COFFIN, Chief Judge.
The major legal problem in this case is to try to fit the fact that owners of four federally subsidized low and middle income housing projects in Boston have to satisfy both federal and municipal authorities as to the reasonableness of their rents within the doctrinal pronouncements of the Supreme Court in Supremacy Clause decisions. This requires us to review the record as to the impact on plaintiffs of the procedures and decisions of the Boston Rent Board in reviewing requests for rental increases and to determine whether the district court’s conclusion that “an impermissible conflict” exists, 410 F.Supp. 1314, 1320 (D.Mass.1976), can stand.
I.
The history and longevity of this litigation are already formidable.1 The factual background is more fully set forth in the district court’s opinion, 410 F.Supp. at 1315-1319. These are two consolidated cases, in which the plaintiffs-appellees are [4]*4owners of large apartment complexes housing low and middle income tenants.2 Their housing developments are federally subsidized and are regulated and financed under § 221(d)(3) of the National Housing Act, 12 U.S.C. § 1715/(d)(3). Under this program, which is administered by the U. S. Department of Housing and Urban Development (HUD), and designed to assist private industry in providing housing for low and middle income families, 12 U.S.C. § 1715/ (a), owners receive insurance on mortgage loans, can borrow at below market interests rates, are subject to reduced equity requirements, and enjoy various tax advantages. See generally Hahn v. Gottlieb, 430 F.2d 1243, 1245-46 (1st Cir. 1970). In order to obtain these advantages, mortgagors must sign regulatory agreements with HUD which oblige the owners to charge tenants at prices “not exceeding” those in the approved schedules. Pertinent to these cases is HUD’s commitment in these agreements to approve rent schedules “necessary to compensate for any net increase, occurring since the last approved rental schedule, in taxes . . . and operating and maintenance expenses”. HUD’s general formula in arriving at approved rent increases is to allow coverage of mortgage retirement and debt service, operating expenses, and, in accordance with 24 C.F.R. § 221.532(a), a six per cent return on equity.
Beginning in 1970 the city of Boston initiated its own system of rent control which applied to § 221(d)(3) housing, amending the ordinance in some respects in 1972. This system required plaintiffs, after having received federal approval of a rent increase, to request the additional approval of the Boston Rent Board. In so doing, they were subjected to further delay, additional expense, more detailed investigation, somewhat different approaches to countable costs, an adversary hearing and possible appeal, and the possibility of either approval of a lower rent increase than that authorized by HUD or an absolute denial.
The Druker plaintiffs brought suit in 1971 and the Kargman plaintiffs later that year; both sets of plaintiffs sought declaratory and injunctive relief invalidating and proscribing enforcement of the Boston Rent Board orders. In 1971 the district court abstained in the Druker case, Druker v. Sullivan, 332 F.Supp. 1126 (D.Mass.1971), affirmed, 458 F.2d 1272 (1st Cir. 1972), on the ground that a possibly dispositive state law issue existed. This proved not to be the case. See 362 Mass. 874, 287 N.E.2d 801 (1972). We subsequently denied plaintiffs’ motion to reinstate the appeal and, instead, ordered the case remanded. We were somewhat ambivalent, recognizing the need for decision without avoidable delay, but hesitant to decide a question of importance to the federal government, the landlords, their tenants and the municipality on the existing record. The summary judgment record gave us, we felt, less than 20-20 vision and we lacked the benefit of any expression of views from HUD itself. We were, as we said, seeing “Hamlet” without the prince.
We therefore remanded to obtain a fuller record and a judgment based thereon. We indicated that:
“ . . . the present record is meager with respect to the purposes, regulations, procedures, accounting, and practical workings of the federal and municipal systems of rent control. It would seem to us, in the light of ITT v. Minter, 435 F.2d 989 (1st Cir. 1970), that such a ventilation of the design and operations of the two regulatory schemes would be helpful in determining if they can co-exist.”
We also renewed a prior district court invitation to HUD that it intervene. Finally, we operated on the naive assumption that both “ventilation” and decision could take place in time for us to review the preemption issue on the merits, if there were an appeal, by the summer of 1973.
Four years later we reach the issue. We are not in a position to say why this fruit [5]*5took so long to ripen. HUD finally did intervene in the Druker case. The Druker and Kargman cases were consolidated. Depositions of a HUD official and a Boston Rent Board official were taken. Fairly voluminous exhibits were prepared. And, finally, the consolidated cases were tried for several days in 1975.
The plaintiffs’ case was directed to underlining the different approaches to setting rent ceilings pursued by HUD and Boston, and their differing results. There were clearly differences in procedure, standards, time-for-decision, and, in the cases at bar, in results.
Defendants conceded differences in approach, but claimed that the approach of the Boston Rent Board, involving closer scrutiny of increased costs, was more realistic, did not jeopardize the financial stability of landlords participating in § 221(d)(3), did not require substantially more time, and more often than not resulted in approving the HUD rent figure.
The district court noted that HUD and the Boston Rent Board differed in their approaches to fixing maximum rents in three ways: (1) While HUD allowed a fixed vacancy rate of seven per cent, Boston restricted vacancy allowances to the actual rate or five per cent whichever might be lower; (2) HUD would treat as operating expenses, countable in the year of expenditure, some items which Boston would treat as capital expenses, allocated over the life expectancy of the item in question; (3) Boston would disallow certain expenses which HUD could accept. All of these combined to yield lower rent ceilings than those HUD’s formula produced. A fourth factor of a different kind was recognized: the fact that the Boston procedure called for adversary proceedings and access to state court review. While these differences existed, the court found that Boston Rent Board procedures alone did not cause any mortgage defaults.
In identifying its standard for determining whether the federal law was preemptive, the court, citing Florida Lime & Avocado Growers, Inc. v. Paul, 373 U.S. 132, 142, 83 S.Ct. 1210, 1217, 10 L.Ed.2d 248 (1963), held out two measures: “either that the nature of the regulated subject matter permits no other conclusion, or that the Congress has unmistakably so ordained.” The court had no difficulty in saying there had been no expression of a clear Congressional design. We agree. Indeed, the statute provides that eligible mortgagors must be subject to rent regulation, either under state or local law or such other method as the Secretary deems appropriate. 12 U.S.C. § 17151 (d)(3). In pursuing further analysis, the court seemed concerned not so much with the nature of the subject matter, or occupation of the field, as it was with, the existence of actual conflict between local rent control and the federal program, noting that the “question for this Court is whether the operation of local rent control has frustrated the purposes” of the Act. 410 F.Supp. at 1320. On this issue it appropriately cited our reference in ITT v. Minter, supra, 435 F.2d at 993, to the standard set forth in Southern Pac. Co. v. Arizona ex rel. Sullivan, 325 U.S. 761, 766, 65 S.Ct. 1515, 89 L.Ed. 1915 (1945); i. e., that preemption must be based on a finding that the local law “palpably infringes” federal policy. The district court then concluded that the differences between the two rent-setting formulas — in allowance for vacancies, allowance of certain operating expenses, and allocations to capital or operating expenses — which yielded a lower Boston-approved rent ceiling for the four projects in this case and differences in timing and complexity between HUD’s ex parte and Boston’s adversary review proceedings, created an “impermissible conflict”,3 410 F.Supp. at 1320.
[6]*6The fact that these differences exist merely pose the preemption question — they do not answer it. See Florida Lime & Avocado Growers, supra, 373 U.S. at 141, 83 S.Ct. 1210. Our task on review is to determine whether the difference between approved rent ceilings, plus the cost and delay encountered in the Boston procedure, so significantly frustrated federal interests in the operation of the § 221(d)(3) program that Boston’s traditionally strong interests in local rent control must yield. We shall first review the evidence and then survey the teachings of the Supreme Court for guidance in determining whether the conflicts revealed by this record rise to a constitutional threshold of impermissibility.
II.
We first emphasize that this is not a case posing the difficulties, nationwide, of federally subsidized housing co-existing with local rent control. Specifically, this is not a case where a national administrator, surveying all of his problems of dealing with local rent control, has concluded that the federal program nationally is in jeopardy because of varying approaches to setting rent ceilings, or administrative entanglements, or the risk that local regulation would discourage participation in the program. Throughout the period of this case HUD’s position was one of trying to coexist. Letters in evidence articulated that policy. At the end of the period covered by this litigation, the policy came under review. At some intermediate point the decision was made to intervene in the Druker — not the Kargman — case. Indeed, at the time of intervention, no clear policy existed; later, as HUD’s concerns increased and criteria for participating in individual cases were developed, a post hoc review of the Druker intervention was made and it was found to have met the criteria.4 Only after this determination was made did HUD take the position, formalized in a regulation, 24 C.F.R. § 403, that local rent control was preempted by the necessities of running the federal program. The efficacy of that regulation is the subject of another case, City of Boston v. Secretary of HUD, C.A. No. 75-902-C (D.Mass.), and we express no views on it. The record now before us is restricted to four housing projects, HUD having intervened on an ad hoc basis in a case involving one of them while still in the process of reviewing its national position.
We note also our agreement with the district court’s view that “the allegations of HUD’s complaint are no substitute for competent evidence of preemption.” 410 F.Supp. at 1321. Whatever power HUD may have to preempt local rent control by regulation, it cannot preempt by the simple act of intervening in on-going litigation. By filing a pleading in the district court, it was submitting its case to the court for fact finding and legal conclusions. And we think it follows that HUD and the plaintiffs were required to prove their case with hard evidence of conflict, and not merely with unsupported pronouncements as to HUD’s “policy”. It is on the basis of the record evidence in this case that we must determine whether there is an actual, impermissible conflict between the local and federal law.
The landlords and HUD have throughout this litigation taken the position that any figure lower than the HUD-approved rent by definition defeats HUD’s purposes and constitutes proof per se of [7]*7“palpable” conflict. The district court agreed, finding that, even though the regulation was framed in terms of maximum rent the rents set by HUD were in fact “no higher than necessary to provide for debt service, expenses and investor return. Thus from a practical standpoint the rents set were the minimum necessary to effect the goals of the National Housing Act.” 410 F.Supp. at 1320. We have carefully examined the evidence in this case, which was largely in the form of documents and depositions, and we are convinced that the district court’s finding on this critical point, which was the foundation of its analysis of conflict, is clearly erroneous.
We believe that HUD may be entitled to considerable leeway in protecting its financial interest, and we might be inclined to defer to its judgment as to the minimum rent necessary to do this and also to assure landlords a promised rate of return. But the contention that HUD’s formula is a minimum of this sort is belied by the actual practice during the period involved here. First, although projects in Boston at this time usually had a waiting list, a seven per cent vacancy rate was allowed in calculating the HUD rent, in effect creating a “cushion” in the rent for hypothetical losses of rental income. Second, if the maximum rent “thrown-off” by HUD’s formula exceeded a landlord’s rental increase request, only the lesser amount requested would be authorized. Third, if the formula yielded a substantial rent increase, it would be granted in phases or installments. The built-in cushion, and the fact that in certain circumstances the authorized rent would be less than the figure yielded by the formula, suggest that HUD did not view the latter as the minimum necessary to protect its financial interest, or the minimum a landlord was entitled to under federal law. Finally, HUD’s policy at this time of cooperative coexistence and selective intervention in rent control disputes seriously undercuts the claim that any rent ceiling differing from HUD’s spelled frustration or defeat for the federal interests.
The only evidence that the HUD formula was actually a minimum was the testimony of HUD’s Director of Loan Management. He asserted that the formula yields a figure “that would have to be considered the amount necessary to carry the project and give the owner his rate of return”, but admitted that the sole basis for this conclusion was his knowledge as to how the formula works. He testified, consistent with his sole reliance on the formula, that any rent lower than that in the regulatory agreement would place the project in jeopardy, but admitted that he could not “make an across the board statement”. Indeed, the fact that HUD did not intervene in the Kargman case, and that intervention was based on a separate case-by-case analysis, makes clear that there was no across-the-board policy that the formula produced a minimum rent.
Thus, we think that the district court’s conclusion that HUD rents in general were “no higher than necessary to provide for debt service, expenses, and investor return”, 410 F.Supp. at 1320, was not supported by the evidence. While the Director’s opinion tends to support the finding, our review of the entire evidence leaves us with a “definite and firm conviction that a mistake has been committed.” United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948).
Since we do not view the HUD rent as a “minimum”, the mere fact that Boston’s rents were lower does not, of itself, present an impermissible actual conflict with federal law. Cf. Florida Lime & Avocado Growers, supra, 373 U.S. at 143, 83 S.Ct. 1210. We must therefore turn to the evidence of what happened to these projects and why, to determine whether any significant federal interest was impaired in these specific instances.
While the landlords are in default on their mortgage payments for a number of these projects,5 the district court was care[8]*8ful to say that it could not attribute the result to local rent control. 410 F.Supp. at 1319. We do not look on the district court’s statement that “Rent Board schedules . did not yield rents sufficient to cover plaintiff’s debt service, expenses and investor return”, id. at 1320, as a finding of fact, since it simply followed from the erroneous premise that HUD rents were “no higher than necessary” to provide for these three items. Indeed, the court found that the Rent Board’s calculations are aimed at providing the landlord with a fair return on his investment. Appellees deliberately did not try this case on a factual basis, endeavoring instead to show that the different approaches were generically antithetical.6 Consequently, all we can say the record supports is speculation that local rent control may have played some part in the difficulties in which the landlords found themselves.
The evidence showed that while at one point the Board may not routinely have allowed, in its calculations, the same six per cent return on equity that HUD provided for, the Board had been required by the Massachusetts Housing Court to honor the standard; that in most cases, the Board eventually approved the HUD endorsed increases; 7 that in cases where the Board reached results different from HUD, there were, not infrequently, local code violations on the part of landlords.8 We also note that the Druker plaintiffs have stipulated that there was no basis to attack the decision of the Board (allowing only $5 of a second increase of $11 approved by HUD) on grounds that the procedure was unfair or the findings were unsupported. One of the findings was that the rent adjustment allowed by the Board met both the federal standard of allowing six per cent return on equity and the Board policy of allowing a fair net operating income. While the Kargman plaintiffs did not file such a stipulation, they made no effort to appeal the Board’s decisions as to their properties. All of this suggests that the Board’s rent levels in these projects were reasonable, and that the landlord’s difficulties were not caused by inadequate Board-approved rents.
The major items of evidence offered to establish that Boston rent control had adverse impacts on the HUD program were a list of the ten cities with the most defaults in mortgage repayments, a handwritten summary of problems encountered in Boston phoned in to HUD from its area office, an assessment prepared by HUD on a nationwide basis of the reasons for default, and testimony by HUD’s Director of the Office of Loan Management. The first two items were objected to as documents prepared for this litigation by staff personnel who were not present for cross-examination as to the data, methods, or assumptions used in developing the documents. But even if they were correctly admitted into evidence, their thrust is far from clear. [9]*9While Boston placed third highest in the number of housing units as to which mortgage payments were in default, the dollar amount was small, there was no attempt to assess the causes of defaults and there was no comparison of default records in cities with and without local rent control. The handwritten summary of reasons for difficulties in Boston, dubiously admissible, was based on undocumented conclusions of the Boston HUD office staff, and asserted that of the 44 projects which were in default, HUD had approved rent increases for 30 but that the Board had not approved 12 of these.9 HUD’s Director of Loan Management conceded that he did not know whether the Board’s refusal to approve rent increases was a primary cause, a secondary cause, or no cause at all for any of the defaults. The only evidence purporting to identify a cause-effect relationship between local rent control and defaults — the only “factual investigation” to support HUD’s determination that local rent control in Boston interferes with the operation of HUD’s insured projects — was an analysis of 846 defaults during a four month period in 1973. Of ten major categories of causes for default, rent problems accounted for only five per cent of the total; in particular, local rent board disapprovals of rent increases and local “rent freezes” accounted for .75 per cent of the total number of defaults.
On the basis of this evidence, the district court concluded, quite properly, that the Rent Board’s procedures did not necessarily result in mortgage defaults. It did find, however, that the additional layer of regulation and lower maximum rents had “some adverse financial consequences”. 410 F.Supp. at 1319. The court also noted that “the matter was exacerbated by delay inherent in the local system” 410 F.Supp. at 1320. While observing that both plaintiffs and defendants contributed» to delay, the court stated: “Regardless of the reasons, however, it took the local board up to six months to process requests for increases. These time-consuming procedures necessarily had adverse financial consequences for plaintiffs.” Id.
This reference to delay is general and atmospheric. We are given no finding of the usual delay or any demonstrable effect on the plaintiffs or, more to the point, on HUD’s operations. While the evidence did show that the longest time the Board took to process a rent increase request was from seven to eight months, and the shortest time 19 days, at the time of trial the average elapsed time between filing and decision was three months.10 By early 1973 the Board’s policy had changed to allow rental increase requests to be filed before final approval by HUD, although Board decision would in such cases await final HUD action. What effect this would have on total delay time is not revealed by the record. Apparently no landlords responded to the invitation. As for HUD’s performance, the Boston Area office had, up to some time in 1972, only one person processing rent increase applications. HUD’s Director of Loan Management admitted that a 1972 report revealed that he was “overwhelmed with work and not able to meet our requirements in a timely manner.” According to a HUD official from the Boston office, there were extended processing delays until December, 1973; beginning in January, 1974, HUD required processing to be completed within 30 days.
One final comment about the operations of HUD and the Board may be noted. There was no evidence of routine or frequent communications between the two organizations. The Board had consulted [10]*10HUD about its basic standards on two occasions. HUD -.had never approached the Board on differences in approach or results. A recent HUD directive had suggested more consultation.
We leave our review of the evidence as to these four projects with an ambivalent feeling. The district court was fully aware that these cases presented “a close question”, 410 F.Supp. at 1320, but nevertheless found impermissible, i. e., “palpable”, conflict. Id. The kind of conflict disclosed by the record below is palpable in the sense of being a perceivable source of irritation to both HUD and plaintiffs. We can appreciate the potential for serious differences in the evaluation of rent increase requests and substantial delays in order to obtain two sets of approvals, which could (1) discourage new entrepreneurs (assuming that the § 221(d)(3) program had continued), (2) severely handicap landlords to the point of putting them out of business, (3) jeopardize the federal investment in the projects, and (4) generally complicate unduly HUD’s administration of the § 221(d)(3) program. On the other hand, we cannot find hard evidence that the different approaches to evaluation of rent increases resulted in plaintiffs receiving less than the six per cent return on equity contemplated by HUD, or that they were in default on their mortgage payments because of the Board’s actions. Nor have we been directed to evidence of specific disadvantage resulting from the delay in awaiting Board decisions except the loss of increased rents during the period awaiting the Board's decision. The record on delay is, as we have noted, a mixed one, with neither HUD nor the Board' having won laurel wreaths. The only actual impact of local rent regulation that appears from the district court’s findings is that the landlords suffered “some adverse financial consequences” arising from the lower maximum rents set by the Board and from the delay inherent in its procedures.
III.
Our task of review is further complicated by the state of the law as to the standard to be used in determining federal preemption. We sense significant activity and some movement in a series of recent Supreme Court decisions. As of the time of our remand, it might correctly have been said that federal preemption could, in the absence of strong evidence of either Congressional intent on the one hand or overpowering state interest on the other, be based on the facts that a national scheme existed and that a state scheme somewhat interfered with it. See, e. g., Cloverleaf Butter Co. v. Patterson, 315 U.S. 148, 62 S.Ct. 491, 86 L.Ed. 754 (1942); Rice v. Santa Fe Elevator Corp., 331 U.S. 218, 67 S.Ct. 1146, 91 L.Ed. 1447 (1947). As recently as 1971, Perez v. Campbell, 402 U.S. 637, 91 S.Ct. 1704, 29 L.Ed.2d 233 (1971), Amalgamated Ass’n of Street, Electric & Motor Coach Employees v. Lockridge, 403 U.S. 274, 91 S.Ct. 1909, 29 L.Ed.2d 473 (1971), and 1973, Burbank v. Lockheed Air Terminal, Inc., 411 U.S. 624, 93 S.Ct. 1854, 36 L.Ed.2d 547 (1973), federal preemption could without too much difficulty be derived from the existence vel non of a far reaching federal legislative structure.
Following on the heels of Burbank, however, there has been a series of cases suggesting a somewhat greater reluctance to find federal preemption. In Goldstein v. California, 412 U.S. 546, 93 S.Ct. 2303, 37 L.Ed.2d 163 (1973), the Court held that even the specific constitutional authority for federal copyright laws did not forbid state legislation proscribing the reproduction of phonograph records. Exclusive federal power was said to be limited to “matters which are necessarily national in import.” Id. at 554, 93 S.Ct. at 2308. State or local legislation, to be preempted, must be “absolutely and totally contradictory and repugnant”. Id. at 553, 93 S.Ct. at 2308. See also Kewanee Oil Co. v. Bicron Corp., 416 U.S. 470, 94 S.Ct. 1879, 40 L.Ed.2d 315 (1974). In New York State Department of Social Services v. Dublino, 413 U.S. 405, 93 S.Ct. 2507, 37 L.Ed.2d 688 (1973), the Court rejected the notion that the breadth of federal legislation alone would support an inference of preemptive intent, and relied on the fact that national AFDC legislation contemplated a cooperative relationship [11]*11with state welfare systems to allow the more restrictive New York Work Rules to eliminate more persons from the eligibility list than would the federal WIN program. The Court said, “Where co-ordinate state and federal efforts exist within a complementary administrative framework, and in the pursuit of common purposes, the case for federal pre-emption becomes a less persuasive one.” 413 U.S. at 421, 93 S.Ct. at 2517. Finally, in Merrill Lynch, Pierce, Fenner & Smith, Inc. v. Ware, 414 U.S. 117, 94 S.Ct. 383, 38 L.Ed.2d 348 (1973), the Court held that California laws voiding a forfeiture by a departing employee of benefits in a profit-sharing plan and disregarding arbitration clauses in actions for collection of wages could prevail, notwithstanding a New York Stock Exchange Rule that any such controversy arising from termination of employment is subject to arbitration. In so doing the Court stated that “the proper approach is to reconcile ‘the operation of both statutory schemes with one another rather than holding one completely ousted.’ [quoting Silver v. New York Stock Exchange, 373 U.S. 341, 357, 83 S.Ct. 1246, 10 L.Ed.2d 389 (1963)].” 414 U.S. at 127, 94 S.Ct. at 390.
It is possible to distinguish these cases on their facts. In Goldstein Congress had been silent on the issue of reproducing phonographic records; here, Congress has spoken, if obliquely, on the subject of maximum rents to be charged in subsidized housing.11 In Dublino the interrelation of federal and state welfare schemes was perhaps more strongly articulated than in the case at bar. And in Ware the federal law in question was interpreted as being aimed at the protection of investors, so that only peripherally could there be said to be a conflict between the Stock Exchange arbitration rule and the "California laws. Nevertheless we cannot avoid the impression that there has been a shift in emphasis, raising the threshold of conflict between federal and state or municipal laws which must be reached before federal preemption can be said to occur. See Note, The Pre-emption Doctrine: Shifting Perspectives on Federalism and the Burger Court, 75 Col.L. Rev. 623 (1975). We look upon this case as within the teaching of Dublino. The federal legislation creating the network of subsidized housing laws is superimposed upon and consciously interdependent with the substructure of local law relating to housing.12 And, as in Dublino, the local law here advances significant and uniquely local interests with which the courts should not lightly interfere. Cf. Lindsey v. Normet, 405 U.S. 56, 68, 92 S.Ct. 862, 31 L.Ed.2d 36 (1972). We also think that the doctrinal teaching of Ware and Silver is applicable: that, given a conflict, especially in an area of strong local concern, reconciliation is to be preferred to complete ouster of a state law or municipal ordinance. All of this has bearing, not on the district court’s formulation of the standard of “impermissible conflict”, but on the quantum of conflict — how much is “impermissible”?
Dublino, while not precisely apposite, is instructive. In that case the federal Work Incentive Program (WIN) incorporated in [12]*121967 into the Social Security Act, required states to provide in their AFDC plans certain supportive services such as vocational rehabilitation, counseling, and child care to “employable” individuals so that they might rejoin the labor force. Welfare recipients were required to participate in WIN in order to continue to receive assistance. In 1971 New York enacted its Work Rules requiring “employables”, in addition, to report personally to pick up assistance checks, to file a biweekly certificate that no suitable job opportunity was available, to attend employment interviews, to report results of job referrals, and to report for suitable employment when available. The Court, noting that AFDC was a “scheme of cooperative federalism”, 413 U.S. at 413, 93 S.Ct. 2507; that New York had legitimate interests in promoting self-reliance, in the allocation of limited funds, and in the fiscal hardships affecting local government; and that the state program was a supplementary effort toward the same end of promoting work opportunities for AFDC recipients, held that WIN did not preempt the Work Rules,13 Added factors in the Court’s decision were that the federal agency administering AFDC, the Department of Health, Education, and Welfare “has never considered the WIN legislation to be pre-emptive”, 413 U.S. at 420, 93 S.Ct. at 2516; and that New York had tried to avoid friction and overlap with WIN, subjecting no person to the state program who was participating in WIN.
Although there was a specific statutory nexus in Dublino connecting the federal and state legislation in a “complementary administrative framework”, 413 U.S. at 421, 93 S.Ct. 2507, prior decisions had severely limited the states’ power to define eligibility for federally funded welfare assistance. Id. at 421-22, 93 S.Ct. 2507. Nevertheless, the Court’s decision, starkly put, was to find no conflict where the operation of state law, “in the pursuit of common purposes”, could result in excluding some persons from assistance who would be eligible under federal law. We think the case for preemption is even less persuasive in the case at bar where Boston, in pursuit of its own interests that are basically consistent with federal purposes, operates an independent regulatory scheme that has no demonstrated impact on the federal program, but merely has some financial consequences for some landlords who participate in the program.
Boston’s rent control ordinance “goes right to the heart of the landlord-tenant relationship”, Villalobo v. McCusker, Housing Court of the City of Boston, 02757 C.A. at p. 6. It is aimed at controlling rent levels, assuring consonant rents for similar accommodations, allowing tenants to participate in the rent setting process, and enforcing various local building, fire and health codes. By contrast, the § 221(d)(3) program is not primarily a rent control program, but has as its basic purpose increasing the supply of low income housing. The federal interest in setting § 221(d)(3) rents must be evaluated in connection with the fundamental purposes of the program: the programmatic goals of inducing private entrepreneurs to enter the field of low and middle income housing, so as to make housing available at reasonable rents to low and middle income families; and a direct financial interest in avoiding financial loss on federal guarantees. The latter, to our knowledge, presents a unique dimension in this ease, as compared with other preemption cases. Nevertheless, our inquiry must be directed to evidence in the record that demonstrates a significant — “not merely trivial or insubstantial” — impairment of the federal interests we have identified.14 See Dublino, supra, 413 at 423, n. 29, 93 S.Ct. 2507.
[13]*13The mere potential for conflict which we have perceived in this case is not enough to justify a court in concluding that federal power precludes an otherwise valid exercise of state sovereignty.15 See Goldstein v. California, supra, 412 U.S. at 554-55, 93 S.Ct. 2303. On the contrary, we think that an assertion of preemption directed to such a purely local situation must be accompanied by pleading and proof of actual, significant frustration of federal purpose. See Florida Lime & Avocado Growers, supra, 373 U.S. at 141-42, 83 S.Ct. 1210. These we have not found. The only concrete impact identified by the district court is “some adverse financial consequences” to the local landlords. But the statute makes clear that it is not a purpose of the act to protect mortgagors from regulation under “State laws or by political subdivisions . . . as to rents, charges and methods of operation”, since the law contemplates this as one possible method of regulation. See 12 U.S.C. § 1715l(d)(3). Unlike the welfare recipients in Dublino, the landlords here are not the primary beneficiaries of the federal law under consideration. Given the nature of the municipality’s interests and the absence of concrete evidence that HUD’s purposes in the § 221(d)(3) program were blocked, we believe that the Court would deem applicable, at least in the absence of a clear position by HUD, the principle of preferring cooperation to that of ousting the local regulatory scheme. See Ware, supra, 414 U.S. at 127, 94 S.Ct. 383.
A critical factor in this balancing process is the policy position of the federal administering agency. As we have already observed, at times relevant to this litigation, HUD’s position with some communities seemed to be one of coexistence with local rent control. Letters in evidence indicate that only if “federal interests were sufficiently imperiled” would HUD invoke federal supremacy. If HUD had in fact offered proof that its investment in any of these projects was jeopardized by the landlords’ poor financial condition, whether or not caused by the Board’s actions, this would be a very different case.16 However, the theory of its determination to assert preemption at least as to Castle Square was simply that the Boston Board required a lower rent ceiling than did HUD. Its complaint in intervention in the Druker case rested on the allegation that the “Rent Board’s decision to limit permissible gross rental below that approved by the Secretary creates a direct conflict between State and Federal law”. It seems clear to us that as of the time of intervention in this litigation, HUD had made no system-wide determination of the incompatibility of federal and local rent control and that, for the reasons we have already given, both its theory and the proof of conflict were inappropriate bases for an assertion of preemption relating to a specific project in a single community.
We therefore hold that the district court’s conclusion of impermissible conflict, on the record in this case, was error. The district court did not, however, consider the impact of HUD’s preemption regulation, 24 C.F.R. § 403.9, on Rent Board actions subsequent to its effective date.17 In addition, [14]*14the Contract Clause and Equal Protection Clause claims raised by the Kargman plaintiffs were not reached by the district court. We must remand for consideration of these issues.18
A question remains. Through the years the difference between the higher HUD approved rents and the lower Board approved rents have been paid by the tenants and deposited in escrow. There now exists the possibility of a petition for certiorari based on this decision, the hearing and disposition in the district court on the Contract and Equal Protection Clause claims, possible appeal to us, and possible certiorari thereafter. We have already recognized, in maintaining the escrow pending this appeal, the improbability of plaintiffs’ recovering from tenants if they finally prevailed. On the other hand, tenants have been paying an incremental rent, which we have now held without basis in the Supremacy Clause, for several years. We are also fully conscious that our expectation of prompt record completion in 1972 proved naive and unrealistic. We therefore think it fair to preserve the escrow fund under this limitation.
1. All funds and interest thereon paid into the fund from its inception to the effective date of HUD’s preemption regulation, shall be held subject to the following conditions.
a. If, at any time within three months from the date of this decision, the district court has decided the Equal Protection and Contract Clause claims, this court will entertain a motion for relief or stay pending appeal by a party on either side of the controversy.
b. If, after three months from the date of this decision the district court has not rendered decision for any reason not connected with dilatoriness on the part of defendants and defendant-intervenors, this court will entertain a motion for distribution of the amount in escrow resulting from the payments described in paragraph 1.
2. Payments based on HUD action since the date of the preemption regulation should continue pending the district court’s decision on this aspect of the case.
The judgment of the District Court is vacated and the case is remanded for proceedings consistent with this opinion.