely but otherwise justifiably." 941 F.2d at 921; see alsoWygant, 476 U.S. at 291, 106 S.Ct. at 1856 (O'Connor, J., concurring) ("public employers are trapped between the competing hazards of liability to minorities if affirmative action is not taken to remedy apparent employment discrimination and liability to nonminorities if affirmative action is taken").

Consideration of post-enactment evidence is especially appropriate here, where the principal relief sought and the only relief granted by the district court, was an injunction. Because injunctions are prospective only, it makes sense to consider all available evidence before the district court, including the post-enactment evidence, which the district court did. Although we recognize the risk of insincerity associated with post-enactment evidence, we believe that risk is minimal here because the Brimmer study consists essentially of an evaluation and re-ordering of pre-enactment evidence--contracts awarded to minority-owned businesses in the three years preceding the Ordinance. For these reasons, we hold the Brimmer affidavit and study were admissible evidence.

c. Sufficiency of the Evidence

We now consider whether the statistical evidence, when combined with the anecdotal evidence, was sufficient to enable the racial preference in the Ordinance to survive summary judgment. The district court believed it was not, stating in its opinion on remand that the City's "additional discovery has not produced any evidence which would cause the court to reconsider its prior ruling."

At least three Courts of Appeals have considered the type and amount of statistical data needed to support a municipal race-based contract preference program after Croson. Associated Gen. Contractors of California v. Coalition for Economic Equity, 950 F.2d 1401 (9th Cir.1991) (upholding program), cert. denied, --- U.S. ----, 112 S. Ct. 1670, 118 L. Ed. 2d 390 (1992); Cone Corp. v. Hillsborough County, Fla., 908 F.2d 908 (11th Cir.) (reversing summary judgment for contractors challenging program), cert. denied, 498 U.S. 983, 111 S. Ct. 516, 112 L. Ed. 2d 528 (1990); O'Donnell Constr. Co. v. District of Columbia, 963 F.2d 420 (D.C.Cir.1992) (striking down program).

In both Cone and AGC of California, where the courts refused to strike down the ordinances, the city and county presented the type of statistics Croson indicated were most probative--data showing minority contractors received a disproportionately low share of contracts given their representation in the total contractor population. Cone, 908 F.2d at 915, AGC, 950 F.2d at 1414; see alsoConcrete Works, 823 F. Supp. 821 (granting summary judgment to city based largely on statistics showing disparity between available minority businesses and contracts awarded to minority businesses).

In O'Donnell, by contrast, the data were conflicting. The District of Columbia presented statistics that minority-owned contractors were "capable of performing 34 percent of the District's construction work," 963 F.2d at 426, but received only 3.4 percent of the total construction contracts. But other evidence showed minorities received 32.4 percent of repairs and improvement contracts, 32.2 percent of architectural contracts, and 24.5 percent of materials management contracts, data the court believed "cast doubt on the notion that the 3.4 percent figure resulted from agency discrimination." 963 F.2d at 426; see alsoAssociated Gen. Contractors of Connecticut v. New Haven, 791 F. Supp. 941, 946 (D.Conn.1992) (striking down minority contractor program where minority and women-owned businesses received a share of contracts "in proportion to the numbers of firms in existence").

In determining whether the statistical evidence was adequate here, we look to its critical component--the "disparity index." The index consists of the percentage of minority contractor participation in City contracts divided by the percentage of minority contractor availability or composition in the "population" of Philadelphia area construction firms. This equation yields a percentage figure which is then multiplied by 100 to generate a number between 0 and 100, with 100 consisting of full participation by minority contractors given the amount of the total contracting population they comprise.

Other courts considering equal protection challenges to similar ordinances have relied on disparity indices in determining whether Croson's evidentiary burden is satisfied. Cone Corp., 908 F.2d at 916; AGC of California, 950 F.2d at 1414; Concrete Works, 823 F. Supp. at 834; see alsoStuart v. Roche, 951 F.2d 446, 451 (1st Cir.1991) (employing similar statistical data to uphold affirmative action promotion program in Boston Police Department), cert. denied, --- U.S. ----, 112 S. Ct. 1948, 118 L. Ed. 2d 553 (1992). Disparity indices are highly probative evidence of discrimination because they ensure that the "relevant statistical pool" of minority contractors is being considered.

The Brimmer study reports a disparity index for City of Philadelphia construction contracts during the years 1979 through 1981 of 4 out of a possible 100. This index is significantly worse than that in other cases where ordinances have withstood constitutional attack. SeeCone Corp., 908 F.2d at 916 (10.78 disparity index); AGC of California, 950 F.2d at 1414 (22.4 disparity index); Concrete Works, 823 F. Supp. at 834 (disparity index "significantly less than" 100); see alsoStuart, 951 F.2d at 451 (disparity index of 10 in police promotion program); compareO'Donnell, 963 F.2d at 426 (striking down ordinance given disparity indices of approximately 100 in two categories).14 Therefore, we find the disparity index probative of discrimination in City contracting in the Philadelphia construction industry prior to enactment of the Ordinance.

The Contractors contend the Brimmer study is methodologically flawed because it considered only prime contractors and because it failed to consider the qualifications of the minority businesses or their interest in performing City contracts. In short, the Contractors maintain the Brimmer study does not indicate why there is a disparity between available minority contractors and their participation in contracting. The Contractors contend that these objections, without more, entitle them to summary judgment, arguing that under the strict scrutiny standard they do not bear the burden of proof and therefore need not offer a neutral explanation for the disparity to prevail.

The Contractors misconceive the allocation of the burden of proof in affirmative action cases. The Supreme Court has indicated that "[t]he ultimate burden remains with [plaintiffs] to demonstrate the unconstitutionality of an affirmative action program." Johnson v. Transport. Agency, Santa Clara County, 480 U.S. 616, 626, 107 S. Ct. 1442, 1449, 94 L. Ed. 2d 615 (1987) (quoting Wygant, 476 U.S. at 277-78, 106 S.Ct. at 1848-49). Thus, the Contractors, not the City, bear the burden of proof here.

The Johnson Court also explained how the burden of proof operates in an affirmative action case, stating that a challenge by a white employee to an employer's voluntary affirmative action program under Title VII:fits readily within the analytical framework set forth in McDonnell Douglas [Corp. v. Green, 411 U.S. 792, 93 S. Ct. 1817, 36 L. Ed. 2d 668 (1973).] ... Once a plaintiff establishes a prima facie case that race or sex has been taken into account ... the burden shifts to the employer to articulate a nondiscriminatory rationale for its decision. The existence of an affirmative action plan provides such a rationale. If such a plan is articulated as the basis for the employer's decision, the burden shifts to plaintiff to prove that [this] justification is pretextual and the plan is invalid. As a practical matter, of course, an employer will generally seek to avoid a charge of pretext by presenting evidence in support of its plan. That does not mean, however, ... that reliance on an affirmative action plan is to be treated as an affirmative defense requiring the employer to carry the burden of proving the validity of the plan. The burden of proving its invalidity remains on the plaintiff. 480 U.S. at 626-27, 107 S.Ct. at 1448-49.

Although Johnson was a Title VII case involving a public employer's affirmative action hiring program, Croson indicated the same approach would apply in a constitutional case involving an affirmative action contracting program. There, the Court stated:

Where there is a significant statistical disparity between the number of qualified minority contractors willing and able to perform a particular service and the number of contractors actually engaged by the locality or the locality's prime contractors, an inference of discriminatory exclusion could arise. See Bazemore v. Friday, 478 U.S. at 398, 106 S.Ct. [3000] at 3008 [92 L. Ed. 2d 315]; Teamsters v. United States, 431 U.S. at 337-339, 97 S.Ct. [1843] at 1856 [52 L. Ed. 2d 396].... Moreover, evidence of a pattern of individual discriminatory acts can, if supported by appropriate statistical proof, lend support to a local government's determination that broader remedial relief is justified. See Teamsters, 431 U.S. at 338, 97 S.Ct. at 1856, 488 U.S. at 509 [109 S.Ct. at 730].

This language demonstrates that where a city defends an affirmative action ordinance as a remedy for past discrimination, issues of proof are handled as they are in other cases involving a pattern or practice of discrimination. Croson's reference to an "inference of discriminatory exclusion" based on statistics, as well as its citation to Bazemore and Teamsters, both Title VII pattern cases, supports this interpretation. Additionally, in Johnson, a Title VII case, the Court explicitly looked to the constitutional standard, stating the plaintiff bears the burden in such a case, and "we see no basis for a different rule regarding a plan's alleged violation of Title VII." 480 U.S. at 626, 107 S.Ct. at 1448. We, too, have indicated statistical proof of discrimination is handled similarly under Title VII and equal protection principles. Newark Branch, NAACP v. Town of Harrison, N.J., 940 F.2d 792, 807 (3d Cir.1991).

Courts of Appeals considering challenges to affirmative action plans under Title VII have indicated that where the plan's proponent adduces evidence supporting an inference of discrimination, the plaintiff must rebut that inference to prevail. Thus, upholding a consent decree establishing an affirmative action program for the Omaha, Nebraska Police Department, the Court of Appeals for the Eighth Circuit directed a verdict for the city where a white police officer "did not present evidence to rebut the City's showing that the plan was remedial and a response to a racial imbalance," based on the disproportionately small percentage of blacks hired by the department. Donaghy v. City of Omaha, 933 F.2d 1448, 1460 (8th Cir.1991), cert. denied, --- U.S. ----, 112 S. Ct. 938, 117 L. Ed. 2d 109 (1992); see alsoStuart, 951 F.2d at 451; Howard v. McLucas, 871 F.2d 1000, 1007 (11th Cir.1989) (both upholding similar consent decrees against reverse discrimination challenges), cert. denied, 493 U.S. 1002, 110 S. Ct. 560, 107 L. Ed. 2d 555 (1989).

The Supreme Court has also offered guidance on the quantum of evidence needed to rebut statistical proof of discrimination. In Bazemore v. Friday, 478 U.S. 385, 106 S. Ct. 3000, 92 L. Ed. 2d 315 (1986), it suggested the approach taken by the Contractors here does not suffice. Reversing a verdict for defendants in a Title VII case, the court expressed disapproval with defendants "strategy at trial," noting they "declare[d] simply that many factors go into making up an individual employee's salary [but] made no attempt ... statistical or otherwise--to demonstrate that when these factors were properly organized and accounted for there was no significant disparity between the salaries of blacks and whites." 478 U.S. at 403 n. 14, 106 S. Ct. at 3010 n. 14.

As the Court of Appeals for the District of Columbia has observed, "[i]mplicit in the Bazemore holding is the principle that a mere conjecture or assertion ... that some missing factor would explain the existing disparities ... generally cannot defeat the inference of discrimination created by ... statistics." Palmer v. Shultz, 815 F.2d 84, 101 (D.C.Cir.1987); see alsoCatlett v. Missouri Highway & Transport. Comm'n, 828 F.2d 1260, 1266 (8th Cir.1987) (relying on Palmer and Bazemore to uphold judgment in sex discrimination case), cert. denied, 485 U.S. 1021, 108 S. Ct. 1574, 99 L. Ed. 2d 889 (1988); Concrete Works, 823 F. Supp. at 839-40 (relying on these decisions to uphold Denver minority contracting preference ordinance).

Under these holdings, the City's statistical evidence has created an inference of discrimination which the Contractors would have to rebut at trial either by proving a "neutral explanation" for the disparity, "showing the statistics are flawed, ... demonstrating that the disparities shown by the statistics are not significant or actionable, ... or presenting contrasting statistical data." Coral Constr., 941 F.2d at 921. A fortiori, this evidence is sufficient for the City to withstand summary judgment. Fed.R.Civ.P. 56 only requires them to establish a genuine issue of fact for trial.15 Indeed, as the Ninth Circuit concluded in Coral

Constr., the Contractors' objections to the study are properly presented to the trier of fact. 941 F.2d at 921 (reversing summary judgment to enable contractors to attack statistics). Accordingly, we believe the City's statistical evidence established a prima facie case of racial discrimination in the award of City of Philadelphia construction contracts.

Consistent with strict scrutiny, we must examine the data for each minority group contained in the Ordinance. The Ordinance applies to businesses owned by persons who are "Black, Hispanic, Asian-American, or Native American." Sec. 17-501(1)(a)-(d). The Census data on which the Brimmer study relies demonstrates that in 1982, the year the Ordinance was enacted, there were construction firms owned in Philadelphia by Blacks, Hispanics, and Asian-Americans, but not Native Americans.16 Therefore, neither the City nor prime contractors could have discriminated against construction companies owned by Native Americans at the time of the Ordinance and we affirm summary judgment as to them.

The Census Report indicates there were 12 construction firms owned by Hispanic persons, 6 firms owned by Asian-American persons, 3 firms owned by persons of Pacific Islands descent, and 1 other minority-owned firm.17 The Brimmer study calculates Hispanic firms represented .15% of the available firms and Asian-American, Pacific-Islander, and "other" minorities represented .12% of the available firms, and that these firms received no City contracts during the years 1979 through 1981. We do not believe these numbers are large enough to create a triable issue of discrimination. The mere fact that .27 percent of City construction firms--the percentage of all of these groups combined--received no contracts does not rise to the "significant statistical disparity" Croson requires. 488 U.S. at 509, 109 S.Ct. at 730.

Nor does it appear that there was any anecdotal evidence of discrimination against construction businesses owned by people of Hispanic or Asian-American descent. The district court found "there is no evidence whatsoever in the legislative history of the Philadelphia Ordinance that an American Indian, Eskimo, Aleut or Native Hawaiian has ever been discriminated against in the procurement of city contracts," 735 F. Supp. at 1299, and neither the City's nor UMEA's briefs call our attention to any witnesses who were members of these groups or who were Hispanic.

We recognize that the small number of Philadelphia-area construction businesses owned by Hispanic or Asian-American persons does not eliminate the possibility of discrimination against these firms. The small number itself may reflect barriers to entry caused in part by discrimination. But plausible hypotheses are not enough to satisfy strict scrutiny, even at the summary judgment stage. Because these groups appear in different subsections of Sec. 17-501 of the Ordinance, Secs. 17-501(b)-(d), in accordance with our earlier discussion of standing, we sever these subsections of the Ordinance. SeeAmerican Subcontractors Ass'n, v. City of Atlanta, 376 S.E.2d at 666 n. 7 (upholding severance of set-aside ordinance as to racial groups other than Blacks because there was no evidence of discrimination against these groups). Nothing in our decision prevents the City from re-enacting a preference for construction firms owned by Hispanic, Asian-American, or Native American persons based on more concrete evidence of discrimination. In sum, we believe the City and UMEA adduced enough evidence of racial discrimination against Blacks in the award of City construction contracts to withstand summary judgment on the compelling government interest prong of the Croson test.

We next decide whether the Ordinance's racial preference was "narrowly tailored" to the compelling government interest of eradicating racial discrimination in the award of City construction contracts. Croson held this inquiry turns on four factors: (1) whether the city has first considered and found ineffective "race-neutral measures," such as enhanced access to capital and relaxation of bonding requirements, (2) the basis offered for the percentage selected, (3) whether the program provides for waivers of the preference or other means of affording individualized treatment to contractors, and (4) whether the Ordinance applies only to minority businesses who operate in the geographic jurisdiction covered by the Ordinance. 488 U.S. at 507-08, 109 S.Ct. at 729-30.

Holding the Ordinance was not "narrowly tailored," the district court believed the Ordinance failed to satisfy most of these criteria. It stated City Council had not considered race-neutral measures, stressed the testimony of a City Council member that the 15 percent number was arbitrarily chosen, and inferred that testimony regarding "front" minority businesses reflected there were not enough legitimate minority enterprises to meet the percentage requirement and therefore that it was too high. 735 F. Supp. at 1298-99.

On appeal, the City contends it enacted the Ordinance only after race-neutral alternatives proved insufficient to improve minority participation in City contracting. It relies on the affidavits of City Council President Joseph Coleman and former Philadelphia Urban Coalition General Counsel Oscar Gaskins who testified regarding the race-neutral precursors of the Ordinance--the Philadelphia Plan, which set goals for employment of minorities on public construction sites, and the Urban Coalition's programs, which included such race-neutral measures as a revolving loan fund, a technical assistance and training program, and bonding assistance efforts. We believe the information in these affidavits sufficiently establishes the City's prior consideration of race-neutral programs to withstand summary judgment.18 SeeConcrete Works, 823 F. Supp. at 841 (relying on similar approaches to conclude program was narrowly tailored).

Unlike the Richmond Ordinance, the Philadelphia Ordinance provides for several types of waivers of the fifteen percent goal. It exempts individual contracts or classes of contracts from the Ordinance where there are an insufficient number of available minority-owned businesses "to ensure adequate competition and an expectation of reasonable prices on bids or proposals," Sec. 17-505(1), (2), and allows a prime contractor to request a waiver of the fifteen percent requirement where the contractor shows he has been unable after "a good faith effort to comply with the goals for DBE participation," Sec. 17-505(3). Furthermore, as the district court noted, the Ordinance eliminates from the program successful minority businesses--those who have won $5 million in city contracts. Also unlike the Richmond program, the City's program is geographically targeted to Philadelphia businesses, as waivers and exemptions are permitted where there exist an insufficient number of MBEs "within the Philadelphia Standard Metropolitan Statistical Area." Sec. 17-505(1), (2).

Other courts have found these targeting mechanisms significant in concluding programs are narrowly tailored. SeeCone Corp., 908 F.2d at 917; Associated Gen. Contractors of Cal. v. City and County of San Francisco, 748 F. Supp. 1443, 1454 (N.D.Cal.1990), aff'd, 950 F.2d 1401 (9th Cir.1991), cert. denied, --- U.S. ----, 112 S. Ct. 1670, 118 L. Ed. 2d 390 (1992); compareCroson, 488 U.S. at 508, 109 S.Ct. at 729 ("Richmond's waiver system ... makes no inquiry into whether or not the particular MBE seeking a preference has suffered from the effects of past discrimination by the city or prime contractors").

A closer question is presented by the Ordinance's fifteen percent goal. The City's data demonstrated that, prior to the Ordinance, only 2.4 percent of available construction contractors were minority-owned. We do not believe the goal must correspond precisely to the percentage of available contractors. Indeed, Croson does not impose this requirement, as the Court stated only that Richmond's 30 percent goal inappropriately assumed "minorities [would] choose a particular trade in lockstep proportion to their representation in the local population." 488 U.S. at 507, 109 S.Ct. at 729. Furthermore, imposing a fifteen percent goal for each contract may reflect the need to account for those contractors who receive a waiver because insufficient minority businesses are available, and the contracts exempted from the program. And given the strength of the Ordinance's showing with respect to other Croson factors, we conclude the City has created a dispute of fact on whether the minority preference in the Ordinance is "narrowly tailored."

For these reasons, we will reverse the district court's grant of summary judgment to the extent it invalidated application of the construction contract provisions to businesses owned by Black persons.

Under the intermediate scrutiny standard, the gender preference is valid if it was "substantially related to an important governmental objective." In striking down the gender preference, the district court found insufficient comparisons between the percentage of city businesses owned by women and the percentage of contracts awarded to women-owned businesses. It also relied on testimony before City Council suggesting the preference was solely the product of political compromise.

The City contends the gender preference is aimed at the "important government objective" of remedying economic discrimination against women, and that the ten percent goal is substantially related to this objective. In assessing this argument, we note that "[i]n the context of women-business enterprise preferences, the two prongs of this intermediate scrutiny test tend to converge into one." Coral Constr., 941 F.2d at 931. At bottom, we can uphold the construction provisions of this program if the City has established a sufficient factual predicate for the claim that women-owned construction businesses have suffered economic discrimination and the ten percent gender preference is an appropriate response.19

Few cases have considered the evidentiary burden needed to satisfy intermediate scrutiny in this context and there is no Croson analogue to provide a ready reference point. In particular, it is unclear whether statistical evidence as well as anecdotal evidence is required to establish the discrimination necessary to satisfy intermediate scrutiny, and if so, how much statistical evidence is necessary. The Supreme Court gender-preference cases are inconclusive. The Court has never squarely ruled on the necessity of statistical evidence of gender discrimination. And its decisions are difficult to reconcile on the point. The Court has upheld gender preferences where no statistics were offered, Schlesinger v. Ballard, 419 U.S. 498, 508-09, 95 S. Ct. 572, 577-78, 42 L. Ed. 2d 610 (1975) (preferential employment treatment for women military officers), struck down gender preferences despite the presence of statistics, Craig v. Boren, 429 U.S. 190, 203-04, 97 S. Ct. 451, 460-61 (1976) (statute allowing 18 year old women but only 21 year old men to purchase 3.2% beer), Weinberger v. Wiesenfeld, 420 U.S. 636, 645, 95 S. Ct. 1225, 1231-32, 43 L. Ed. 2d 514 (1975) (statute allowing survivors' benefits for widows but not widowers), and also decided cases both ways by relying in part on statistics, Hogan, 458 U.S. at 728, 102 S. Ct. at 3338 (striking down female-only nursing school), Califano v. Webster, 430 U.S. 313, 318 & n. 5, 97 S. Ct. 1192, 1195 & n. 5, 51 L. Ed. 2d 360 (1977) (upholding federal statute allowing women to eliminate more low-earning years from calculation of their retirement benefits than men).

Lower court cases are similarly diverse. In Coral Constr., 941 F.2d at 933, the Ninth Circuit emphasized a single, albeit lengthy, affidavit from a woman business owner in affirming summary judgment upholding a contract preference ordinance, although the court elsewhere noted three affidavits had been presented, id. at 917-18. Yet in AGC of California, the same court, relying on Hogan, upheld a general challenge to a gender preference in San Francisco's contracting ordinance but expressly left open the possibility of a future challenge regarding application of the ordinance "to an industry where women are not disadvantaged." 813 F.2d at 942. Additionally, two circuits are divided over whether evidence of governmental discrimination against women is required to establish the important government interest needed to satisfy a sex-based affirmative action program. CompareCoral Constr., 941 F.2d at 932 (no such evidence needed) withMichigan Road Builders, 834 F.2d at 595 (evidence needed). Logically, a city must be able to rely on less evidence in enacting a gender preference than a racial preference because applying Croson's evidentiary standard to a gender preference would eviscerate the difference between strict and intermediate scrutiny.

The Supreme Court has stated that an affirmative action program survives intermediate scrutiny if the proponent can show it was "a product of analysis rather than a stereotyped reaction based on habit." Metro Broadcasting, Inc. v. F.C.C., 497 U.S. 547, 582-83, 110 S. Ct. 2997, 3018-19, 111 L. Ed. 2d 445 (1990) (citation and internal quotation marks omitted). We believe this standard requires the City to present probative evidence in support of its stated rationale for the gender preference, discrimination against women-owned contractors. CompareMetro Broadcasting, 497 U.S. at 580, 110 S.Ct. at 3017 ("a host of empirical evidence" supported the linkage between minority ownership of radio stations and broadcasting diversity, the challenged program's stated goal).

The City has not produced enough evidence of discrimination here. In its brief, the City relies on statistics in the City Council Finance Committee Report and one affidavit from a woman engaged in the catering business, but this evidence only reflects the participation of women in City contracting generally, rather than in the construction industry, which is the only cognizable issue here.

The evidence offered by the City regarding women-owned construction businesses is insufficient to create an issue of fact. Significantly, the Brimmer study contains no disparity index for women-owned construction businesses in City contracting, such as that presented for minority-owned businesses.20 Given the absence of probative statistical evidence, the City must rely solely on anecdotal evidence to establish gender discrimination necessary to support the Ordinance. But the record contains only one three-page affidavit alleging gender discrimination in the construction industry. The only other testimony on this subject consists of a single, conclusory sentence of one witness who appeared at a City Council hearing.21 This evidence is not enough to create a triable issue of fact regarding gender discrimination under the intermediate scrutiny standard. Therefore, we will affirm the grant of summary judgment invalidating the gender preference for construction contracts. We see no impediment to the City re-enacting the preference if it can provide probative evidence of discrimination.

Finally we address the two-percent preference for businesses owned by handicapped persons. The district court struck down this preference under the rational basis test, believing Croson required some evidence of discrimination against business enterprises owned by handicapped persons and therefore that the City could not rely on testimony of discrimination against handicapped individuals. 735 F. Supp. at 1308.

A classification will pass the rational basis test if it is "rationally related to a legitimate government purpose," Cleburne, 473 U.S. at 440, 105 S.Ct. at 3254. The Supreme Court recently reaffirmed the permissiveness of this test in Heller v. Doe, --- U.S. ----, ---- - ----, 113 S. Ct. 2637, 2642-43, 125 L. Ed. 2d 257 (1993), indicating that "a [statutory] classification" subject to rational basis review "is accorded a strong presumption of validity," and that "a state ... has no obligation to produce evidence to sustain the rationality of [the] classification." Moreover, "the burden is on the one attacking the legislative arrangement to negative every conceivable basis which might support it, whether or not the basis has a foundation in the record." Id. (citation omitted).

The City stated it sought to minimize discrimination against businesses owned by handicapped persons and encourage them to seek City contracts. We agree with the district court that these are legitimate goals. But unlike the district court, we believe the two-percent preference is rationally related to this goal.

The City offered anecdotal evidence of discrimination against handicapped persons. Prior to amending the Ordinance in 1988 to include the preference, City Council held a hearing where eight witnesses testified regarding employment discrimination against handicapped persons both nationally and in Philadelphia. Four witnesses spoke of discrimination against blind people, and three testified to discrimination against people with other physical handicaps. Two of the witnesses, who were physically disabled, spoke compellingly of discrimination they and others had faced in the work force. One of these disabled witnesses testified he was in the process of forming his own residential construction company. Additionally, two witnesses testified that the preference would encourage handicapped persons to own and operate their own businesses.

We believe, under the rational basis standard, that the Contractors did not carry their burden of negativing every basis which supports the legislative arrangement, seeHeller, --- U.S. at ----, 113 S. Ct. at 2643, and that City Council was entitled to infer discrimination against the handicapped from this evidence and was entitled to conclude the Ordinance would encourage handicapped persons to form businesses to win City contracts. Therefore, we will reverse the district court's grant of summary judgment invalidating this aspect of the Ordinance and remand for entry of an order granting summary judgment to the City and UMEA on this issue.

For the foregoing reasons, we will vacate the district court's grant of summary judgment on the non-construction provisions of the Ordinance, reverse the grant of summary judgment on the construction provisions of the Ordinance as applied to businesses owned by Black persons and handicapped persons, affirm the grant of summary judgment on the construction provisions of the Ordinance as applied to businesses owned by Hispanic, Asian-American, or Native American persons or women, and remand for further proceedings in accordance with this opinion.