OPINION OF THE COURT
Pigott, J.
In this third appeal by plaintiffs Campaign for Fiscal Equity, Inc. (CFE), et al., we address the cost of providing children in New York City’s public schools with a sound basic education. The State estimated this cost to include a minimum of $1.93 billion, in 2004 dollars, in additional annual operating funds. [20]*20We conclude that this estimate was a reasonable one and that the courts should defer to this estimate, appropriately updated.
I.
More than a decade ago, we held that the Education Article of the New York State Constitution requires the State “to offer all children the opportunity of a sound basic education” (Campaign for Fiscal Equity v State of New York, 86 NY2d 307, 316 [1995] [CFE /]). Plaintiffs had sought a declaratory judgment against the State, claiming that students in New York City public schools were not receiving a basic education and that the State’s public school financing system was unconstitutional.1
Mindful of the fundamental value of education in our democratic society, we agreed with plaintiffs’ interpretation of the Education Article. The State must ensure that New York’s public schools are able to teach “the basic literacy, calculating, and verbal skills necessary to enable children to eventually function productively as civic participants capable of voting and serving on a jury” (86 NY2d at 316). In assessing adequacy of education, this standard is the constitutional minimum or floor that we had acknowledged earlier, in Board of Educ., Levittown Union Free School Dist. v Nyquist (57 NY2d 27, 47-48 [1982]). Accordingly, we held that plaintiffs’ cause of action under the Education Article survived a motion to dismiss (86 NY2d at 318-319), reminding plaintiffs that they would “have to establish a causal link between the present funding system and any proven failure to provide a sound basic education to New York City school children” (86 NY2d at 318).
Plaintiffs succeeded in establishing that causal link, in a 1999-2000 trial concerning the 1997-1998 school year. In 2003, we decided that this trial record supports the conclusion that, because of inadequate funding for their public schools, children in New York City “are not receiving the constitutionally-mandated opportunity for a sound basic education” (Campaign for Fiscal Equity v State of New York, 100 NY2d 893, 919 [2003] [CFE II])- In CFE I, we had understood a sound basic education as teaching skills that enable students to undertake civil re[21]*21sponsibilities meaningfully. In CFE II, we defined “sound basic education” more exactly, as the “opportunity for a meaningful high school education, one which prepares [children] to function productively as civic participants” (100 NY2d at 908 [emphasis added]).
We determined that New York City public schools provided inadequate teaching, because they were unable to attract and retain qualified teachers (100 NY2d at 909-911). They were deficient in at least two instrumentalities of learning: libraries and computers (100 NY2d at 913). Moreover, although plaintiffs had not proven “a measurable correlation between building disrepair and student performance, in general” (100 NY2d at 911), they sufficiently demonstrated “that large class sizes negatively affect student performance in New York City public schools” (100 NY2d at 912).
Whether measured by “inputs” or by “outputs,” i.e. school completion rates and test results (100 NY2d at 914-919), New York City schoolchildren, we determined, were not receiving the opportunity for a sound basic education. Finally, we concluded that plaintiffs had established the causation element of their claim by showing that increased funding can provide better teachers, facilities and instrumentalities of learning, and that such improved inputs in turn yield better student performance (100 NY2d at 919-925).
Accordingly we directed the State to ensure, by means of “[r]eforms to the current system of financing school funding and managing schools . . . that every school in New York City would have the resources necessary for providing the opportunity for a sound basic education” (100 NY2d at 930). Noting that “the political process allocates to City schools a share of state aid that does not bear a perceptible relation to the needs of City students” (id.), we instructed the State to ascertain the actual cost of providing a sound basic education in New York City, rather than the state as a whole (id.). We also held that “the new scheme should ensure a system of accountability to measure whether the reforms actually provide the opportunity for a sound basic education” (id.). We gave the State a deadline of July 30, 2004 by which to implement the necessary measures (id.) and remitted to Supreme Court for further proceedings in accordance with our opinion (100 NY2d at 932).
Within a matter of weeks, Governor Pataki issued an executive order creating the New York State Commission on Educa[22]*22tion Reform, charged with recommending, to the Executive and the Legislature, education financing and other reforms that would ensure that all children in New York State have an opportunity to obtain a sound basic education. The Commission, chaired by Frank G. Zarb, published its final report on March 29, 2004.
The Zarb Commission retained Standard and Poor’s (S & P) School Evaluation Services to calculate the additional spending required to provide a sound basic education, directing S & P to use a “Successful Schools” model that studies the expenditures of school districts with a proven track record of high student performance. The method had been used by the New York State Board of Regents in its Proposal on State Aid to School Districts for 2004-05.
The Zarb Commission developed three alternative criteria for identifying successful school districts. One option was based on New York’s 2007-2008 performance standard set in accordance with the federal No Child Left Behind Act of 2001. Another was similar to the first but substituted the 2006-2007 performance standard. The third criterion was the same approach to identifying successful schools that the Board of Regents had used; it picked out school districts — 281 of 699 — in which at least 80% of the students performed at or above a proficient level, over a three-year period, in seven tests required by the Board of Regents: two fourth-grade examinations and five examinations required for high school graduation.2
Reasoning that not all successful schools operate in a manner that is economical, the Zarb Commission instructed S & P to apply a cost-effectiveness filter: once successful school districts were identified by the methods just described, they were to be ranked according to expenditures and those in the lower-spending half were to be used to create an average. The Board of Regents had noted the necessity for applying such an efficiency filter, because “districts that perform at high levels often enjoy a very substantial wealth base, and therefore also spend at very high per pupil[ ] levels” (Regents Proposal on State Aid to School Districts for 2004-05, at 48, available on the Internet at <http://emsc32.nysed.gov/stateaidworkgroup/2004[23]*2305%20RSAP/RSAP0405.htm>, cached at <http://www. nycourts.gov/reporter/webdocs/Regents_Item.htm>).
Drawing on an extensive review of pertinent research literature, S & P applied three weightings to the resulting base expenditures, in order to take into account the greater spending required for students with special needs. The coefficients by which the base expenditures were multiplied were 2.1 for students with disabilities, 1.35 for economically disadvantaged students, and 1.2 for students with limited English proficiency. S & P cautioned that it was not in a position to recommend explicitly the set of weightings it applied.
Adjustments were also made to account for the local purchasing power of the dollar, using two, alternative cost indices, the New York Regional Cost Index (NY RCI), provided by the New York State Education Department and based on differences in labor market costs, and the Geographic Cost of Education Index (GCEI), provided by the National Center for Education Statistics. The latter index was developed by Jay Chambers, one of the principal authors of the New York Adequacy Study, a cost analysis project cosponsored by plaintiff CEE in 2003-2004. The GCEI attempts to measure the attractiveness of employment within a particular district, one of the key determinants of the cost of providing education. S & P used what was then the most recent, publicly available version of the GCEI, a 1997 update by Chambers. Finally, amounts were adjusted for inflation to reflect January 2004 purchasing power.
S & P thus calculated “sound basic education” spending estimates for each school district, using the two regional cost in-dices and the four alternative criteria for identifying successful school districts. The spending estimates did not include capital, debt or transportation costs. Finally, these figures were compared with amounts actually spent in 2002-2003, in order to identify “spending gaps.”
Applying the GCEI, the estimated spending gaps for New York City ranged from $1.93 billion to $2.53 billion and the statewide spending gaps from $2.45 billion to $3.39 billion, depending on which criterion for successful school districts was used. (New York City’s spending gap thus comprised 74% to 79% of the State’s total gap.) When the NY RCI was applied, the estimated spending gaps were larger, ranging from $4.05 billion to $4.69 billion for New York City, and $4.61 billion to [24]*24$5.57 billion statewide.3 Applying the GCEI and the Board of Regents approach to identifying successful schools, the spending gap for New York City, in 2004 dollars, was $1.93 billion.
Governor Pataki convened the Legislature in extraordinary session on July 20, 2004, and proposed a program bill to the Senate, incorporating the Zarb Commission’s methodology. The Senate passed an amended version of the bill. Ultimately, the legislation was not enacted. In both versions of the bill, the Legislature would have found that the actual costs of providing a sound basic education should be determined using the Board of Regents approach to identifying successful schools (which had picked out 281 of 699 school districts), the S&P weightings for students with special needs, the GCEI, and the cost-effectiveness filter (see 2004 Extraordinary Session NY Senate Bill S 1-A, § 1, at 2 [July 20, 2004]; 2004 NY Senate Bill S 7684-B, § 2, at 2 [July 20, 2004]).
In other words, Governor Pataki and the Senate endorsed the approach that generated a minimum figure of $1.93 billion as the estimated spending gap in operating expenses for New York City. In his State Education Reform Plan, submitted in the course of this litigation, the Governor concluded that “the S&P analysis as adopted by the Zarb Commission and by State defendants determined that $2.5 billion in additional revenues statewide (equating to $1.9 billion in New York City) was a valid determination of the cost of providing a sound basic education in New York City” (State Education Reform Plan, at 14 [Aug. 12, 2004]).
In his program bill memorandum, Governor Pataki made it clear that he intended New York City schools to receive additional funding that exceeded the minimum cost of a sound basic education. Proposals for a Dedicated State Fund for Sound Basic Education and for a New York City local state aid match would, when coupled with projected increases in state school aid and federal aid, generate “approximately $4.7 billion in additional support over the next five years” (Governor’s Program Bill Mem in Support of 2004 Extraordinary Session NY Senate Bill S 1-A, at 4). As noted, the proposed legislation was not enacted. The Legislature, however, passed a bill on August 10, [25]*252004, providing $300 million in additional education aid to New York City.
Once the deadline of July 30, 2004 we had set in CFE II had passed, Supreme Court set out to determine whether the measures we had declared necessary had been carried out. It appointed a blue-ribbon panel of referees “to hear and report with recommendations” on whether the steps taken by the State brought compliance with CFE II.
The Referees conducted numerous hearings, in which they heard from many witnesses, including the Mayor of New York City, the Chancellor of the New York City School District, and representatives of the New York State Division of the Budget and Education Department. They received extensive written submissions, including four compliance plans: the Governor’s State Education Reform Plan, drawing on the Zarb Commission; plaintiffs’ Plan for Compliance, which included the New York Adequacy Study, a cost analysis conducted by the American Institutes for Research (AIR) and Management Analysis and Planning, Inc. (MAP); the Regents Proposal on State Aid to School Districts for 2004-05; and a proposal from the City of New York.
Although they accepted the “successful school districts” methodology of the Zarb Commission, the Referees rejected its cost-effectiveness filter, used a 1.5 weighting for economically disadvantaged students in place of the S & P coefficient of 1.35, and insisted on the use of an updated GCEI, prepared for plaintiff CFE’s New York Adequacy Study. They concluded that the spending gap in New York City was $5.63 billion in 2004-2005 dollars, rejecting the State’s contention that additional funding in the amount of $1.93 billion would ensure the opportunity for a sound basic education in New York City’s public schools.
The Referees adopted CFE’s capital funding program, “Building Requires Immediate Capital for Kids” (BRICKS), recommending that the State be required to ensure that $9,179 billion in 2004-2005 dollars would be available as funding for capital improvements over the following five years. Additionally, the Referees recommended that costing-out studies be carried out every four years, supervised by the Board of Regents, “until it becomes clear that reforms to the State’s education finance formulas have rendered such studies no longer necessary to assure all New York City students the opportunity for a sound ha-[26]*26sic education” (Report and Recommendations of the Judicial Referees, at 39).
On the question of accountability, the Referees concluded that existing state systems, identifying schools that perform poorly and sanctioning failing schools, already provide adequate accountability, and that no new Office of Educational Accountability should be created. They recommended, however, that the current system should be enhanced by the development of a comprehensive “sound basic education” plan by the New York City Department of Education.
Supreme Court confirmed the Judicial Referees’ Report and Recommendations. The Appellate Division vacated that confirmation (29 AD3d 175 [2006]). It found support in the record for the State’s “cost-effectiveness” approach, as well as for its weighting for economically disadvantaged students. Noting that record support, the Appellate Division observed that
“Supreme Court should not have substituted the Referees’ opinion for that of the State . . . [and] converted a factor that was arguable and reasonable for the Legislature and Governor to consider into an incontrovertible fact. As long as the State’s choices remained within the range of professionally accepted practices in determining the costs of a sound basic education, Supreme Court should have left the conclusions for legislative and gubernatorial consideration and determination.” (29 AD3d at 184.)
Citing Governor Pataki’s proposal to increase funding of the New York City School District by $4.7 billion (over a period of five years), the Appellate Division directed the Governor and Legislature to appropriate at least $4.7 billion in additional operating funds (phased in over four years).4 The Appellate Division also directed the Governor and Legislature to “implement a capital improvement plan that expends $9,179 billion [27]*27over the next five years or otherwise satisfies the city schools’ constitutionally recognized capital needs” (29 AD3d at 191).
Plaintiffs CFE et al. appeal pursuant to CPLR 5601 (a) and (b) (1). The state defendants cross-appeal under CPLR 5601 (b) (1).
II.
The Judicial Referees’ Report, dated November 30, 2004, commands our attention as well as our respect; it is likely that much of value may be learned from the Referees’ careful consideration of methods of ascertaining the cost of a sound basic education and reforms to the current system of public school financing. Nevertheless, we hold that Supreme Court erred by, in effect, commissioning a de novo review of the compliance question. The role of the courts is not, as Supreme Court assumed, to determine the best way to calculate the cost of a sound basic education in New York City schools, but to determine whether the State’s proposed calculation of that cost is rational. Supreme Court should not have endorsed an examination in which the cost of a sound basic education in New York was calculated anew, when the state budget plan had already reasonably calculated that cost. In this respect, we agree with the Appellate Division. It was error to confirm the Referees’ Report.
We differ from the Appellate Division, however, in two respects. First, we observe that the state plan found that the cost of providing a sound basic education in New York City was $1.93 billion in additional annual operating funds, and that Governor Pataki’s proposal to provide $4.7 billion in additional funding amounted to a policy choice to exceed the constitutional minimum. Second, in light of recently enacted legislation designed to allow the State to remedy inadequacies in New York City schools facilities, we reject as unnecessary the Appellate Division’s directive regarding capital improvement.
Therefore, we modify the order of the Appellate Division, in two ways. We declare that the constitutionally required funding for the New York City School District includes, as demonstrated by this record, additional operating funds in the amount of $1.93 billion, adjusted with reference to the latest version of the GCEI and inflation since 2004. We vacate the requirement that the Governor and the Legislature implement a capital improvement plan that either expends $9,179 billion over the following five years “or otherwise satisfies the city schools’ constitutionally recognized capital needs.” As modified, we affirm.
[28]*28III.
In CFE II, we expressed the necessity for courts to tread carefully when asked to evaluate state financing plans. On the one hand, the Judiciary has a duty “to defer to the Legislature in matters of policymaking, particularly in a matter so vital as education financing, which has as well a core element of local control. We have neither the authority, nor the ability, nor the will, to micromanage education financing.” (100 NY2d at 925.) On the other hand, “it is the province of the Judicial branch to define, and safeguard, rights provided by the New York State Constitution, and order redress for violation of them” (id.).
The need for deference, where appropriate, is no less important for this Court than it is for the Judiciary as a whole. We are the ultimate arbiters of our State Constitution (see e.g. Cohen v State of New York, 94 NY2d 1, 11 [1999]). Yet, in fashioning specific remedies for constitutional violations, we must avoid intrusion on the primary domain of another branch of government. We have often spoken of this tension between our responsibility to safeguard rights and the necessary deference of the courts to the policies of the Legislature. “While it is within the power of the judiciary to declare the vested rights of a specifically protected class of individuals, in a fashion recognized by statute . . . the manner by which the State addresses complex societal and governmental issues is a subject left to the discretion of the political branches of government” (Matter of New York State Inspection, Sec. & Law Enforcement Empls., Dist. Council 82, AFSCME, AFL-CIO v Cuomo, 64 NY2d 233, 239-240 [1984] [citations omitted]). When we review the acts of the Legislature and the Executive, we do so to protect rights, not to make policy.
Our deference to the Legislature’s education financing plans is justified not only by prudent and practical hesitation in light of the limited access of the Judiciary “to the controlling economic and social facts,” but also by our abiding “respect for the separation of powers upon which our system of government is based” (Matter of 89 Christopher v Joy, 35 NY2d 213, 220 [1974]). We cannot “intrude upon the policy-making and discretionary decisions that are reserved to the legislative and executive branches” (Klostermann v Cuomo, 61 NY2d 525, 541 [1984]).
Deference to the Legislature is especially necessary where it is the State’s budget plan that is being questioned. Devising a [29]*29state budget is a prerogative of the Legislature and Executive; the Judiciary should not usurp this power. The legislative and executive branches of government are in a far better position than the Judiciary to determine funding needs throughout the state and priorities for the allocation of the State’s resources.
We have therefore spoken of the “formidable burden” of proof imposed on “one who attacks the budget plan” (Wein v Carey, 41 NY2d 498, 505 [1977]). Indeed, the burden is
“realistically, impossible as to some categories of estimates. But there are some estimates that could be demonstrated on their face to be unreasonable. An extreme example would be a tripling of the estimates of personal income tax revenue, without a change in the tax rate, in a period in which the economy appears to be on a plateau or in decline.” (Id.)
The illustrations we gave in Wein v Carey, while extreme, were meant to show how patently irrational a state financing plan must be, before judicial deference will give way. Judicial intervention in the state budget “may be invoked only in the narrowest of instances” (id.).
When we remitted in CFE II, we did so in order that Supreme Court would determine, when our deadline had passed, whether the State had implemented the reforms we required — legislation that would ensure that New York City schools have the resources necessary for providing the opportunity for a sound basic education and that would ensure accountability. CFE II called for the State to present evidence of its reforms, both predating CFE II (see 100 NY2d at 927) and following CFE II, and for Supreme Court to determine whether they satisfied our directives.
In light of our language in CFE II and our jurisprudence as a whole concerning deference to the Legislature in matters of policymaking, it was incumbent upon Supreme Court to begin by making a finding as to whether the State’s estimate of the cost of providing a sound basic education in New York City was a reasonable estimate. Then the court should have proceeded to determine whether the state plan, as of July 30, 2004, incorporated that sound basic education expenditure in its proposed budget and would, if enacted, ensure a system of accountability. Supreme Court should not have provided a panel of referees with a mandate to make recommendations as between compli[30]*30anee proposals — the State’s, the plaintiffs’, the City’s, the Regents’. The State, not Supreme Court, was ordered to ascertain the cost of a sound basic education in New York City.
IV
We do not believe that Governor Pataki’s proposed State Education Reform Plan was unreasonable. In particular, we do not find irrational the Governor’s acceptance of the Board of Regents approach to identifying successful schools, the S & P weightings for students with special needs and the cost-effectiveness filter (2004 Extraordinary Session NY Senate Bill S 1-A, § 1, at 2 [July 20, 2004]). As a result, we do not find unreasonable the assertion that “$2.5 billion in additional revenues statewide (equating to $1.9 billion in New York City) was a valid determination of the cost of providing a sound basic education in New York City” (State Education Reform Plan, at 14 [Aug. 12, 2004]). There is substantial record support for that statement.
First, the use of the cost-effectiveness filter is rationally defensible. The variation in spending between New York school districts is very large.5 As S & P explained, averaging the expenditures of all successful schools would
“mask a considerable range of per-pupil spending among the individual districts ... If the concept of ‘adequacy’ means spending no less, but not necessarily more, than is necessary to produce high achievement levels, then there is reasonable cause to adjust the base expenditure by a measure of cost effectiveness. This can be done by ranking the successful districts under each scenario by their base expenditure, and computing the average of the lowest 50% (in terms of spending), which is the same approach used by the New York Board of Regents in its recent study of educational costs. An analysis of the average achievement levels of the lower-spending half of districts shows that they closely resemble the average achievement levels of the upper-spending half of districts . . . .” (Standard & [31]*31Poor’s Resource Adequacy Study for the New York State Commission on Education Reform, at 46-47 [Mar. 2004].)
The essential premise of the cost-effectiveness filter is that the higher-spending half of the successful districts is spending more than the constitutional minimum — either because those districts spend less efficiently than some others or because they have chosen to do more for their students than the Constitution requires. The State, in adopting S & P’s approach, implicitly concluded that New York City could attain minimal constitutional standards while spending less than this higher-spending group of successful districts. The premise, and the conclusion, are no doubt debatable, but we cannot say they are irrational, and they are therefore entitled to deference from the courts.
The S & P weightings for children with special needs also have record support. While S & P did not recommend any particular weighting over another, the coefficients that S & P applied were drawn from an extensive review of relevant research. Indeed the pertinent footnote to S & P’s Resource Adequacy Study cites no fewer than 37 articles, reports and other scholarly works (Standard & Poor’s Resource Adequacy Study for the New York State Commission on Education Reform, n 16, at 89-92 [Mar. 2004]).
The S & P calculations — applying a 2.1 weighting for students with disabilities, 1.35 for economically disadvantaged students, and 1.2 for students with limited English proficiency, and reaching the conclusion that the spending gap for the New York City School District is $1.93 billion — were reasonable. Although we recognize that legitimate arguments can be made for raising the coefficient for economically disadvantaged students to 1.5, we do not believe that the figure of 1.35 lacks grounding in prudent reason.
Accordingly, we declare that the constitutionally required funding for the New York City School District includes additional operating funds in the amount of $1.93 billion, adjusted with reference to the latest version of the GCEI and inflation since 2004.
V
Turning to capital improvements, we emphasize again, as we did in CFE I, that New York’s public schoolchildren “are entitled to minimally adequate physical facilities and classrooms which provide enough light, space, heat, and air to permit chil[32]*32dren to learn” (86 NY2d at 317). The Appellate Division directed the Governor and the Legislature to implement a capital improvement plan that either expends $9.179 billion over the following five years “or otherwise satisfies the city schools’ constitutionally recognized capital needs” (29 AD3d at 191). In choosing its words thus, the Appellate Division was perhaps mindful of the fact that in CFE II we did not expressly require the State to calculate the amount of capital funding necessary to remedy deficiencies in facilities, with the result that the State did not carry out a costing-out study for capital funding needs in the way it did for operating costs.
The part of the Appellate Division order that requires a capital improvement plan should be vacated as unnecessary. In 2006, the Legislature set forth a capital construction program totaling $2.6 billion, that includes $1.8 billion for the New York City School District (L 2006, ch 58, part A-2; L 2006, ch 61, part I). Crucially, the Legislature increased the cap for the New York City Transitional Finance Authority by $9.4 billion to help fund the cost of capital improvements, and permitted New York City to pledge future state building aid in order to repay borrowed funds (L 2006, ch 58, part A-3). We are of course bound to decide this case on the record before us. But since the parties now agree that the funds envisaged by the Legislature this year would be sufficient to remedy facilities deficiencies, we believe that there is no need for further judicial direction.
Finally, insofar as the Appellate Division vacated Supreme Court’s order confirming the Referees’ Report, it struck Supreme Court’s call for state costing-out studies every four years and its requirement that the New York City Department of Education prepare a comprehensive “sound basic education” plan, to ensure accountability. We agree with these results. In particular, we agree with the City of New York, an amicus in this case, that a new and costly layer of city bureaucracy is not constitutionally required. It is undisputed that there are minimally adequate accountability mechanisms now in place for the evaluation of New York schools (including the Schools Under Registration Review process and the state standards required by the federal No Child Left Behind Act of 2001).
Accordingly, the order of the Appellate Division should be modified, without costs, by granting judgment declaring in accordance with this opinion and, as so modified, affirmed.