BOGGS, C.J., delivered the opinion of the court. MERRITT, J. (p. 379), delivered a separate opinion concurring in the judgment and in Section IV. MOORE, J. (pp. 379-94), delivered a separate opinion concurring in part and dissenting in part.
OPINION
BOGGS, Chief Judge.The Section 8 voucher program is a voluntary program through which the federal government provides rent subsidies to eligible low-income families who rent from participating landlords. See 42 U.S.C. § 1437f(a). The Fair Housing Act, also known as Title VIII, bars discrimination “against any person in the terms, conditions, or privileges of sale or rental of a dwelling ... because of race.... ” 42 U.S.C. § 3604(b). In this case, Graoch, the owner of Autumn Run Apartments in Louisville, seeks a declaratory judgment that it did not violate the FHA by withdrawing from the Section 8 program. Its claim presents two questions regarding the interplay between Section 8 and the FHA. First, can a landlord’s withdrawal from the Section 8 program ever violate the FHA solely because it has a disparate impact on members of a protected class? Second, if so, what are the standards for measuring disparate impact?
The district court answered the first question in the negative and therefore granted summary judgment in favor of Graoch without reaching the second question. We reach the same final result, but in a different way. Disagreeing with the position taken by the Second and Seventh Circuits, we hold that a plaintiff can, in principle, rely on evidence of some instances of disparate impact to show that a landlord violated the Fair Housing Act by withdrawing from Section 8. We also hold, however, that in this case the Metro Human Relations Commission did not even allege facts making the statistical comparison necessary to state a prima facie case of disparate-impact discrimination. Consequently, we affirm.
I
The Housing Authority of Jefferson County (“HAJC”) coordinates the local disbursement of Section 8 funds in the Louisville area. See 42 U.S.C. § 1437f(b)(1) (“The Secretary is authorized to enter into annual contributions contracts with public housing agencies pursuant to which such agencies may enter into contracts to make assistance payments to owners of existing dwelling units in accordance with this section.”). In March 2003, Graoch notified the HAJC that it intended to withdraw *370from the Section 8 voucher program, stating that it would honor existing leases by Section 8 tenants but would not renew those leases or sign any new Section 8 leases.
The Kentucky Fair Housing Council (“FHC”) and three Autumn Run tenants receiving Section 8 assistance — Joyce McNealy, Tina Gray, and Angela Thornton — filed a complaint with the Metro Human Relations Commission. The Commission found probable cause to believe that Graoch’s withdrawal from the Section 8 program constituted unlawful racial discrimination because it had a disparate impact on blacks. On Graoch’s motion, it then stayed administrative proceedings to give Graoch the opportunity to seek declaratory relief in federal court. Graoch responded by initiating this case.1
The parties agreed to a series of factual stipulations. Eighteen families receiving Section 8 assistance lived at Autumn Run when Graoch announced that it was withdrawing from the Section 8 program. Seventeen of those families were black. As of 2003, 6,270 of the 8,849 Jefferson County residents receiving Section 8 vouchers were black. Finally, as of the 2000 census, 18.9% of Jefferson County residents were black and 24% were members of black households. Absent from the joint stipulations, however, was any information regarding the races of the non-Section 8 tenants at Autumn Run. Indeed, the only reference contained in the record to the racial makeup of Autumn Run as a whole, or of its non-Section 8 tenants, is a remark by Graoch’s counsel, quoted by the Commission in its probable cause finding: “the tenants at Respondent project Autumn Run Apartments were 90% minority and 10% white.”
Graoch stated that it chose to withdraw from the Section 8 program because of disputes with the HAJC regarding rent payments made on behalf of Section 8 tenants. Graoch claimed that the HAJC held Graoch “to an impossible standard” in enforcing the quality standards for Section 8, “abating rent for conditions which Graoch was either not made aware of prior to inspection, or which it attempted to fix only to be cited upon re-inspection for not making its repairs to the arbitrary satisfaction of the inspector.” See 24 C.F.R. § 982.401 (stating “the housing quality standards ... for housing assisted” through the Section 8 program).
Based on this record, Graoch moved for summary judgment. First, it argued that the Commission failed to state a prima facie case that Graoch’s withdrawal from Section 8 violated the FHA because it had a disparate impact on blacks. Second, it argued that its withdrawal from Section 8 did not violate the FHA even if the Commission did state a prima facie case because the decision to withdraw resulted from a “business necessity.” The district court granted summary judgment for Graoch, holding that a party offering only evidence that a landlord’s withdrawal from the Section 8 program had a disparate impact on members of a protected class cannot establish a prima facie case that the landlord violated the FHA.
The Commission appealed. We review the district court’s decision de novo. See Trustees of the Mich. Laborers’ Health Care Fund v. Gibbons, 209 F.3d 587, 590 (6th Cir.2000). Summary judgment is appropriate “if the pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving par*371ty is entitled to a judgment as a matter of law.” Fed.R.Civ.P. 56(c). Ordinarily, we may affirm a grant of summary judgment on any basis fairly presented by the record, as long as the opposing party “is not denied an opportunity to respond to the new theory.” Herm v. Stafford, 663 F.2d 669, 684 (6th Cir.1981). In this case, however, both parties agree that we should remand the case to the district court if it is necessary to consider the adequacy of Graoch’s business justification for its withdrawal from Section 8. As a result, we consider only whether the Commission has stated a prima facie case.
II
Before delving into the specific implications of withdrawal from Section 8 and the even-more-specific details of this case, we must address as a general matter the law governing disparate-impact claims under the FHA. We previously have held that a plaintiff can establish a violation of the FHA by showing either disparate treatment or disparate impact.2 Larkin v. Mich. Dep’t of Soc. Servs., 89 F.3d 285, 289 (6th Cir.1996). We have stated that, to show disparate impact, a plaintiff must demonstrate that a facially neutral policy or practice has the effect of discriminating against a protected class of which the plaintiff is a member. Blaz v. Barberton Garden Apartment, No. 91-3896, 1992 WL 180180, at *3, 1992 U.S.App. LEXIS 18508, at *8 (6th Cir. July 29,1992) (unpublished). We have not decided, however, what framework we should use to determine whether a plaintiff has done enough to survive summary judgment on a disparate-impact claim against a private defendant under the FHA.
We find it useful to compare the frameworks that we have applied to three similar kinds of claims: disparate-treatment claims against private defendants under the FHA, disparate-impact claims against private defendants under Title VII, and disparate-impact claims against governmental defendants under the FHA. In disparate-treatment cases under the FHA, we apply “the three-part burden of proof test established in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973).” Selden Apartments v. U.S. Dep’t of Hous. & Urban Dev., 785 F.2d 152, 159 (6th Cir.1986) (internal citation and quotation marks omitted). First, the plaintiff must state a prima facie case by showing that he is a member of a protected class, that he applied to and was qualified to rent or purchase certain housing, that he was rejected, and that the housing remained available thereafter. Maki v. Laakko, 88 F.3d 361, 364 (6th Cir.1996). Second, the defendant may then articulate a legitimate non-discriminatory basis for its challenged decision. Selden Apartments, 785 F.2d at 160. Third, if the defendant does proffer such a basis, the plaintiff must establish that the articulated reason is pretextual. Ibid. The burden of persuasion always remains with the plaintiff.
We imported this framework from our disparate-treatment cases under Title VII. See Larkin, 89 F.3d at 289; compare Daniels v. Bd. of Educ., 805 F.2d 203, 207 (6th Cir.1986) (addressing a disparate-treatment claim under Title VII) with Selden, 785 F.2d at 159 (applying the same framework to a disparate-treatment claim under the FHA). We agreed with the reasoning used by the Second Circuit, which found *372“persuasive” the “parallel between Title VII and Title VIII”:
The two statutes are part of a coordinated scheme of federal civil rights laws enacted to end discrimination; the Supreme Court has held that both statutes must be construed expansively to implement that goal. See, e.g., Trafficante v. Metro. Life Ins. Co., 409 U.S. 205, 211-12, 93 S.Ct. 364, 34 L.Ed.2d 415 (1972) (Fair Housing Act must be generously construed to foster integration); Griggs v. Duke Power Co., 401 U.S. 424, 429-36, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971) (Title VII should be interpreted broadly to achieve equal employment opportunity). Courts and commentators have observed that the two statutes require similar proof to establish a violation.
Huntington Branch, NAACP v. Town of Huntington, 844 F.2d 926, 935 (2d Cir.), aff'd on other grounds, 488 U.S. 15, 109 S.Ct. 276, 102 L.Ed.2d 180 (1988) (cited in Blaz, 1992 WL 180180, at *3, 1992 U.S.App. LEXIS 18508, at *8, and Bangerter v. Orem City Corp., 46 F.3d 1491, 1501 (10th Cir.1995), which in turn was cited in Larkin, 89 F.3d at 289). In sum, we concluded that we generally should evaluate claims under the FHA by analogizing them to comparable claims under Title VII. Larkin, 89 F.3d at 289.
In disparate-impact cases under Title VII, we apply a different version of the McDonnell Douglas burden-shifting framework than in disparate-treatment cases. We follow the same three steps: the plaintiff states a prima facie case; the defendant responds; the plaintiff rebuts the response. See Kovacevich v. Kent State Univ., 224 F.3d 806, 830 (6th Cir.2000). To state a prima facie case, however, a Title VII plaintiff basing his claim on disparate impact must “identify ] and challeng[e] a specific employment practice, and then show an adverse effect by offering statistical evidence of a kind or degree sufficient to show that the practice in question has caused the adverse effect in question.” Ibid. To respond, the defendant must articulate a “legitimate business reason” for the challenged decision, rather than simply a non-discriminatory basis. Ibid. Finally, to rebut the defendant’s response, the plaintiff can show “either that the employer’s reason is a pretext for discrimination, or that there exists an alternative employment practice that would achieve the same business ends with a less discriminatory impact.” Ibid. Relying on the analogy between Title VII and the FHA, several other circuits have applied essentially this approach to disparate-impact claims under the FHA. Hack v. President & Fellows of Yale Coll., 237 F.3d 81, 98-102 (2d Cir.2000); Langlois v. Abington Hous. Auth., 207 F.3d 43, 51 (1st Cir.2000); Mountain Side Mobile Estates P’ship v. Sec’y of Hous. & Urban Dev. ex rel. VanLoozenoord, 56 F.3d 1243, 1251 (10th Cir.1995). So has a district court in our circuit. Marbly v. Home Props. of N.Y., 205 F.Supp.2d 736, 740 (E.D.Mich.2002).
In a disparate-impact case against a governmental defendant under the FHA, however, we did not mention the burden-shifting framework or the elements of a prima facie case. See Arthur v. City of Toledo, 782 F.2d 565, 575-77 (6th Cir.1986). Instead, we stated that, to determine “whether conduct which produces a discriminatory effect but which did not have a discriminatory intent” violates the FHA, we must consider three factors: (1) the strength of the plaintiffs showing of discriminatory effect; (2) the strength of the defendant’s interest in taking the challenged action; and (3) whether the plaintiff seeks to compel the defendant affirmatively to provide housing for members of minority groups or merely to restrain the defendant from interfering with individual *373property owners who wish to provide such housing. Id. at 575 (adopting a modified version of the four-factor standard used in Metropolitan Housing Development Corp. v. Village of Arlington Heights, 558 F.2d 1283, 1290 (7th Cir.1977) (“Arlington Heights II”)).3 Neither our opinion in Arthur nor the Seventh Circuit’s opinion in Arlington Heights II makes clear whether this three-factor balancing approach is the way to evaluate the claim as a whole after the parties have completed all three steps in the burden-shifting framework, or whether the burden-shifting framework does not apply. Compare 2922 Sherman Ave. Tenants’ Ass’n v. Dist. of Columbia, 444 F.3d 673, 680 (D.C.Cir.2006) (contrasting “the Seventh Circuit’s four-factor inquiry” to “the Second Circuit[’s] ... burden-shifting framework,” thereby indicating it reads Arlington Heights II not to require a prima facie case) with Resident Advisory Bd. v. Rizzo, 564 F.2d 126, 148 n. 32 (3d Cir.1977) (“We read the Seventh Circuit’s opinion in Arlington Heights II as requiring no more than we do in order for a plaintiff to establish a prima facie case.... To the extent that the Seventh Circuit would seem to go beyond this standard in its statement of ‘critical factors,’ our impression is that the court is setting forth a standard upon which ultimate Title VIII relief may be predicated, rather than indicating the point at which the evidentiary burden of justifying a discriminatory effect will shift to the defendant.”).
In formulating a framework to analyze disparate-impact claims against private defendants under the FHA, we think that it is best to merge the approaches we used in Arthur and in our Title VII disparate-impact cases by reading the Arthur standard as a supplement, not an alternative, to the burden-shifting framework. Treating the Arthur standard as an alternative would conflict with our statement in Lar-kin that similar claims under Title VII and the FHA generally should receive similar treatment. See Larkin, 89 F.3d at 289. And when applied to private defendants, the Arthur standard nicely captures the inquiry that we must perform at the third step in the burden-shifting framework. To determine whether a defendant’s proffered business reason is a pretext for discrimination, or whether an alternative practice exists that would achieve the same business ends with a less discriminatory impact, we must consider the strength of the plaintiffs statistical evidence of disparate impact and the strength of the defendant’s interest in maintaining the challenged practice. Those are the first two Arthur factors. The third Arthur factor, distinguishing defendants who wish only not to provide housing themselves from those who wish to stop others from providing housing, is not relevant to claims against private defendants, who lack the authority to control third-party behavior (though it is relevant when applied to governmental defendants, who through zoning regulations or land-use permit requirements might have the authority to control the housing development and allocation decisions of third parties). Consequently, we think that our approaches to disparate-impact claims under Title VII and disparate-impact claims against governmental *374defendants under the FHA are compatible, not inconsistent.
Borrowing from our Title VII cases, then, we hold that disparate-impact claims against private defendants under the FHA should be analyzed using a form of the McDonnell Douglas burden-shifting framework:
—First, a plaintiff must make a prima facie case of discrimination by “identifying and challenging a specific [housing] practice, and then show[ing] an adverse effect by offering statistical evidence of a kind or degree sufficient to show that the practice in question has caused the adverse effect in question,” Kovacevich, 224 F.3d at 830 (disparate impact claim under Title VII);
—Second, if the plaintiff makes a prima facie case, the defendant must offer a “legitimate business reason” for the challenged practice, ibid.;
—Third, if the defendant offers such a reason, the plaintiff must demonstrate that the defendant’s reason is “a pretext for discrimination, or that there exists an alternative [housing] practice that would achieve the same business ends with a less discriminatory impact.” Ibid. In order to evaluate the plaintiffs showing, we consider the strength of the plaintiffs showing of discriminatory effect against the strength of the defendant’s interest in taking the challenged action. See Arthur, 782 F.2d at 575 (disparate-impact claim against a governmental defendant under the FHA).
Ill
Graoch argues that, instead of analyzing the Commission’s claims through this burden-shifting framework, we should exempt landlords from any possible liability under the FHA for a complete withdrawal from the Section 8 voucher program.4 The Second Circuit, in Salute v. Stratford Greens Garden Apartments, 136 F.3d 293 (2d Cir.1998), and the Seventh Circuit, in Knapp v. Eagle Property Management Corp., 54 F.3d 1272 (7th Cir.1995), have created such a categorical exemption for Section 8 withdrawals. We decline to do so.
The FHA prohibits discrimination “against any person in the terms, conditions, or privileges of sale or rental of a dwelling ... because of race.... ” 42 U.S.C. § 3604(b). The Supreme Court held that Title VII, which uses similar language, “proscribes not only overt discrimination but also practices that are fair in form, but discriminatory in operation.” Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971). The same analysis justifies the existence of disparate-impact liability under the FHA. “What is required by Congress is the removal of artificial, arbitrary, and unnecessary barriers to [access to housing] when the barriers operate invidiously to discriminate on the basis of racial or other impermissible classification.” Ibid, (discussing Title VII).
Of course, not every housing practice that has a disparate impact is illegal. We use the burden-shifting framework described above — and especially the final inquiry considering the strength of the plaintiffs statistical evidence and the strength of the defendant’s business reason' — -to distinguish the artificial, arbitrary, and unnecessary barriers proscribed by the FHA from valid policies and practices crafted to *375advance legitimate interests. Our analysis is derived from our interpretation of the statute, and we cannot create categorical exemptions from it without a statutory basis.
In Knapp and Salute, our sister circuits seem to have taken a more circumscribed view of disparate-impact liability under the FHA. The Knapp court stated:
“[T]he courts must use their discretion in deciding whether, given the particular circumstances of each case, relief should be granted under the statute.” [Arlington Heights II,] 558 F.2d at 1290.... [T]his court has recognized that disparate impact analysis is not appropriate in certain contexts. See NAACP v. American Family Mutual Ins. Co., 978 F.2d 287, 290 (7th Cir.1992) (claim alleging failure to insure in certain areas violated the Act not conducive to disparate-impact analysis), cert. denied, 508 U.S. 907, 113 S.Ct. 2335, 124 L.Ed.2d 247(1993); Village of Bellwood v. Dwivedi 895 F.2d 1521, 1533 (7th Cir.1990) (“Some practices lend themselves to disparate impact method, others not.”).
Knapp, 54 F.3d at 1280. This passage appears to say that some disparate-impact claims should fail even if the plaintiff could prevail under the standard burden-shifting framework. We cannot endorse this view.
When we adjudicate statutory claims, “our role is limited to interpreting what Congress may do and has done.” Patterson v. McLean Credit Union, 491 U.S. 164, 188, 109 S.Ct. 2363, 105 L.Ed.2d 132 (1989). While reasonable minds could disagree as to whether the FHA contemplates disparate impact liability, we have concluded that it does. See Larkin, 89 F.3d at 289; Arthur, 782 F.2d at 575. Nothing in the text of the FHA instructs us to create practice-specific exceptions. Absent such instruction, we lack the authority to evaluate the pros and cons of allowing disparate-impact claims challenging a particular housing practice and to prohibit claims that we believe to be unwise as a matter of social policy. Accord Salute, 136 F.3d at 312 & n. 23 (Calabresi, J., dissenting) (expressing “doubt whether the Seventh Circuit’s approach” in Knapp is “consistent with” the general use of the burden-shifting framework to analyze disparate-impact claims under the FHA).
The passage from Knapp also could mean something more limited: that if no disparate-impact challenge to a particular practice ever could succeed under the burden-shifting framework, then a court categorically may bar all disparate-impact challenges to that practice. This reading is consistent with our statement that “at least under some circumstances a violation of [the Fair Housing Act] can be established by a showing of discriminatory effect without a showing of discriminatory intent.” Arthur, 782 F.2d at 574 (quoting Arlington Heights II, 558 F.2d at 1290) (alteration in original). It is also consistent with the cases relied on by the Seventh Circuit in Knapp. In NAACP v. American Family Mutual Insurance Co., for example, the court held that insurers never can face disparate-impact liability for “charging higher rates or declining to write insurance for people who live in particular areas,” reasoning that “[i]nsurance works best when the risks in the pool have similar characteristics.” 978 F.2d at 290. Conducting an inquiry analogous to our inquiry at the third step in the burden-shifting framework, it concluded that the strength of the insurer’s interest in declining to insure in certain areas outweighed the strength of any possible disparate impact the policy could have. See id. at 290-91.
Likewise, in Village of Bellwood v. Dwivedi, the court concluded that claims *376of “racial steering” were not subject to disparate-impact analysis because racial steering necessarily involves disparate treatment, not a facially neutral policy that has a discriminatory effect. 895 F.2d at 1533. We agree that categorical bars are justified when they result from a generalized application of the burden-shifting framework. We see no reason to require courts to engage in the Sisyphean task of working through the burden-shifting framework in each individual case when the plaintiff has no chance of success (for example, when a plaintiff brings a disparate-impact challenge to a landlord’s decision to charge rent).
We do not believe, however, that it is impossible for a disparate-impact challenge to a landlord’s withdrawal from Section 8 to succeed under the burden-shifting framework. Consider the strongest possible disparate-impact challenge to a landlord’s withdrawal from Section 8: All of the non-Section 8 tenants at a housing complex are white. The landlord decides to participate in the Section 8 voucher program. A number of Section 8 participants move in, including the plaintiff. All are black. The landlord quickly decides to withdraw from Section 8 and refuses to renew the leases of existing Section 8 tenants. He explains his decision by saying simply that he no longer wished to participate in the program, without claiming that his participation cost him money, time, business opportunities, or any other benefit.
At the first stage of the burden-shifting framework, the hypothetical plaintiff could state a prima facie case: she could identify a challenged practice (the landlord’s withdrawal from Section 8) and present strong statistical evidence that the practice had a disparate impact on blacks. At the second stage of the burden-shifting framework, the landlord could offer a business reason for his decision: he wanted out of the Section 8 program. At the third stage, then, the plaintiff would have to show either that the landlord’s explanation was a pretext for discrimination or that an alternate practice would serve the same business ends with less discriminatory effect. To evaluate the plaintiffs showing, we would consider the strength of her evidence of disparate impact and the strength of the business interests stated by the landlord. We have little doubt that a reasonable fact-finder presented with this hypothetical set of facts could conclude that the landlord’s withdrawal from Section 8 was pretextual, and therefore that the plaintiffs claim should survive summary judgment under the burden-shifting framework.
Graoch makes two arguments that could undermine this conclusion. First, it contends that because Section 8 is a voluntary program we should place exceptional weight on a landlord’s interest in being able to withdraw without the threat of litigation. Yet to say that Section 8 participation is “voluntary” is only to say that a landlord does not break the law by declining to participate.5 As such, Graoch’s *377claim proves too much: almost every action that could create disparate-impact liability under the FHA is voluntary. Cf. Buckeye, 263 F.3d at 631-32 (vacated) (approving building site plans); Maki, 88 F.3d at 364 (varying of rent based on number of occupants). The mere fact that a landlord often can withdraw from Section 8 without violating the terms of Section 8 or the FHA does not mean that withdrawal from Section 8 never can constitute a violation of the FHA.
Second, Graoch claims that there is no coherent way to permit disparate-impact challenges to a landlord’s withdrawal from Section 8 without also permitting such challenges to a landlord’s non-participation in Section 8. It reasons that “[t]he actions of both non-participating and participating owners have the same impact on minorities” and therefore that “no principled way exists” to distinguish the two groups. Knapp, 54 F.3d at 1280. While we agree that a landlord should never face disparate-impact liability for non-participation in Section 8, we do not think that withdrawal and non-participation are functionally identical. Withdrawal affects an identifiable group: tenants receiving Section 8 assistance. Non-participation could be said to have a theoretical impact on members of a protected class, but the size and composition of the affected group is indeterminate.
Moreover, the FHA does not impose liability based solely on the existence of a disparate impact. It imposes liability only when a plaintiff can meet the ultimate burden of persuasion by showing either that the landlord’s business reason for a practice is a pretext for discrimination or that an alternative practice would serve the same business goal with less discriminatory effect. That burden is harder to meet in a challenge to non-participation than in a challenge to withdrawal. A nonparticipating landlord presumptively can appeal to his interests in not wanting to spend time learning about the program and not wanting to become entangled in government bureaucracy, while a withdrawing landlord who fails to cite any reason why participation in Section 8 hurt his business cannot do so. And an unexplained withdrawal from Section 8, combined with strong evidence of disparate impact, might permit a fact-finder to infer that a landlord entered Section 8 expecting to attract Section 8 tenants of one race, then withdrew based on discriminatory animus when she instead attracted tenants of a different race. Thus, our view that it is possible to bring disparate-impact challenges to withdrawals from Section 8 is consistent with our view that one cannot bring a disparate-impact challenge to Section 8 non-participation.
IV
Because we reject a categorical rule against disparate-impact challenges to withdrawals from Section 8, we must address whether the Commission can state a prima facie case of disparate impact discrimination against Graoch in this case. To state a prima facie case of disparate-impact discrimination under the FHA, a plaintiff must (1) identify and challenge a specific housing practice, and then (2) show that the practice had an adverse effect on members of a protected class by offering statistical evidence of a kind or degree sufficient to show that the practice in question has caused the adverse effect in question. The Commission plainly satisfied the first requirement by identifying and challenging Graoch’s practice of no longer accepting Section 8 vouchers. We conclude that it did not satisfy the second requirement, however, because it failed even to allege any facts making the statistical comparison necessary to determine *378whether Graoch’s withdrawal from Section 8 had a disparate impact on blacks.
In Arthur, we explained that there are “two types of discriminatory effects which a facially neutral housing decision can have”:
The first occurs when that decision has a greater adverse impact on one racial group than on another. The second is the effect which the decision has on the community involved; if it perpetuates segregation and thereby prevents interracial association it will be considered invidious under the Fair Housing Act independently of the extent to which it produces a disparate effect on different racial groups.
782 F.2d at 575 (citing Arlington Heights II, 558 F.2d at 1290). In a disparate-impact case under the FHA, a plaintiff must present statistical evidence showing the existence of at least one of the two effects to state a prima facie ease. Here, the Commission presented three pieces of data: seventeen of the eighteen families who received Section 8 assistance and lived at Autumn Run when Graoch announced its withdrawal are black; as of 2003, 6,270 of the 8,849 Jefferson County residents receiving Section 8 vouchers were black; and as of the 2000 census, 18.9% of Jefferson County residents were black and 24% were members of black households. It made no other allegations regarding the racial makeup of Autumn Run or the surrounding community.
To determine whether a plaintiff can show the first type of discriminatory effect described in Arthur, “[t]he correct inquiry is whether the policy in question had a disproportionate impact on the minorities in the total group to which the policy was applied.” Betsey v. Turtle Creek Assocs., 736 F.2d 983, 987 (4th Cir.1984) (emphasis added); cf. Isabel v. City of Memphis, 404 F.3d 404, 412 (6th Cir.2005) (applying essentially the same inquiry in a Title VII case). In Betsey, a landlord managing several adjacent buildings adopted a policy banning children from Building Three. 736 F.2d at 985. The policy led to the eviction of 54.3% of the non-white tenants in Building Three, but only 14.1% of white tenants. Id. at 988. Relying on the discrepancy between those percentages, the court held that the plaintiffs had made a prima facie case even though “the percentage of blacks at [the apartment complex as a whole] continues to exceed by a substantial margin both the percentage of black renters in the election district in which [the complex] is located as well as in Montgomery County as a whole.” Id. at 986-87. This approach respects the commonsense notion that a policy applied to a population that is 95 percent black does not have a disproportionate adverse effect on blacks even if 90 percent of the people it burdens are black.
Here, Graoch applied its “no Section 8” policy to all residents of Autumn Run. Consequently, to determine whether the policy had a greater adverse effect on blacks than whites, we should compare the percentage of blacks among the Section 8 tenants whose leases were not renewed (about 94 percent) to the percentage of blacks among the total pool of tenants at Autumn Run. Yet the Commission has not presented any data regarding the total tenant pool and has not even alleged that there are more whites among the non-Section 8 tenants than among the Section 8 tenants. The only information we have about Autumn Run’s non-Section 8 tenants is an isolated comment by Graoch’s lawyer that “the tenants at Respondent project Autumn Run Apartments were 90% minority and 10% white.” If we consider this information (which the Commission does not dispute), it indicates that the racial mix of Autumn Run’s Section 8 and non-Sec*379tion 8 tenants was essentially the same, overwhelmingly non-white. Because the Commission has not even alleged — and very likely could not show — that Graoch’s withdrawal from Section 8 harmed a disproportionate percentage of the black tenants at Autumn Run, we conclude that the Commission has not stated a prima facie case based on the first type of discriminatory effect. There can be no prima facie case of racial disparate impact without a showing of adverse impact on a racial group. For all that the Commission alleges, any vacancies caused by non-renewal of Section 8 leases to black tenants will be filled by black non-Section 8 tenants, who already fill the very large majority of Autumn Run’s apartments.
The Commission has not stated a prima facie ease based on the second type of discriminatory effect either. It has never claimed that Graoch’s withdrawal from Section 8 had a segregative effect nor alleged any facts from which we could infer a segregative effect: it has not provided any information about the racial makeup of the community surrounding Autumn Run or about the likely effect of withdrawal on the racial makeup of Autumn Run itself.
Because the Commission cannot meet its burden at the first step of the burden-shifting framework, its case cannot go forward. Graoch is not required to present a legitimate business reason justifying its decision to withdraw from Section 8.
V
For the foregoing reasons, the district court’s grant of summary judgment in favor of Graoch is affirmed.
. Graoch named only the Commission as a defendant; the FHC moved to intervene as of right, but the district court denied that motion.
. Although Graoch is the plaintiff in this declaratory judgment action, when discussing a generic claim we use "plaintiff” to refer to the party alleging a violation of the FHA and "defendant” to refer to the party that allegedly violated the FHA.
. We followed the same procedure in another case involving a disparate-impact claim against a governmental defendant under the FHA. See Buckeye Cmty. Hope Found. v. City of Cuyahoga Falls, 263 F.3d 627, 640 (6th Cir.2001). The Supreme Court subsequently granted certiorari in that case, reversed our decision on other grounds, and vacated our decision regarding the plaintiff's FHA claim because the plaintiff abandoned that claim before the Supreme Court. See City of Cuyahoga Falls v. Buckeye Cmty. Hope Found., 538 U.S. 188, 199-200, 123 S.Ct. 1389, 155 L.Ed.2d 349 (2003).
. Even were we to accept this argument, landlords still could face liability under a disparate-treatment theory, as opposed to a disparate-impact theory, if they refused to accept Section 8 vouchers from some tenants (who were predominantly black) but accepted Section 8 vouchers from others (who were predominantly white).
. Nothing in the language of Section 8 indicates that, in making the program voluntary, Congress intended to do more than give landlords the option not to participate. Although Congress created the Section 8 program six years after passing the FHA, see Housing and Community Development Act of 1974, Pub.L. No. 93-383, 88 Stat. 653 (codified as amended at 42 U.S.C. §§ 1437-371), it did not include language indicating that Section 8 landlords should be exempt from any FHA requirements. Nor did Congress add any exceptions to the FHA when it later amended Section 8 "to minimize the burdens of Section 8 participation in order to make the program more attractive to landlords.” Salute, 136 F.3d at 298 (discussing the passage of Pub.L. No. 104-134, § 203(a), (d), 110 Stat. 1321 (1996)).