The Fair Housing Act prohibits discrimination in housing because of 'race, color, religion, or national origin.' 42 U.S.C. § 3604 (1970). Plaintiffs argue that defendants by utilizing the 90% rule have violated this prohibition. They reason that the 90% rule excludes all public assistance recipients except for the very small number who can obtain an acceptable co-signer, that a large majority of public assistance recipients in New York City are black or Puerto Rican, and that therefore the use of the 90% rule is racially discriminatory. The premise of plaintiffs' argument is that '(w) elfare recipiency . . . must be seen as the 'functional equivalent' of race.' Appellee's brief at 36. Such an equivalency between race and income has been rejected by the Supreme Court in James v. Valtierra, 402 U.S. 137, 91 S. Ct. 1331, 28 L. Ed. 2d 678 (1971). Plaintiffs in that case, relying on a line of reasoning similar to that advanced by plaintiffs here, challenged an article of the California state constitution which provided that no low-rent housing projects should be developed, constructed or acquired by a state public body until the project was approved by the majority of those voting at a community election. Despite implicit recognition of the correlation between racial minority and low income, the Court refused to equate the two factors. The Court, in upholding the validity of the provision, said: 'The Article requires referendum approval for any low-rent public housing project, not only for projects which will be occupied by a racial minority.' Id. at 141, 91 S.Ct. at 1333.

As in Valtierra, supra, the rule under consideration here cannot be said to rest on distinctions based on race. See 402 U.S. at 141, 91 S. Ct. 1331. While blacks and Puerto Ricans do not have the same access to Lefrak apartments as do whites, the reason for this inequality is not racial discrimination but rather the disparity in economic level among these groups. While a showing of a disproportionate effect on non-whites is sufficient to require application of the compelling state interest standard in the context of an equal protection challenge to government action, see, e.g., Hunter v. Erickson, 393 U.S. 385, 391--392, 89 S. Ct. 557, 21 L. Ed. 2d 616 (1969), such an analysis is inappropriate in the context of a purely private action asserting a claim of racial discrimination. A businessman's differential treatment of different economic groups is not necessarily racial discrimination and is not made so because minorities are statistically overrepresented in the poorer economic groups. The fact that differentiation in eligibility rates for defendants' apartments is correlated with race proves merely that minorities tend to be poorer than is the general population. In order to utilize this correlation to establish a violation of the Fair Housing Act on the part of a private landlord, plaintiffs would have to show that there existed some demonstrable prejudicial treatment of minorities over and above that which is the inevitable result of disparity in income. Cf. James v. Valtierra,supra. Just as this court will not impose even on the government an affirmative duty to construct low-income housing when the decision not to build is not racially motivated, Citizens Committee for Faraday Wood v. Lindsay,507 F.2d 1065, 1071 (2d Cir. 1974); Acevedo v. Nassau County, 500 F.2d 1078, 1080--1081 (2d Cir. 1974), so we will not impose an affirmative duty on the private landlord to accept low income tenants absent evidence that his motivation is racial rather than economic in origin.

The conclusion that defendants' seemingly neutral rule is in fact aimed at excluding a racial minority finds no support in the record. The percentage of blacks in appellants' apartments (about 19.8%) closely approximates the percentage of blacks in the population of New York City (21%). There is no claim that black public assistance recipients are treated differently from white public assistance recipients. Nor is there any evidence that public assistance recipients have been excluded from Lefrak apartments for any reason other than inability to meet the financial standard. On the other hand, the evidence indicates that when the financial standard could be met as, for example, by obtaining an acceptable co-signer, public assistance recipients, including those belonging to minority ethnic groups, were rented apartments on the same terms as were any other qualifying applicants. To use ethnic distribution of tenants excluded from appellants' apartments as a basis for inferring discriminatory practices in this situation is like imputing discriminatory motives to producers of high priced goods because such goods more often find their way into the hands of wealthier white consumers than into the hands of the poor. As this court has said in the context of the equal protection clause, '(t)he mere fact that a requirement, otherwise proper, may have a greater impact on the poor, does not render it invalid. . . .' English v. Town of Huntington, 448 F.2d 319, 324 (2d Cir. 1971).

A landlord in the private sector6 is entitled to choose whom he will accept as tenants as long as he does not discriminate on one of the statutorily condemned bases. Certainly he may seek assurance that prospective tenants will be able to meet their rental responsibilities. '(T)here is no requirement that welfare recipients, or any other individuals, may secure apartments . . . without regard to their ability to pay.' Male v. Crossroads Associates, 469 F.2d 616, 622 (2d Cir. 1972).

Plaintiffs claim that the 90% rule, and indeed any economic standard computed on a percentage-of-income basis, is an inappropriate measure of a public assistance recipient's rent-paying ability because, unlike the working person the amount which a recipient receives for his non-shelter needs remains the same regardless of the amount of his shelter allowance7 and because increased shelter allowances can be obtained by recipients if approved by the New York City Department of Social Services. While it may be true that a public assistance recipient's ability to pay rent is related to the Department of Social Services' willingness to approve shelter allowances, the private landlord in choosing his tenants is free to use any grounds he likes so long as no discriminatory purpose is shown. See Madison v. Jeffers, 494 F.2d 114, 116--117 (4th Cir. 1974); Pughsley v. 3750 Lake Shore Drive Cooperative Building, 463 F.2d 1055, 1056 (7th Cir. 1972). His choice is not limited by any obligation to accommodate a special class of low income applicants.

Plaintiffs' reliance on Griggs v. Duke Power Co., 401 U.S. 424, 91 S. Ct. 849, 28 L. Ed. 2d 158 (1971) and other cases interpreting the Fair Employment Act, 42 U.S.C. § 2000e et seq. (1970) is misplaced. The 'business necessity' test developed in that context, whereby employers must demonstrate the business necessity of employment tests which have an unequal impact on minority job applicants, has never been applied in any Fair Housing Act case, either public or private, and we find it to be inapposite here. Cf. Jefferson v. Hackney, 406 U.S. 535, 549 n. 19, 92 S. Ct. 1724, 32 L. Ed. 2d 285 (1972).

Plaintiffs have contended that the 90% rule is unevenly applied in that the income of welfare recipients was, in effect, dismissed out of hand as insufficient while the income of the working family was carefully explored and all non-cash bonuses included in order to enhance the possibility of acceptance. Not only is this contention irrelevant to the issue of racial discrimination but defendants have denied that their rule is thus applied. In any event, plaintiffs' contention ignores the fact that even if liberal allowance is made for non-cash benefits and increased shelter allowances, welfare recipients would still have to pay a disproportionate amount (from 47.8% to 64.8%) of their income for rent in order to live in appellants' apartments. Conversely, if an individual could meet the minimum income level necessary to comply with the 90% rule, calculated at trial to be $10,500 per annum, he would be precluded from remaining a welfare recipient. Consequently, defendants' so-called 'uneven application' of their rule, if it occurred, is merely an acknowledgment of irrefutable mathematical facts and is certainly not indicative of racially discriminatory behavior.

Similarly because there has been no finding of racially-motivated discrimination, appellants have not violated the Civil Rights Act of 1866, 42 U.S.C. § 1982 (1970). See Jones v. Alfred H. Mayer Co., 392 U.S. 409, 421, 88 S. Ct. 2186, 20 L. Ed. 2d 1189 (1968); Madison v. Jeffers, 494 F.2d 114, 116--117 (4th Cir. 1974); Pughsley v. 3750 Lake Shore Drive Cooperative Building, 463 F.2d 1055, 1056 (7th Cir. 1972).

Accordingly, the judgment of the district court is reversed.

With due respect, the majority opinion seems to me to be grounded upon the erroneous concept that, in order to establish a violation of the Fair Housing Act, 42 U.S.C. § 3601 et seq., direct evidence of a racially discriminatory motive or purpose on the part of the alleged violator must be adduced. In my view such proof is not required.

This case should be governed by the principle, firmly established by the Supreme Court in its interpretation and enforcement of analogous civil rights legislation, to the effect that, where a facially neutral practice has a serious and substantial de facto discriminatory impact, it prima facie violates a statutory prohibition against racial discrimination unless the alleged violator can show that the practice is necessary for non-racial reasons. Griggs v. Duke Power Co., 401 U.S. 424, 91 S. Ct. 849, 28 L. Ed. 2d 158 (1971). In Griggs the question was whether an employer's use of employeetesting procedures which had the effect of excluding a disproportionate number of Negroes from employment violated the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq. (Title VII) in the absence of proof of a discriminatory intent on the part of the employer. In holding that the proof was sufficient to establish such a violation, Chief Justice Burger, speaking for a unanimous court, stated:

'The Act proscribes not only overt discrimination but also practices that are fair in form, but discriminatory in operation. The touchstone is business necessity. If an employment practice which operates to exclude Negroes cannot be shown to be related to job performance, the practice is prohibited.

'We do not suggest that either the District Court or the Court of Appeals erred in examining the employer's intent; but good intent or absence of discriminatory intent does not redeem employment procedures or testing mechanisms that operate as 'built-in headwinds' for minority groups and are unrelated to measuring job capability.

'The Company's lack of discriminatory intent is suggested by special efforts to help the undereducated employees through Company financing of two-thirds the cost of tuition for high school training. But Congress directed the thrust of the Act to the consequences of employment practices, not simply the motivation. More than that, Congress has placed on the employer the burden of showing that any given requirement must have a manifest relationship to the employment in question.

'Nothing in the Act precludes the use of testing or measuring procedures; obviously they are useful. What Congress has forbidden is giving these devices and mechanisms controlling force unless they are demonstrably a reasonable measure of job performance'. 401 U.S. at 431, 432, 436, 91 S. Ct. at 853, 854, 856. (Emphasis supplied)

See also Olzman v. Lake Hills Swim Club, 495 F.2d 1333, 1340--1341 (2d Cir. 1974) (holding that a facially neutral rule with respect to club guest privileges prima facie violates the Civil Rights Act of 1964, § 201(a) and (e), 42 U.S.C. § 2000a(a), (e), where its effect is to weed out Black guests, and that the club must bear the burden of showing that the rule has a reasonable non-racial purpose); cf. Chance v. Board of Examiners, 458 F.2d 1167 (2d Cir. 1972); Dean v. Ashling, 409 F.2d 754 (5th Cir. 1969).

Congress, in passing the Fair Housing Act, sought to eliminate artificial and arbitrary barriers to housing that operate to discriminate against persons because of their race, just as it had through Title VII acted to remove similar racially discriminatory barriers to employment. It must be recognized that if these laws were to depend for their effectiveness upon proof of the landlord's subjective intent, which is rarely obtainable, they would largely be rendered useless. Racial discrimination cannot, of course, be condoned because it is accomplished through a sophisticated or indirect method, cf. Lane v. Wilson, 307 U.S. 268, 275, 59 S. Ct. 872, 83 L. Ed. 1281 (1939). Nor should it be excused on the theory that it is the product of thought-lessness rather than willfulness. In either event, the harmful effect is the same. Enforcement of the anti-discrimination provisions of the Fair Housing Act must therefore be judged by objective standards, in line with the principle that the Act be given a 'generous construction,' Trafficante v. Metropolitan Life Insurance Co., 409 U.S. 205, 212, 93 S. Ct. 364, 34 L. Ed. 2d 415 (1972). In recognition of the old adage that actions speak louder than words, to make out a prima facie violation of the Act it should be sufficient to show that the challenged practice excludes a disproportionately high percentage of minority persons as compared with non-minority. The burden of going forward with a non-racial justification should then shift to the person using the practice.

Plaintiffs in this case have not suggested that 'welfare recipients, or any other individuals, may secure apartments . . . without regard to their ability to pay,' Male v. Crossroads Associates, 469 F.2d 616, 622 (2d Cir. 1972). Nor does anyone question the importance to a landlord of a prospective tenant's payment of rent, upon which the landlord depends for the successful operation of his real estate enterprise. Toward that end the landlord, of course, may adopt reasonably appropriate economic standards or tests designed to assure the tenant's future ability to pay rent on an on-going basis. See, e.g., United States v. Grooms, 348 F. Supp. 1130, 1134 (M.D.Fla.1972). But where the formula has the effect of excluding a disproportionately high percentage of a minority group, the landlord should in fairness be prepared to demonstrate the business necessity of his facially neutral rule. 'It is no answer that defendants would have exploited whites as well as blacks,' Clark v. Universal Builders, 501 F.2d 324, 331 (7th Cir. 1974).

The Supreme Court's decision in James v. Valtierra, 402 U.S. 137, 91 S. Ct. 1331, 28 L. Ed. 2d 678 (1971), so heavily relied upon by the majority, is inapposite. That case did not deal with the interpretation of a federal civil rights statute, which is the issue before us, but with the constitutionality of a state law under attack as violative of the Equal Protection Clause, an entirely different issue requiring application of totally different principles.1 In deciding that a state's constitutional provisions for community referenda with respect to proposed housing projects did not violate the Equal Protection Clause, the Court was not concerned with statutory interpretation but with the constitutionality of a presumptively valid statute. It concluded that the community's interest in determining the level of local governmental expenditure justified the state constitutional amendment under attack, even though some persons might be disadvantaged by the mandatory referendum procedure. Here we are dealing with the entirely different question of determining the scope and meaning of a statute which expressly prohibits discrimination in housing because of 'race, color, religion, or national origin.' In my view, in order to give effect to Congress' objective, this legislation must be interpreted as prohibiting conduct which, according to objective standards, has the effect of discriminating because of race, unless a non-racial justification is shown. See United States v. Grooms, 348 F. Supp. 1130 (M.D.Fla.1972). Otherwise the law would become for the most part a dead letter, in view of the difficulty of proving discriminatory intent upon the part of the person alleged to have violated it.

Applying these principles here, plaintiffs made out a clear prima facie showing that the 90% Rule2 has had a disproportionately high racial impact. The evidence is undisputed that the effect of the 90% Rule is to exclude from appellant's apartments all but a handful of welfare recipients who apply. Indeed there was direct evidence that the rental agent for Lefrak apartments, Anthony Cuccia, told plaintiff Boyd, 'We don't rent to welfare recipients.' Jerry Richter, Vice-President of Lefrak, further advised plaintiff's attorney that Lefrak had a policy of not renting to public assistance recipients.

To exclude public assistance recipients in New York City is the equivalent of excluding minority persons. According to the 1970 census 77% of all welfare recipients in New York City were minority persons, the great majority of whom were Black and Puerto Rican. Since that time the figure has increased. A 1971 study, for instance, reveals that 90.1% of those receiving assistance under the program for Aid to Dependent Children are minority persons. A review of all categories of public assistance, see Goodwin v. Wyman, 330 F. Supp. 1038 (S.D.N.Y.1971), aff'd per curiam, 406 U.S. 964, 92 S. Ct. 2420, 32 L. Ed. 2d 664 (1972), indicates that minorities predominate substantially.

Although minority persons would, under the 90% Rule, continue to be eligible to rent appellants' apartments, they constitute but a small number of those, minority and non-minority, who are eligible. There was expert testimony to the effect that 92.5% of Black and Puerto Rican households would be excluded. Such testimony indicated that under the 90% Rule eligibility of white households in New York City would be four times as great as that of Black households and ten times as great as that of Puerto Rican households.3 Furthermore, accepting the fact that appellants have rented apartments to a substantial number of wealthier minority persons, not on welfare, who can satisfy the 90% Rule,4 the percentage of minority households in appellants' apartments would be much higher under a less stringent rule. To compound the disproportionately high racial impact of the 90% Rule, the evidence before the trial judge further disclosed that of the public assistance recipients who applied to appellants for apartments the percentage of non-minority appellants who were accepted was approximately twice the percentage of Black applicants accepted. The 90% Rule, therefore, operates as 'built-in headwinds' for minority groups.

Faced with the discriminatory consequences of their conduct appellants failed to offer any satisfactory non-racial explanation, such as business necessity. There was no showing, for instance, that experience in the rental of apartments of the type here under consideration had demonstrated that the 90% Rule was reasonably necessary to insure tenants' payment of rent and that there had been losses, substantial defaults, or failure to collect back rental payments under less stringent rules. Nor was there proof that welfare recipents as tenants have a greater incidence of rent failures or defaults than other tenants. Plaintiffs, in contrast, offered persuasive evidence that the 90% Rule was not adopted as a 'business necessity' measure. Fifty-six percent of all New York City households renting apartments in the Lefrak rental range here under consideration do not meet the 90% Rule, i.e., the household's weekly net income is not equal to at least 90% of the monthly rental of its apartment. Indeed, in New York City 47.4% of the white renters and 57.6% of the Black and Puerto Rican renters exceed that ratio. The trial judge further found that 53.1% of the white households renting apartments in the Lefrak rental range, 70.4% of the Black households, and 74.8% of the Puerto Rican households, do not satisfy the 90% Rule. Still there is no showing that landlords renting to these tenants have suffered losses or been faced with bankruptcy. Furthermore, the 90% Rule fails to take into consideration the fact that the amount allocated by the Department of Social Services ('DSS') to a welfare recipient for payment of his rent is not fixed at a specific figure but is equal to the recipient's actual rent when approved by DSS, and, in the event he defaults in the payment of his rent, the rent is thereafter paid by a 'two party' DSS check (i.e., payable jointly to the recipient and landlord), which does not affect the level and amount of subsistence separately paid to the recipient. Indeed DSS may even pay the rent directly to the landlord, N.Y. Social Services Law § 131--a(7). The welfare recipient's ability to pay, therefore, is not properly measurable by his or her aggregate income. Furthermore, the recipient's income includes non-cash benefits (e.g., food stamps, Medicaid)5 not available to the non-welfare tenant. Id., § 131--a. The ability of welfare recipients to pay Lefrak rents is also attested to by the existence, unknown to Lefrak, of some 461 welfare-recipient households in its apartments in 1972 and the fact that, out of 15,484 Lefrak apartments, only 108 dispossess notices were issued in 1972 and 43 in 1973.

Thus, according to the trial judge's findings of fact, which are fully supported and are not asserted by the majority to be clearly erroneous, the 90% Rule has a disproportionately high racially discriminatory impact and appellants have failed to show that the Rule is based on business necessity or other non-racial grounds. Application of the principles prescribed by the Supreme Court in Griggs therefore mandates affirmance of the district court's decision.