dissenting.
The federal government makes payments for the benefit of local education agencies (LEAs) to ameliorate costs incurred in educating federally connected children. Such children include those residing on Indian lands. 20 U.S.C. § 7703(a)(1)(C). Quite reasonably, Congress wants the impact aid payments to be applied for the intended (and expressed) purpose, not merely used as a federal supplement for a state’s general education *1170needs. But the matter is complicated because some states have equalized education spending, taking money from wealthy LEAs to meet the needs of children in less fortunate ones. In truly equalized states the special costs of educating federally connected children are borne system-wide, rather than by the LEAs where those children reside. Congress recognized the complication and dealt with it quite specifically in the statutes. That is the backdrop for and focus of this litigation.
Plaintiff school districts contend that New Mexico has not equalized education spending and therefore cannot count impact aid for LEAs as a local resource.1 They are correct. Because the majority’s decision defies express statutory language, I respectfully dissent.
The relevant statute, 20 U.S.C. § 7709, prohibits a state from counting federal Impact Aid payments as a resource in determining the amount of state aid to be allocated to that LEA for free public education.2 But there is an exception for a state with “a program ... that equalizes expenditures for free public education among local educational agencies in the State.” 20 U.S.C. § 7709(b)(1). A state’s educational spending is equalized—
if, in the second fiscal year preceding the fiscal year for which the determination is made, the amount of per-pupil expenditures made by, or per-pupil revenues available to, the local educational agency in the State with the highest such per-pupil expenditures or revenues did not exceed the amount of such per-pupil expenditures made by, or per-pupil revenues available to, the local educational agency in the State with the lowest such expenditures or revenues by more than 25 percent.
20 U.S.C. § 7709(b)(2)(A) (emphasis added). In making that determination Congress directed the Secretary to “disregard local educational agencies [LEAs] with per-pupil expenditures or revenues above the 95th percentile or below the 5th percentile of such expenditures or revenues in the State” as part of its equalization determination. 20 U.S.C. § 7709(b)(2)(B)(I) (emphasis added); see also 20 U.S.C. §§ 7702, 7703 and 7709(b)(1).
These requirements are unambiguous. A percentile is a mathematical concept not admitting of multiple interpretations; it is a simple, straightforward method of ranking an array of values. Attached to this dissent is Exhibit A. It lists all of the 89 New Mexico LEAs along with the per-pupil revenue for each.3 Analysis of that array yields a value of $3,650.40 for the *117195th percentile and $2,803.80 for the 5th percentile. Five districts are above the 95th percentile and five districts fall below the 5th percentile; they are excluded from further analysis. After the exclusion, Gadsden district has the lowest per-pupil revenues ($2,829.00). When those revenues are multiplied by 125% the result is $3,536.00, an amount less than the highest non-excluded district, Maxwell — with per-pupil revenues of $3,591.00. The 25% test is not met. New Mexico has not equalized school funding as statutorily required, therefore federal impact aid cannot be counted as part of the LEAs’ resources.
Eschewing a plain and simple reading of the statute, the Department of Education has adopted regulations which it and, derivatively, the State of New Mexico used to determine whether New Mexico has an equalized educational funding system. The regulations essentially contradict and undermine the provisions of the statute that authorizes them, 20 U.S.C. § 7709.4
Rather than abide the statutory command to apply a mathematical function to an array of numbers, the Department adopted regulations directing a complex and mystifying formula for determining which LEA’s fall into the 5th and 95th percentiles of per-pupil expenditures. 34 C.F.R. 222.162, et. seq. The regulations start correctly by requiring LEAs to be arrayed according to their per-pupil expenditures (or revenues). Next, and inexplicably, they require a calculation of 5 percent of the total number of pupils in the state. The number of LEAs necessary to use up 5% of the state’s student population at both ends of the list are eliminated from consideration in applying the 25% formula. Then the per-pupil expenditures of the highest and lowest remaining LEAs are used to determine whether there is less than a 25% disparity in funding and, therefore, “equalization.” That seems quite at odds with the statute’s directive to disregard LEAs “with per-pupil expenditures or revenues above the 95th percentile or below the 5th percentile of such expenditures or revenues in the State.” I cannot join the majority in putting the Court’s imprimatur on an unwieldy scheme utterly at odds with clear statutory language.
Review of this regulatory framework begins and ends with Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). In reviewing an agency’s construction of a statute it administers, we must first ask whether Congress has directly spoken on the precise question at issue. If the statute is silent or ambiguous, the inquiry is whether the agency’s interpretation is a permissible construction of the statute. Id. at 842-43, 104 S.Ct. 2778. However, when the statute is clear and unambiguous, further “interpretation” is not only unnecessary, it is prohibited. Id.; see also Regions Hosp. v. Shalala, 522 U.S. 448, 457, 118 S.Ct. 909, 139 L.Ed.2d 895 (1998). Unlike the majority, I am not convinced we need to look beyond the statute to find the precise formula for measuring equalization.
Statutes should be read to avoid rather than beget ambiguity. Here, § 7709 instructs the Secretary to “disregard LEAs with per-pupil expenditures or revenues above the 95th percentile or below the 5th *1172percentile of such expenditures or revenues in the State.” The focus is upon LEAs, not upon the number of pupils. Moreover, Congress was concerned with ranking LEAs by expenditure (or resource) “percentiles,” rather than by the number or percentage of students. The words Congress chose are not coincidental, nor are we allowed to indulge in presumptions, despite our own personal predilections. Conn. Nat’l Bank v. Germain, 503 U.S. 249, 253-54, 112 S.Ct. 1146, 117 L.Ed.2d 391 (1992) (“[C]ourts must presume that a legislature says in a statute what it means and means in a statute what it says there.”) (citations omitted).5
The Department’s approach not only undermines the plain language of the statute, but also its purpose. “There is ... no more persuasive evidence of the purpose of a statute than the words by which the legislature undertook to give expression to its wishes.” United States v. Am. Trucking Ass’ns, 310 U.S. 534, 543, 60 S.Ct. 1059, 84 L.Ed. 1345 (1940). The stated purpose in providing Impact Aid is to “fulfill the Federal responsibility to assist with the provision of educational services to federally connected children in a manner that promotes control by local educational agencies with little or no Federal or State involvement.” 20 U.S.C. § 7701 (emphasis added). Focusing on the needs of individual LEAs rather than on the entire state school population also flows into the understanding of § 7709, for its focus is on examining “the amount of per-pupil expenditures made by, or per-pupil revenues available to, the local educational agency in the State....” 20 U.S.C. § 7709(b)(2)(A) (emphasis added).
In support of its conclusion that the statute is ambiguous, the majority emphasizes 1994 events saying that Congress codified the disparity standard in 20 U.S.C. § 7709 “essentially as it appeared in the former regulations,” which allowed for the percentile calculation to be based on the total student enrollment in the state. Majority Op. at 15. The statement might be persuasive if it were not manifestly untrue. The entire basis for this case is the difference in language between the 1994 statute, which by its terms does not allow for a population based calculation, and the pre-existing regulations that expressly did. If Congress had adopted a standard that was “essentially” the same as the pre-existing regulations, it would not have altered the language of the statute to require that the calculation be based on per-pupil expenditures or revenues of “local educational agencies.” While the majority may consider such a difference an “unexplained and slight alteration,” such a conclusion ignores the strong presumption that Congress chooses its words carefully and with full knowledge of administrative interpretations. “Congress is presumed to be aware of an administrative ... interpretation of a statute and to adopt that interpretation when it re-enacts a statute without change.” Lorillard v. Pons, 434 U.S. 575, 98 S.Ct. 866, 55 L.Ed.2d 40 (1978) (emphasis added).6 The fact that Congress adopted a disparity standard different from the clearly articulated preexisting regulations is evidence that Con*1173gress rejected, rather than adopted, the total student enrollment standard.
The majority observes, “There may exist some other, and perhaps even better, way to eliminate anomalies in school district funding in New Mexico.” Maj. Op. at 1168. Nevertheless it concludes the Department’s construction is a permissible one. I posit that the Department’s method of handling the 5th/95th percentile exclusion may be the superior one, but it is not the one ordained by Congress.
EXHIBIT A
[[Image here]]
*1174[[Image here]]
. When impact aid is counted as part of an LEA’s local resources it reduces the amount of state aid to the LEA.
. 20 U.S.C. § 7709(a) General prohibition Except as provided in subsection (b) of this section, a State may not—
(1) consider payments under this sub-chapter in determining for any fiscal year—
(A) the eligibility of a local educational agency for State aid for free public education; or
(B) the amount of such aid; or
(2) make such aid available to local educational agencies in a manner that results in less State aid to any local educational agency that is eligible for such payment than such agency would receive if such agency were not so eligible.
.It uses the 1999-2000 data supplied by the parties and was produced using Microsoft's ‘'Excel” spreadsheet program. Once the districts and per-pupil spending were listed the program's percentile function ^PERCENTILE ([first cell-Moscero district]:[last cell-Des Moines district],0.95)) was applied to the array and returned the numbers for the 95th percentile and 5th percentile. No manipulation or massaging of the data was necessary.
. The Department ignored 1994 statutory changes, clinging instead to contrary past practices. See Majority Op. at 1164-66. It now argues, and the majority accepts, that its historical practice of using "pupils in attendance in the schools of those agencies” is congruent with the statutory language "per-pupil expenditures or revenues.” As Justice Scalia said in a different context, "they are as similar as chalk and cheese.” Blakely v. Washington, - U.S. -,-, 124 S.Ct. 2531, 159 L.Ed.2d 403 (2004).
. The subsection's title "Computation,” 20 U.S.C. § 7709(b)(2), suggests that Congress had a particular formula in mind, the one it adopted.
. "This selectivity that Congress exhibited in incorporating provisions and in modifying certain FLSA practices strongly suggests that but for those changes Congress expressly made, it intended to incorporate fully the remedies and procedures of the FLSA”. Id. at 582, 98 S.Ct. 866 (emphasis added).