American Association of Cosmetology Schools v. Devos

UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA

AMERICAN ASSOCIATION OF :
COSMETOLOGY SCHOOLS, :
:
Plaintiff, : Civil Action No.: 17-0263 (RC)
:
v. : Re Document Nos.: 8, 18
:
ELISABETH DEVOS, in her official capacity :
as Secretary of Education, :
:
Defendant. :

MEMORANDUM OPINION

GRANTING PLAINTIFF’S MOTION FOR SUMMARY JUDGMENT;
DENYING DEFENDANT’S MOTION FOR SUMMARY JUDGMENT;
DENYING PLAINTIFF’S MOTION FOR A PRELIMINARY INJUNCTION

I. INTRODUCTION

In this case, the Court considers whether the Department of Education (“DOE”) acted

arbitrarily and capriciously with respect to cosmetology schools when it decided to

presumptively use earnings data that does not account for unreported income. Although the

DOE was justified in using reported income as the presumptive measure of overall income, it

arbitrarily and capriciously made rebutting that presumption overly difficult.

In setting standards that determine which proprietary schools’ graduates are entitled to

federally backed student loans, the DOE looks to the rates at which the schools’ graduates are

“gainfully employed.” To determine whether graduates are gainfully employed, the DOE has

adopted a test that compares the graduates’ income levels to their levels of debt. To determine

the graduates’ income, the DOE presumptively uses the Social Security Administration’s

(“SSA”) income data. This data does not account for income that is not reported to the Internal
Revenue Service. Schools may appeal the DOE’s use of SSA data through “alternate earnings

appeals,” which, if successful, allow them to use alternate measures of income before the debt-

to-earnings rates become final. To submit such an appeal, a school is required to use either state-

sponsored data pertaining to over half of its graduates during the relevant timeframe or gather

income data on almost all of its graduates through a survey. Schools that fail the debt-to-

earnings test for a long enough time lose eligibility for federal loans. Schools at immediate risk

of losing federal-loan eligibility are required to warn their students and prospective students that

they may be ineligible for student loans in the near future.

During the notice-and-comment period, several commenters argued that use of SSA data

would be unfair to, among others, cosmetology programs, because their graduates

disproportionately underreport their income due to high levels of cash-based and self-

employment-based earnings, including tips. The commenters suggested that the Bureau of Labor

Statistics’s survey-based data would better account for unreported income.

The DOE rejected the commenters’ objection to the data and their proposed solution. In

rejecting the objection, the DOE reasoned that graduates who do not report their income are

subject to civil and criminal penalties, and noted that the SSA data is only one means of

determining income; programs can submit alternate earnings appeals to show the DOE that its

graduates’ average incomes are actually higher than the SSA data indicates. In rejecting the

proposed solution, the DOE noted that the Bureau of Labor Statistics data cannot be tied to

particular programs, and thus would undermine the purpose of the regulations—to identify

individual programs whose graduates are not gainfully employed.

The American Association of Cosmetology Schools (“AACS”) sued under the

Administrative Procedure Act, arguing that these responses were unsatisfactory, and thus the

2
DOE arbitrarily and capriciously failed to adequately consider the unreported-income issue when

it promulgated the regulations. AACS is a nonprofit association of cosmetology schools, many

of which are at risk of failing the DOE’s debt-to-earnings test. In fact, at least three schools have

already posted warnings to their students, because they could not feasibly appeal their failing

grades. Defendant moves for summary judgment on two procedural grounds and on the basis

that the regulations were not arbitrary or capricious.

Both of the DOE’s procedural arguments concern subject-matter jurisdiction. It first

argues that AACS does not challenge final agency action, because none of AACS’s member-

schools have unsuccessfully appealed their failing debt-to-earnings rates. The DOE also argues

that AACS does not have standing to bring this action, because it did not identify any individual

member-schools with standing in the complaint. The Court holds that AACS is challenging final

agency action, because at least three schools have already had to post warnings, and raising an

administrative appeal is not a prerequisite to suit under the APA. The Court further holds that

AACS has standing to bring this case, because AACS indeed identified adversely affected

members and meets all other requirements to establish associational standing to sue on behalf of

its members.

The DOE argues that it did not act arbitrarily and capriciously because it provided a

reasoned explanation for its rejection of commenters’ concerns about unreported income and

their proposed solution. The Court holds that, although the DOE was rational in rejecting the

commenters’ proposed alternative, it did not adequately address the issue of unreported income.

Neither side disputes that SSA data is highly accurate, even though it may be insufficient in

certain circumstances. But, as the commenters pointed out—and the government did not

dispute—there is a significant problem with this data. Despite the illegality of failing to report

3
earnings to the IRS, many programs’ graduates fail to report substantial portions of their income.

The DOE proffered two responses to the commenters’ problem. First, it argued that civil and

criminal penalties deter underreporting. Second, it noted that institutions may submit an

alternate earnings appeal to have their graduates’ earnings more accurately calculated.

The existence of penalties—which existed when the commenters’ data was collected and

will continue to exist into the future, absent some added deterrent—is irrelevant to the issue of

undercounting income. In comparison, an alternative means of measuring earnings data is

responsive to the problem. The DOE justifiably used SSA data as the default measure of

earnings, but supplemented that default methodology with an alternate means of determining

income. However, by inexplicably requiring high response rates to submit state-sponsored or

survey-based alternate earnings calculations, the DOE narrowly circumscribed the alternate-

earnings appeal process, making it unfeasible for certain programs to appeal their designations.

To remedy the DOE’s arbitrary and capricious narrowing of appellate recourse—and to avoid

upending the entire administrative scheme—the Court removes barriers to appeal, making it

more widely available for programs subject to the regulations.

II. REGULATORY BACKGROUND

Congress passed Title IV of the Higher Education Act of 1965 (“HEA”) to make

postsecondary education more widely available to the general public. See 20 U.S.C. § 1070(a).

To broaden the availability of student loans and lower their cost to students, Title IV authorizes

the United States government to enter into agreements with postsecondary educational

institutions to allow students at their schools to obtain federally guaranteed loans. See id.

§ 1094. To protect the taxpayer, the government sets certain conditions that participating

institutions must meet. See id.; see also Ass’n of Private Sector Colls. & Univs. v. Duncan

4
(APSCU I), 681 F.3d 427, 435 (D.C. Cir. 2012) (“[S]chools receive the benefit of accepting

tuition payments from students receiving federal financial aid, regardless of whether those

students are ultimately able to repay their loans. Therefore, Congress codified statutory

requirements in the HEA to ensure against abuse by schools.”). Many of those conditions center

on the financial success of proprietary institutions’ graduates, because such success makes it

more likely that the borrower will repay his own loans. See generally 20 U.S.C. § 1094. Under

20 U.S.C. § 1002(b)(1)(A), students at “proprietary institutions of higher education” may receive

federal loan assistance only if their school “provides an eligible program of training to prepare

students for gainful employment in a recognized occupation.” See also Compl. ¶ 8, ECF No. 1

(emphasis added).

From 2009 to 2011, the DOE promulgated regulations defining the term “gainful

employment” (“GE”) in terms of a debt-to-income test and a debt-repayment test. See Program

Integrity: Gainful Employment—Debt Measures, 76 Fed. Reg. 34,386 (June 13, 2011); see also

Ass’n of Private Colls. & Univs. v. Duncan, 870 F. Supp. 2d 133, 153–54 (D.D.C. 2012). This

Court determined that parts of that rule were arbitrary and capricious under the Administrative

Procedure Act (“APA”). See Ass’n of Private Colleges & Universities, 870 F. Supp. 2d at 154,

158.

The DOE again set out to define the term “gainful employment,” this time focusing only

on program graduates’ overall debt compared to their earnings, which it refers to as “D/E” rates.

See Program Integrity: Gainful Employment, 79 Fed. Reg. 64,890, 64,891 (Oct. 31, 2014) (to be

codified at 34 C.F.R. pt. 600, 668). “The D/E rates measure evaluates the amount of debt

(tuition and fees and books, equipment, and supplies) students who completed a GE program

5
incurred to attend that program in comparison to those same students’ discretionary and annual

earnings after completing the program.” Id.; see also 34 C.F.R. § 668.404.

The GE regulations use two tests to determine whether programs “pass,” each of which is

designed to determine whether programs’ graduates are earning enough income to justify the

government’s guarantee of their loans. The first test divides the program’s graduates’ average

loan payment amounts by their average discretionary income amounts.1 The resulting rate is

called the program’s “discretionary income rate.”2 34 C.F.R. §§ 668.402, 668.404. A program

passes if its graduates’ loan payments make up, on average, 20 percent or less of their overall

discretionary income. See 34 C.F.R. §§ 668.403. So, for example, if a program’s graduates were

spending an average of $200 per month on student loan payments, its graduates’ average

monthly discretionary income would have to be $1000 or more for the program to pass.

The second test divides the program’s graduates’ average annual loan payments by their

average annual income. The resulting rate is called the program’s “annual earnings rate.”

§§ 668.402, 668.404. A program passes if its graduates’ average annual loan payment is 8

percent or less of its graduates’ average annual earnings. See 34 C.F.R. § 668.403–04. Using

another example, if a program’s graduates paid, on average, $800 per year in student loans, they

would need to make an average of $10,000 per year to pass.

A program is classified as “failing” the GE regulations if its discretionary income rate is

greater than 30 percent and its annual earnings rate is more than 12 percent. Id. at

§ 668.403(c)(2). If a program does not pass either test and its discretionary income rate or its

1
Discretionary income is determined via a formula, not by measuring each graduate’s
actual level of discretionary income. 34 C.F.R. § 668.404(a)(1).
2
Under the regulations, average income is defined as mean or median income—
whichever is higher. 34 C.F.R. § 668.404(a).

6
annual earnings rate falls in between the passing and failing rates, then the school is designated

as “in the zone” between passing and failing. See id. § 668.403(c)(3); see also 79 Fed. Reg. at

64,891. A program loses HEA loan eligibility if (1) it fails for two out of three consecutive

years, or (2) it is either in the zone or failing for four consecutive years. 34 C.F.R.

§ 668.403(c)(4)(i)–(ii).

To measure program graduates’ income rates in calculating a program’s discretionary

income rate or annual earnings rate, the DOE presumptively uses income data from the Social

Security Administration. 34 C.F.R. §§ 668.405, 668.404(c)(1). The SSA data accounts only for

earnings that were reported to the IRS for the purpose of calculating income tax. See 79 Fed.

Reg. 64,890.

If the DOE’s presumptive D/E rates, as calculated using SSA data, would leave a

program either failing or in the zone, it may file an “alternate earnings appeal” to request an

alternative formulation to determine the program’s final D/E rates. 34 C.F.R. §§ 668.406(a),

668.409(a)(5). This allows the school to argue that more accurate data exists from which its

graduates’ average level of earnings can be calculated. See id. §§ 668.406(b), 668.409(a). The

process requires the school to use either an institutional survey of the institution’s graduates or a

state-sponsored data system, if such a system exists. Id. § 668.406(b)(1). The data generally

must be from the same calendar year as the SSA data that the DOE used to calculate the

program’s D/E rate. Id. § 668.406(a). The state-sponsored data-system avenue requires that the

data include more than 50 percent of a program’s graduates during the relevant period. See id.

§ 668.406(d). With limited exceptions, the institutional-survey avenue requires a school to

7
obtain the earnings data of all of a program’s graduates during the relevant period.3 Id.

§ 668.406(c). The DOE never explained how it came up with the survey-response or state-based

data requirements. See 79 Fed. Reg. at 64,995–96; see generally 79 Fed. Reg. 64,890. While an

appeal is pending, the school is not considered at risk of losing eligibility the following year. 34

C.F.R. § 668.406(e)(2).

Notably, if a program successfully appeals the DOE’s use of SSA data for the

determination of its D/E rates, it can avoid any negative repercussions altogether. Under 34

C.F.R. § 668.409(a)(1), the DOE’s “final D/E rates” can include either the presumptive SSA data

or a program’s successfully proposed alternative. Only after finalizing each program’s D/E rate

does the DOE finally determine whether the program is passing, failing, or in the zone. Id.

§ 668.409(a)(2).

However, if a school does not timely appeal its designation or is unsuccessful in seeking

an alternative earnings calculation, it begins to suffer serious consequences. Under the Gainful

3
Although Plaintiff seems to suggest that this avenue required the responses of only 50
percent of its graduates, see Decl. of Mez Varol, President of International Academy (“Varol
Decl.”) ¶¶ 7–8, ECF No. 8-3; Tr. Oral Arg. (“Tr.”) at 17, ECF No. 25, this is not supported by
the text of the regulations. Under 34 C.F.R. § 668.406(c) and (b)(3), to submit an alternate
earnings appeal based on an institutional survey, the “institution must . . . [i]n accordance with
the standards included on an Earnings Survey Form developed by [the National Center for
Educational Studies (“NCES”)], conduct a survey to obtain annual earnings information of” “all
the students who completed the program during the same cohort period that the [DOE] used to
calculate the final D/E rates,” with narrow, enumerated exceptions. See also 79 Fed. Reg. at
64995 (“An institution that wishes to submit an appeal by providing survey data must include in
its survey all the students who completed the program during the same cohort period that the
Secretary used to calculate the final D/E rates under § 668.404 or a comparable cohort period,
provided that the institution may elect to exclude from the survey population all or some of the
students excluded from the D/E rates calculation under § 668.404(e).”). The regulations do not
outline what the standards of an Earnings Survey Form might be. J.A. 606, ECF No. 26-1.
Regardless of the NCES’s standards, however, the regulations require the institution to “survey .
. . all of the students who completed the program during the same cohort period that the
Secretary used to calculate the final D/E rates.” 34 C.F.R. § 668.406(b)(3).

8
Employment rules, schools that could lose HEA loan eligibility the following year must provide

written warnings to current and prospective students informing them that their “program has not

passed standards established by the U.S. Department of Education,” and that if the institution

does not obtain a passing grade, the students may not be able to receive federal student grants or

loans the following year. 34 C.F.R. § 668.410(a)(2)(i). The warning requirement was

implemented to protect prospective students from programs that might put their loan eligibility

into jeopardy. See 79 Fed. Reg. at 64,894. A program becomes ineligible for federal student

loans under the HEA after failing the D/E rates for two out of three consecutive years or by

having a combination of failing and in-the-zone rates for four consecutive years. See 34 C.F.R.

§ 668.403(c)(4)(i)–(ii).

Federal courts have jurisdiction to review final agency action. 5 U.S.C. § 704. The APA

requires reviewing courts to set aside an agency action that is “arbitrary, capricious, an abuse of

discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A). In cases like this

one, prior to final publication, the agency is required to submit proposed regulations for a “notice

and comment” period, during which interested persons may submit data and views to the agency

for its consideration. See 5 U.S.C. § 553.

III. FACTUAL BACKGROUND

The facts in this case are largely undisputed. Plaintiff, the American Association of

Cosmetology Schools brings this action under the APA, seeking to enjoin the DOE from

enforcing its GE rules. Compl. at 1. AACS is a nonprofit association of accredited cosmetology

programs. Compl. ¶ 1. At core, AACS takes issue with the DOE’s use of the SSA annual

earnings data. See Compl. ¶¶ 15, 35–38. Plaintiff notes that the SSA data is incomplete because

it tends to undercount the income of self-employed individuals and individuals who receive a

9
significant amount of their income in the form of tips and cash. J.A. 797, ECF No. 26-1; see also

Decl. of Anthony Civitano (“Civitano Decl.”) ¶¶ 25, 28, ECF No.8-1. These two groups, the

argument goes, can underreport their income more easily than persons earning paychecks from

employer payrolls. J.A. 606; Civitano Decl. ¶ 28. Thus, cosmetology school graduates—who

Plaintiff claims are disproportionately self-employed and paid in cash—on average have higher

incomes than the SSA’s numbers reflect, in many cases as much as double. J.A. 595, 787;

Civitano Decl. ¶ 28.

According to Plaintiff, the DOE was made aware of “this well-known reported income

gap phenomenon” by several commenters during the notice-and-comment period of the GE rule

promulgation. Compl. ¶¶ 47–48. Indeed, the DOE explicitly acknowledges that these comments

included claims that “about half of earnings in service occupations such as cosmetology” are

made up of tips, and that many people in the industry are self-employed. See 79 Fed. Reg. at

64,955. A report by Stanford professor Dr. Eric Bettinger, which was submitted to the agency

during the notice-and-comment period, found that both tip income and self-employment income

are, on average, underreported by around 60%. See J.A. 594; see also Decl. of Katherine Brodie

(“Brodie Decl.”) at 493, ECF No. 8-4.

Even aside from formal public comments, AACS cites to several sources suggesting that

the underreporting of cash-based earnings—including in the cosmetology industry—is widely

known to be a serious problem. See Brodie Decl. ¶ 3. For example, AACS provides the House

Budget Committee testimony of Russell George, Treasury Inspector General of the Tax

Administration, wherein Mr. George states that “self-employed individuals who formally operate

. . . businesses . . . are estimated to report only about 68 percent of their income for tax purposes”

and “self-employed individuals operating businesses on a cash basis report just 19 percent of

10
their income to the IRS.” Brodie Decl. at 1258. Although these materials were apparently not

submitted in connection with notice-and-comment and have little direct relevance, they lend

credence to the general idea that the income gap phenomenon is a significant problem.

The DOE acknowledged that the use of SSA data was imperfect, but generally viewed

civil and criminal penalties as sufficient deterrents to underreporting. See 79 Fed. Reg. at

64,955–56. The pertinent portions of the DOE’s responses to the public comments are as

follows:

We do not agree that our reliance on reported earnings is flawed because of its
treatment of self-employment earnings and tips . . . . We acknowledge that some
self-employed individuals may fail to report, or underreport, their earnings.
However, [the Internal Revenue Code] requires self-employed individuals to file a
return if the individual earns $400 or more for the taxable year. . . .
Underreporting subjects the individual to penalty or criminal prosecution. . . .

With respect to the earnings of workers who regularly receive tips for their
services, [the Internal Revenue Code] requires individuals to report to IRS their
tip earnings for any month in which those tips exceeded $20, and . . . individuals
who fail to do so are subject to penalties. . . .

For these reasons, we do not agree with the commenters’ assertion that aggregate
earnings data provided by SSA . . . are unreliable with respect to workers in
occupations that involve significant tip income or a high percentage of income
from self-employment. . . . Moreover, the regulations allow an institution to
submit an alternate earnings appeal using State databases or a survey.

79 Fed. Reg. at 64,955–56.

At least one commenter also proposed an alternate means of calculating income. See id.;

J.A. 53, 785; see also Brodie Decl. at 1193–94. That commenter suggested using data gathered

by the Bureau of Labor Statistics, which relies on population surveys rather than formally

reported income. See J.A. 597, 785; Brodie Decl. at 1193–94. According to Dr. Bettinger, this

data would more accurately capture the earnings of, among others, cosmetology-school

graduates, because, unlike SSA data, it takes unreported income into account. See Brodie Decl.

at 684–85, 1193–94. “[T]hese adjustments would significantly augment the SSA aggregate

11
earnings reported for these occupations, increasing the median earnings by 19 percent and the

mean earnings by 24 percent.” 79 Fed. Reg. at 64,890, 64,955. In responding to this comment,

the DOE not only discounted the commenter, but concluded that no commenter had provided a

viable alternative to the SSA data. The DOE specifically stated that

[i]mputing some percentage of added earnings to account for underreported tips
and other compensation could only be done by generalizations drawn from some
source of data on earnings, but none has been suggested that would permit doing
so in a way that would distinguish between programs.

To assess the bias that the commenter asserted . . . the commenter relies on
earnings data from . . . surveys of households. The survey samples data on a
selection of all households, and relies on earnings data as provided by the
individuals included in the survey. As the commenter noted, there are no data . . .
that allow one to associate a particular respondent with a particular GE program.

Unlike the approach taken in these regulations, which captures all earnings of the
cohort of students completing a program and credits those earnings to the program
completed by the wage earners, the analysis proposed by the commenter does the
reverse: It extrapolates from earnings reported by those survey recipients who
identify their occupation as one that appears related to GE programs of that
general type, and then projects an increase in aggregate earnings for all GE
programs in the category of programs that appears to include that occupation. In
fact, even if the respondents were all currently employed in occupations for which
a category of GE programs trains students, the respondents’ earnings will almost
certainly have no connection with a particular GE programs [sic] we are
assessing. Because any inference drawn from . . . respondents’ earnings could
only benefit a whole category of programs—improving the D/E rates for every
program in that category—using such inferences would mask poorer performing
programs and thwart a major purpose of the GE assessment.

In addition, by the time the survey is conducted, the respondent can be expected
to identify his or her current or most recent job, which may be different than the
occupation for which training was received years before in a GE program. Thus,
to draw a usable inference about D/E earnings from data gathered in the CPS one
must connect a particular GE program now being offered and evaluated with
earnings and occupations disclosed by the CPS respondents years, even decades,
into their careers, during which they may have worked in different kinds of
occupations.

Id. at 64890, 64,955–56. When the DOE rejected the entirety of the underreporting problem

identified by commenters, it stated that the commenters’ most “important[]” shortcoming was the

12
lack of an alternative. Id. at 64,956 (“More importantly, the critique fails to demonstrate either

that a different and more reliable source of earnings data is available and should reasonably be

used instead of the SSA data, or that adjustments must be made based on CPS data.”). A court in

this district held that the DOE considered alternatives and that the SSA data was the best-

available data, notwithstanding the fact that it was imperfect. Ass’n of Private Sector Colls. &

Univs. v. Duncan (APSCU II), 110 F. Supp. 3d 176, 195 (D.D.C. 2015). The D.C. Circuit

affirmed that opinion. 640 F. App’x 5 (D.C. Cir. 2016).

AACS suggests—and the DOE does not directly dispute—that it is very difficult for

cosmetology schools to gather their own data for use in administrative appeals. Many

cosmetology schools operate in states that do not maintain state-sponsored data and the schools

anticipate that they will be unable to obtain the student responses required to use institutional

data. See Decl. of Mez Varol, President of International Academy (“Varol Decl.”) ¶¶ 8–9, ECF

No. 8-3; Decl. of Robert Rosenberg, President of Artistic Nails & Beauty Academy (“Rosenberg

Decl.”) ¶¶ 8–9, ECF No. 8-2. The articles and testimony that Plaintiff cites generally support the

idea that it is difficult to gather accurate earnings data for self-employed, heavily cash-earning

individuals. See generally Brodie Decl. ¶ 3. Thus, AACS argues, schools will be effectively

unable to appeal their classification as having failed, or fallen into the zone, the D/E rates.

Plaintiff argues that, despite no school having been deemed ineligible for federal student

loans under the HEA, the schools are already suffering harm due to the GE rules. Plaintiff’s

affiants state that the mere publication of the 2015 D/E rates has resulted in a drop in enrollment,

in one case by at least 30 percent. Varol Decl. ¶ 13; Rosenberg Decl. ¶¶ 22–23. For the handful

of schools that have failed the D/E ratio and must post a warning to their students, Plaintiff states

that such a warning would result in “insurmountable” drops in enrollment that would force those

13
schools to end certain programs. See Rosenberg Decl. ¶¶ 24–25 (“Based on my experience to

date, the compelled . . . warning . . . will result in an insurmountable drop in enrollments . . .

[and] ultimately force the school to end its barbering course within weeks,” which is a necessary

course to cover its operating expenses, and thus lead to closure of the school.); Varol Decl.

¶¶ 17–18 (reasoning that because the warning would lead to a 50% drop in enrollment and “the

school needs a constant flow of new enrollees . . . to remain in business,” publication of the

warning would “likely lead to the closure of the school almost immediately”). At least three

AACS member schools have already had to post the warnings. See Decl. of Edward M. Cramp

(“Cramp Decl.”) ¶ 3, ECF No. 24-1.4 AACS specifically identified three schools—located in

California, Illinois, and Texas—after the parties briefed summary judgment. Id.

4
Plaintiff has not provided any information specific to these schools that would meet the
“irreparable harm” standard for a preliminary injunction. The D.C. Circuit “has set a high
standard for irreparable injury.” Mexichem Specialty Resins, Inc. v. EPA, 787 F.3d 544, 555
(D.C. Cir. 2015) (quoting Chaplaincy of Full Gospel Churches v. England, 454 F.3d 290, 297
(D.C. Cir. 2006)). Not only must a party show that a harm is beyond remediation, but it also
must show that it is “of such imminence that there is a clear and present need for equitable
relief.” See id. (quoting Chaplaincy of Full Gospel Churches, 454 F.3d at 297). Mere imminent
economic harm is generally insufficient to establish irreparable injury. Wisconsin Gas Co. v.
FERC, 758 F.2d 669, 674 (D.C. Cir. 1985) (“It is . . . well settled that economic loss does not, in
and of itself, constitute irreparable harm.”). “Recoverable monetary loss may constitute
irreparable harm only where the loss threatens the very existence of the movant’s business.” Id.
A movant asserting that its business’s existence is threatened must explain the threat with
specificity. See Nat’l Mining Ass’n v. Jackson, 768 F. Supp. 2d 34, 52 (D.D.C. 2011); see also
Nat’l ATM Council, Inc. v. Visa Inc., 11-cv-1803, 2017 WL 2257327, at *5 (D.D.C. May 22,
2017).
Plaintiff has made no specific showing that the schools who currently have warnings
posted are in imminent danger of closing. None of Plaintiff’s affiants has personal knowledge of
the current financial situations of schools with posted warnings, nor do they identify any
particular school that will have to close between now and July 2. At oral argument, Plaintiff’s
counsel conceded that he did not have any “specific data of irreparable harm with respect to any
of th[e] [three] schools” that posted the warnings, and referred to the information he did have as
“anecdotal.” Tr. at 20. Blanket statements that schools will have to close soon after posting the
warnings are insufficient to establish irreparable harm. See Mexichem Specialty Resins, Inc. v.
EPA, 787 F.3d at 555.

14
 IV. LEGAL STANDARD

In a typical case, the Court must grant summary judgment to a movant who “shows that

there is no genuine dispute as to any material fact and the movant is entitled to judgment as a

matter of law.” Fed. R. Civ. P. 56(a); see also Winston & Strawn, LLP v. McLean, 843 F.3d 503,

505 (D.C. Cir. 2016). But in the context of the APA, the Court’s review of the administrative

record is limited. Sierra Club v. Mainella, 459 F. Supp. 2d 76, 89 (D.D.C. 2006)

(citing National Wilderness Inst. v. United States Army Corps of Eng’rs, 2005 WL 691775, *7

(D.D.C. 2005)). It is the agency’s role to resolve issues of fact and regulate in accordance with

those facts. See id. The district court’s review is confined to determining whether, as a matter of

law, the evidence in the administrative record supports the agency’s decision. Citizens for

Responsibility & Ethics in Washington v. SEC, 916 F. Supp. 2d 141, 145 (D.D.C. 2013).

“Summary judgment thus serves as the mechanism for deciding, as a matter of law, whether the

agency action is supported by the administrative record and otherwise consistent with the APA

standard of review.” Id. (citing Richards v. INS, 554 F.2d 1173, 1177 & n. 28 (D.C. Cir. 1977)).

V. ANALYSIS

The DOE argues that the Court should enter summary judgment in its favor for two

reasons. First, it argues that the Court does not have subject-matter jurisdiction. Def.’s Cross-

Mot. Summ. J. (“Def.’s Mot. Summ. J.”) at 16–18, ECF No. 18-1. It contends that AACS

member schools have not been subject to final agency action—a prerequisite to suit under the

APA—because none of them have lost their federal student loan eligibility. Def.’s Mot. Summ.

J. at 16–18. And because AACS did not identify any school that has incurred an injury in its

complaint, the DOE also argues that AACS lacks standing to sue on behalf of its members.

Def.’s Reply Supp. Def.’s Cross-Mot. Summ. J. (“Def.’s Reply”) at 8, ECF No. 23. Second,

15
Defendant contends that it did not act arbitrarily or capriciously, because the DOE reasonably

concluded that underreporting of income is adequately deterred by legal penalties and the SSA

data is supplemented by an alternate-earning appeal process, and because the alternatives

proposed during notice and comment would not allow the DOE to differentiate among programs

to identify which are not producing gainfully employed graduates. AACS argues instead that it

is entitled to summary judgment in its favor. See generally Pl.’s Mot. Summ. J., ECF No. 8.5

The Court holds that AACS is entitled to summary judgment. Because the GE

designations need not be appealed to be final and requiring failing schools to issue written

warnings constitutes final agency action, at least some AACS member schools have been subject

to final agency action. AACS has associational standing to bring claims on those schools’ behalf

because it identified the particular programs that have been injured.

Moreover, the DOE acted arbitrarily and capriciously with respect to the underreporting

issue identified by commenters. The DOE openly acknowledged that underreporting is an issue,

even identifying cosmetology schools by name. Commenters pointed to data that seem to show

the issue is significant. In response to public comments, the DOE addressed the underreporting

issue by noting that underreporters are subject to civil and criminal penalties and that programs

may seek to have separate earnings data used through an alternate earnings appeal. The first

justification is a non sequitur given the context that penalties existed when the commenters’ data

was gathered and will continue to exist in the same form into the foreseeable future. The second

justification also comes up short. Although SSA data is a useful starting point that seems to be

highly accurate with respect to reported income, it is incomplete. To deal with that shortcoming,

5
In a telephonic conference, Plaintiff agreed to use its motion for a preliminary
injunction as its motion for summary judgment.

16
the DOE created an alternate-earnings appeal process, but in doing so unjustifiably made appeals

difficult to mount. To remedy the DOE’s arbitrary and capricious appeal requirements, the Court

removes barriers to submitting alternate earnings appeals.

But the Court goes no further. The DOE adequately explained its decision to reject the

alternatives presented during the notice-and-comment period: it stated that the alternative data

was too disconnected from particular programs to allow the DOE to identify which programs are

underperforming. The explanation that the DOE provided sufficed to show that it was not acting

arbitrarily or capriciously. Given that no better data or methodology was presented during

notice-and-comment and AACS has failed to show that it is DOE’s responsibility to come up

with a better formula for each distinct type of program, it is incumbent on each school to provide

better data or methodology—if it exists—during the appeal process.

A. Subject-Matter Jurisdiction

The DOE first argues that there has been no “final” agency action over which the Court

has jurisdiction for an as-applied challenge. See Def. Mot. Summ. J. at 16–18. It contends that

the earliest a program could be deemed ineligible for federal loan guarantees would be 2018, and

that schools could still submit an alternate earnings appeal. Def.’s Mot. at 17–18. Without a

“final agency action” having been taken against any AACS member school, the DOE argues that

AACS does not have associational standing to pursue this case. See Def.’s Mot. at 17 (“Plaintiff

here points to no [final] action taken against it or any of its member cosmetology schools.”);

Def.’s Reply at 8. However, the DOE concedes that the schools that have already had to post

warnings have been subject to final agency action. See Tr. Oral Arg. (“Tr.”) at 40–42, 50, ECF

No. 25. The Court will first analyze whether the DOE has taken final agency action, and then

turn to whether AACS has standing to maintain this action. Because requiring schools to post

17
warnings constitutes final agency action and some schools have already posted those warnings,

the Court concludes that final agency action has been taken against, at least, three member-

schools, and the AACS has associational standing to pursue claims on these schools’ behalf.

1. AACS Challenges Final Agency Action

The DOE argues that AACS has not identified a final agency action that DOE has applied

to AACS or a member-school. Def.’s Mot. Summ. J. at 16–17. In support of this argument,

DOE notes that “no program can become ineligible for Title IV funding because no such

decision has been issued,” and that member-schools can submit alternate earnings appeals to

avert losing federal funding. Def.’s Mot. Summ. J. at 17–18. As noted above, the DOE

concedes that the schools that have already had to post warnings have been subject to final

agency action. See Tr. at 40–42, 50.

Plaintiff seeks judicial review of the DOE’s GE regulations pursuant to the APA. Compl.

¶ 53. The APA provides for judicial review of “final agency action.” 5 U.S.C. § 704; see also

Hall v. Sebelius, 689 F. Supp. 2d 10, 17 (D.D.C. 2009) (“Courts may not . . . review non-final

agency actions.”). A final agency action is one that both “marks the consummation of the

agency’s decision making process” as opposed to action that is “of a merely tentative or

interlocutory nature,” and determines the rights or obligations of the regulated entities. Bennett

v. Spear, 520 U.S. 154, 177–78 (1997). “The fact that an agency will apply a regulation if

certain events take place in the future does not satisfy the final agency action requirement for

purposes of an as-applied challenge.” Nat’l Wildlife Fed’n v. EPA, 945 F. Supp. 2d 39, 45

(D.D.C. 2013). A party generally need not administratively appeal an action for it to be

considered “final” under the APA. 5 U.S.C. § 704 (“Except as otherwise expressly required by

statute, agency action otherwise final is final for the purposes of [§ 704] whether or not there has

18
been . . . an appeal to superior agency authority.”). This is because, unlike with the doctrine of

exhaustion, “the finality requirement is concerned with whether the initial decisionmaker has

arrived at a definitive position on the issue that inflicts an actual, concrete injury.” Williamson

Cty. Reg’l Planning Comm’n v. Hamilton Bank, 473 U.S. 172, 193 (1985); accord Darby v.

Cisneros, 509 U.S. 137, 144 (1993).

In Dole v. United Steelworkers of America, the Supreme Court analyzed OSHA

disclosure rules, which the Court characterized as rules that “protect by providing access to

information about what dangers exist and how these dangers can be avoided.” 494 U.S. 26, 28

(1990). In deciding what constitutes final agency action, the Court said that

[t]he promulgation of a disclosure rule is a final agency action that represents a
substantive regulatory choice. An agency charged with protecting employees
from hazardous chemicals has a variety of regulatory weapons from which to
choose: It can ban the chemical altogether; it can mandate specified safety
measures, such as gloves or goggles; or it can require labels or other warnings
alerting users to dangers and recommended precautions. An agency chooses to
impose a warning requirement because it believes that such a requirement is the
least intrusive measure that will sufficiently protect the public, not because the
measure is a means of acquiring information useful in performing some other
agency function.

Id. at 33–34.

Consistent with the reasoning under Dole, the warning DOE regulations require for

schools failing the GE rules constitutes final agency action once they are triggered. The

requirement for schools to post warnings was not an interlocutory measure aimed at acquiring

information for a later agency action, but rather a policy implemented with the aim of protecting

students. See 79 Fed. Reg. at 64,901 (“[T]he [warning] requirements . . . balance the need to

provide prospective students with critical information at a time when they can most benefit from

it with ensuring that the administrative burden for institutions is not unnecessarily increased.”).

It was the consummation of the DOE’s decisionmaking process with respect to schools that

19
failed under the GE rules and did not file appeals, thus triggering the regulations’ warning

obligations. Thus, at least some of the AACS schools have been subject to this final agency

action for two reasons. First—and most obviously—at least three of AACS’s member-schools

have already posted the required warnings. See Cramp Decl. ¶ 3. Indeed, the government

admitted as much at oral argument, and conceded that those schools have been subject to final

agency action. See Tr. at 40–42, 50. Second, neither AACS nor its member-schools need

administratively appeal the warning requirement once they are subject to it in order to challenge

the regulations under the APA. See 5 U.S.C. § 704. Therefore, the Court finds that AACS is

challenging final agency action as defined under the APA. 6

6
The flip-side of the DOE’s final-agency-action argument is that, to the extent Plaintiff
actually seeks to assert a facial challenge, such a claim is precluded. Because AACS challenges
final agency action and does not bring a facial challenge, the Court rejects the DOE’s argument,
which is apparently made in the alternative. See Tr. at 12. As the Supreme Court has
acknowledged on multiple occasions, “the distinction between facial and as-applied challenges is
not so well defined that it has some automatic effect or that it must always control the pleadings
and disposition in every case involving a constitutional challenge.” Citizens United v. FEC, 558
U.S. 310, 331 (2010); see also Elgin v. Dep’t of Treasury, 567 U.S. 1, 15 (2012); In re Methyl
Tertiary Butyl Ether Prod. Liab. Litig., 674 F. Supp. 2d 494, 504 (S.D.N.Y. 2009) (“There are
few areas of the law that are as confused and conflicted as the law governing facial challenges.”).
In the administrative-law context, the distinction between facial and as-applied challenges is
even more elusive, and has little analytical value. Plunkett v. Castro, 67 F. Supp. 3d 1, 21 n.8
(D.D.C. 2014). “The concept of facial and as-applied challenges comes from constitutional law,
not administrative law, and in that context scholars have recognized that often the as
applied/facial dichotomy represents nothing more than a distinction without a difference.” Id.
Indeed, focus on the dichotomy “has only served to confuse each and obscure the real issues that
animate the outcome in a given case.” Id. (quoting Alex Kreit, Making Sense of Facial and As-
Applied Challenges, 18 Wm. & Mary Bill Rts. J. 657, 660 (2010)). As a result, in the
administrative law context, courts often classify challenges based on whether it appears, from the
complaint, that the plaintiff is asserting that under no set of circumstances could the regulation be
in accord with the statute. See id. at 20 (noting that a “[c]ourt could not have classified [a]
challenge as a classic facial challenge because . . . it [was] not difficult to identify situations in
which [the regulations] could be validly applied even under [the p]laintiff’s reading of the
statute.” (internal quotation marks and alterations omitted)). Because AACS does not bring such
a challenge but instead seeks review of final agency action, the Court rejects the DOE’s
argument.

20
 2. AACS Has Associational Standing

Because the Court has a sua sponte obligation to ascertain that it has subject-matter

jurisdiction and Defendant mentioned the issue in its reply, the Court next analyzes whether

AACS has standing to bring claims on behalf of any of its member schools that has suffered an

injury as a result of the GE regulations. The DOE asserts that AACS lacks standing because it

was required to identify, by name, at least one member-school who actually had to post the

warnings at the time of pleading. See Def.’s Reply at 6–8. The Court concludes that AACS

indeed possesses such standing.

“An organization can have standing on its own behalf . . . or on behalf of its members.”

Abigail All. for Better Access to Developmental Drugs v. Eschenbach, 469 F.3d 129, 132 (D.C.

Cir. 2006) (internal citations omitted). Standing based on an organization’s own injury—

“organizational standing”—requires an organization, “like an individual plaintiff, to show actual

or threatened injury in fact that is fairly traceable to the alleged illegal action and likely to be

redressed by a favorable court decision.” Equal Rights Ctr. v. Post Properties, Inc., 633 F.3d

1136, 1138 (D.C. Cir. 2011) (internal quotations omitted). “Associational standing,” in

comparison, allows an organization to sue on behalf of its members to protect their interests.

Common Purpose USA, Inc. v. Obama, --- F. Supp. 3d ---, No. 16-cv-0345, 2016 WL 7496105,

at *4 (D.D.C. Dec. 30, 2016).

For an organization to sue on behalf of its members through associational standing, it

must show that (1) “its members would otherwise have standing to sue in their own right,” (2)

“the interests it seeks to protect are germane to the organization’s purpose,” and (3) “neither the

claim asserted nor the relief requested requires the participation of individual members in the

lawsuit.” United Food & Commercial Workers Union Local 751 v. Brown Grp., Inc., 517 U.S.

21
544, 553 (1996) (quoting Hunt v. Wash. State Apple Advert. Comm’n, 432 U.S. 333, 343 (1977)).

The third requirement “ha[s] been understood to preclude associational standing when an

organization seeks damages on behalf of its members.” Id. at 554. Thus, when an association

seeks prospective relief on behalf of its members, actual participation by individual members is

generally not required. Warth v. Seldin, 422 U.S. 490, 515 (1975) (“If in a proper case the

association seeks a declaration, injunction, or some other form of prospective relief, it can

reasonably be supposed that the remedy, if granted, will inure to the benefit of those members of

the association actually injured.”). But see Ctr. for Auto Safety v. Nat'l Highway Traffic Safety

Admin., 793 F.2d 1322, 1329 n.44 (D.C. Cir. 1986) (“Courts have required individual

participation in circumstances where there are conflicts of interest within the organization or

when a specific factual setting is needed to illuminate the issues.”).

Courts are somewhat split on whether an organization must identify affected members by

name to meet the first requirement—that members would otherwise have standing to sue in their

own right. In Summers v. Earth Island Institute, the Supreme Court required an association to

“identify members who have suffered the requisite harm” to establish associational standing.

555 U.S. 488, 499 (2009). Lower courts have found that, although an association must identify

members, it need not identify members by name at the pleading stage to fulfill the first prong of

the test for associational standing. See, e.g., Ass’n of Am. Physicians & Surgeons, Inc. v.

Sebelius, 901 F. Supp. 2d 19, 31 (D.D.C. 2012) (compiling cases); Bldg. & Const. Trades

Council of Buffalo & Vicinity v. Downtown Dev., Inc., 448 F.3d 138, 145 (2d Cir. 2006).

However, these same courts have suggested that an association may need to name specific

members at the summary-judgment stage. See Downtown Dev., Inc., 448 F.3d at 144–45 (“The

defendants’ argument that the persons allegedly injured must be identified by name might have

22
some validity if this litigation were at the summary judgment stage.”). This naming requirement

is imposed because standing “must be supported in the same way as any other matter on which

the plaintiff bears the burden of proof, [that is], with the manner and degree of evidence required

at the successive stages of the litigation.” Ass’n of Am. Physicians & Surgeons, Inc., 901 F.

Supp. 2d at 31 (quoting Bennett v. Spear, 520 U.S. 154, 167–68 (1997). At the pleading stage,

“general factual allegations of injury . . . may suffice . . . [because] we presume that general

factual allegations embrace those specific facts that are necessary to support the claim.” Bennett,

520 U.S. at 168 (quoting Lujan v. Defs. of Wildlife, 504 U.S. 555, 561 (1992)) (internal quotation

marks and citation omitted); accord Ass’n of Am. Physicians & Surgeons, Inc., 901 F. Supp. 2d

at 31. Once discovery on the issue is complete, the adequacy of the specific standing allegations

can be tested. Downtown Dev., Inc., 448 F.3d at 144–45.

Cases in this district suggest that a plaintiff–association must identify injured parties at

the pleading stage, not necessarily that they be named in the complaint. In Western Wood

Preservers Institute v. McHugh, the court, ruling on a motion to reconsider, merely identified

that courts had divergent views on whether an individual member needed to be identified by

name at the pleading stage. 292 F.R.D. 145, 148 (D.D.C. 2013). Rather than state that any court

was correct on the issue, the court simply concluded that the existence of diverging viewpoints

did not make the court’s underlying decision “clearly erroneous.” Id. In the two cases that the

Western Wood Preservers Institute court identified, the court did not squarely address the issue

of whether an individual member must be named, so long as it is identified. In Californians for

Renewable Energy v. Department of Energy, the court analyzed the injury of the only named

associational member because “it is ‘not enough to aver that unidentified members have been

injured.’” 860 F. Supp. 2d 44, 48 (D.D.C. 2012) (quoting Chamber of Commerce v. EPA, 642

23
F.3d 192, 199 (D.C. Cir. 2011)) (emphasis added). To conclude that an individual member must

be named to ensure that it is identified is unwarranted. In the second case cited by the court in

Western Wood Preservers Institute, the court merely held that because the plaintiffs did not

respond to the defendants’ argument that an organization must “specifically identif[y] . . .

members who suffered the requisite harm,” the plaintiffs conceded they did not have

associational standing. Common Cause v. Biden, 909 F. Supp. 2d 9, 21 n.6 (D.D.C. 2012), aff’d

on other grounds, 748 F.3d 1280 (D.C. Cir. 2014). These cases do not require a plaintiff to

specifically name an injured member in the complaint so long as it is identified. The DOE did

not take a contrary position at oral argument once it became clear that AACS would name

specific members in a supplemental filing. Tr. at 41–43.

AACS meets all three requirements for associational standing. AACS has shown that at

least some of its members would have standing to sue in their own right. AACS identified

certain member-schools that did not appeal the final GE rates, and thus were required by the

regulations to issue the warnings. Civitano Decl. ¶ 32 (stating that “AACS member schools who

cannot appeal . . . and who did not file a notice of appeal . . . [were] required to deliver the

warning . . . no later than February 8, 2017). And even assuming that AACS was required to

identify specific schools by name at the summary judgment stage, it did so, naming three schools

who have already had to post the warnings. Cramp Decl. ¶ 3.

Second, this lawsuit is germane to the purpose of AACS. AACS is a nonprofit

association representing cosmetology programs, most of whom are subject to the GE rules

promulgated by the DOE. Compl. ¶ 1; Civitano Decl. ¶ 6. Ensuring that member schools

maintain Title IV federal loan assistance eligibility is in line with the purpose of the group,

which is to represent the interests of its member cosmetology schools.

24
 Third, this case does not require the participation of individual member-schools. AACS

seeks injunctive and declaratory relief on behalf of the member-schools, not damages. See

Compl. at 10. Thus, as a general rule, individual participation is not necessary. See Warth, 422

U.S. at 515. The DOE has not identified anything about this case that leads the Court to believe

individual participation is necessary for any other reason; there are no apparent conflicts of

interest that could arise, and the Court does not need individualized facts from the member-

schools to resolve the case. See Ctr. for Auto Safety, 793 F.2d at 1329 n.44. Thus, AACS has

established that it has associational standing to pursue the claims on behalf of its member schools

that have suffered injury as a result of the GE regulations.

B. Arbitrary and Capricious

Finally, having run the gauntlet of the DOE’s threshold arguments, the Court now

reaches the merits of AACS’s challenge to the GE regulations. AACS argues that the DOE acted

arbitrarily and capriciously in disregarding the widespread underreporting of income in the

cosmetology industry. See Pl.’s Mot. Summ. J. at 12. AACS specifically contends that the DOE

(1) knowingly used income data that was faulty and that would lead to seriously flawed results,

despite commenters’ concerns, and (2) did not consider credible alternatives. See Pl.’s Mot.

Summ. J. at 12. In support of its first argument, AACS asserts that not only is the SSA data

incorrect, but that the DOE actually acknowledged that it failed to take underreporting into

account in response to commenters. See Pl.’s Mot. Summ. J. at 17–19. Plaintiff further asserts

that the DOE failed to adequately explain why it did not account for underreporting, because

civil and criminal penalties are irrelevant to the issues of gainful employment. See Pl.’s Mot.

Summ. J. at 19–20. To support its argument that viable alternatives existed, Plaintiff cites to

proposals by commenters during notice and comment, and argues that the DOE could have

25
conducted a survey to estimate unreported income levels in the industry and adjusted each

school’s graduates’ earnings accordingly. See Pl.’s Mot. Summ. J. at 8–9. The Court will

analyze each of these arguments in turn, but first addresses the DOE’s argument that the

availability of appeal gives its regulations more leeway.

1. The Court Analyzes Whether the Entire Scheme is Arbitrary or Capricious

The DOE argues that the Court should analyze whether the adoption of the entire D/E-

calculation scheme was arbitrary and capricious as applied to AACS member schools, as

opposed to only focusing on the DOE’s presumptive use of SSA data. Def.’s Mot. Summ. J. at

8–9, 17–18; Def.’s Reply at 4–5. AACS argues that the Court must look at each piece

separately, meaning that the alternate-earnings appeal process has no bearing on whether the

DOE’s decision to presumptively use SSA data was arbitrary and capricious. The Court thus

addresses, as a threshold matter, whether the existence of the alternate-earnings appeal process

alters the Court’s standard of review with respect to the DOE’s use of SSA data. See Connors v.

Cedar Coal Co., No. 90-cv-0260, 1991 WL 102640, at *2 (D.D.C. June 4, 1991) (“As a

threshold matter the Court must determine the appropriate standard of review to apply . . . .”).

In general, the court must look to an agency’s final rule as a whole to determine whether

the agency gave substantial consideration to issues raised during the notice-and-comment period.

See Chamber of Commerce of United States of Am. v. NLRB, 118 F. Supp. 3d 171, 217–18

(D.D.C. 2015). One cannot assert that a particular portion of a regulation is arbitrary or

capricious because it does not address a shortcoming when another part of the regulation

squarely addresses it. See id.

The sole authority that AACS cites for the proposition that the Court should consider the

DOE’s use of SSA data in a vacuum is Leather Industries of America, Inc. v. EPA. There, the

26
plaintiff challenged the EPA’s adoption of certain caps on pollutants. 40 F.3d 392, 399 (D.C.

Cir. 1994). The caps themselves were not strict requirements, but rather one potential means of

meeting EPA standards. See id. (“In defending the . . . caps, the EPA suggests that because the .

. . caps do no more than offer land appliers an additional option, rather than impose a mandatory

requirement, they should withstand review. Land appliers need not comply with [the caps] . . .

.”). But if an entity’s pollutants did happen to meet the cap, it avoided significant regulatory

control, recordkeeping obligations, and labeling requirements. See id. Because the caps

provided “significant relief from the otherwise controlling regulatory standards,” the D.C. Circuit

reasoned that, by the same token, the caps created “not insignificant burdens” for entities who

could not meet them. Id. The court thus held the caps to the same standard that it would have

had it been the sole means of passing regulatory muster. Id.

Later, in American Iron and Steel Institute v. EPA, the D.C. Circuit analyzed EPA limits

on water pollutants promulgated under the Clean Water Act. 115 F.3d 979, 985 (D.C. Cir.

1997). To discharge pollutants into the Great Lakes, the Clean Water Act requires entities to

have permits, each of which comes with specific discharge limits based on, among other things,

the quality of water and the levels of pollution-control technology. Id. at 990. To formulate

specific discharge limits for entities, the EPA used two different methodologies, known as “Tier

I” and “Tier II.” Id. at 991. The difference between Tier I and Tier II is the amount of

toxicological data they required to calculate discharge limits. Id. Tier I data required

significantly more toxicological data than Tier II. Tier II was used as a fallback methodology

when insufficient toxicology data was available. Id. at 991 n.5. Tier II also resulted in more

stringent toxicology limits, making compliance more difficult for certain regulated entities. Id. at

993. The plaintiff, the American Iron and Steel Institute, whose limits were determined by Tier

27
II, complained that the regulations unfairly subjected it to more stringent requirements than it did

to entities in situations where more toxicological data was available. See id. The D.C. Circuit

upheld the tiered approach because the methodologies were based on the best-available data in

the face of imperfect data, and differing compliance burdens on industry did not automatically

make something arbitrary or capricious. Id. Distinguishing Leather Industries of America, Inc.,

the circuit held that “[t]he Tier II methodology, by scaling the uncertainty factors to reflect

existing data, properly correlates risk with knowledge.” Id.

These cases are no doubt similar. The major issue that the D.C. Circuit found with the

“optional” regulatory caps in Leather Industries was that meeting them excused certain regulated

entities from significant legal obligations. 40 F.3d at 399. Thus, if the caps were arbitrary,

entities would be subjected to burdensome regulations randomly based on the types of pollutants

they produce, while others would avoid the scheme altogether. Id. In comparison, in American

Iron and Steel Institute, the differing standards were created based on the relative availability of

data. 115 F.3d at 993. Although entities able to use the Tier I methodology were ultimately

subjected to laxer standards, those standards were the result of having more accurate data

available. See id. It was not as if those subjected to Tier II standards were forced to adhere to

more stringent standards separate from the overall regulatory scheme; the EPA decided to “err on

the side of overprotection” in the face of uncertainty. Leather Industries, 40 F.3d at 408

(quotation marks omitted) (quoting Am. Petroleum Inst. v. Costle, 665 F.2d 1176, 1186 (D.C.

Cir. 1981)). Plaintiff has not cited to any cases, following Leather Industries or otherwise,

suggesting that agencies cannot require regulated entities to undergo different compliance

processes based on the relative availability of data. See generally Pl.’s Opp’n.

28
 There is no reason for the Court to analyze the DOE’s use of SSA data in a vacuum. As a

general matter, the Court looks to the entire regulatory scheme to determine whether an agency

acted arbitrarily and capriciously. Chamber of Commerce of United States of Am., 118 F. Supp.

3d at 217–18. Leather Industries, the sole case relied upon by AACS, does not affect the Court’s

analysis. The DOE’s GE regulations do not exempt programs from the regulatory scheme

altogether if they happen to have a certain level of reported income. See 34 C.F.R. § 668.409.

Instead, the GE regulations permit multiple avenues of compliance based on the relative

availability of information, as in American Iron and Steel Institute. See id. The DOE

presumptively uses SSA data to determine graduates’ earnings, because it is a highly accurate

indicator of reported income, and no better source of earnings data broadly exists. See APSCU

II, 110 F. Supp. 3d at 195; 79 Fed. Reg. at 64,956. But for situations where reported income is

an insufficient representation of a program’s average graduate-earnings, the DOE created a

separate—and unavoidably more burdensome—process for calculating earnings based on

surveying an individual program’s specific graduates. See 34 C.F.R. §§ 668.406, 668.409.

Although the alternate-earnings appeal process makes compliance with the regulatory regime

more difficult for certain entities, the difficulty is the result of imperfect data, not an arbitrary

designation that reported earnings are superior to unreported earnings. Thus, rather than look at

the DOE’s use of SSA data in a vacuum, the Court analyzes the DOE’s regulatory scheme as a

whole to determine whether it is arbitrary and capricious as applied to AACS member schools.

See American Iron and Steel Institute, 115 F.3d at 993.

2. The DOE Acted Arbitrarily and Capriciously with
Respect to the Problem of Underreporting

AACS argues that the DOE uses “income data that is manifestly bad in the context of the

application of the GE Rule” to its member-schools. Pl.’s Mot. at 17; see also Compl. ¶ 1.

29
Plaintiff notes that the DOE even acknowledged that the data was flawed when it stated that it

was aware that “some self-employed individuals may fail to report, or underreport, their

earnings,” and that experts submitted evidence that tip income and self-employment income are

underreported by as much as 60% in, among others, the cosmetology industry. Pl.’s Mot. at 18–

19. AACS further argues that the DOE’s justification for accepting the flaws in the data—that

there are civil and criminal penalties for failing to report taxable income—is not reasoned, and

has no rational connection to the issues raised by commenters. Pl.’s Mot. at 19–20. DOE

responds that its justification regarding civil and criminal penalties was rational and that

programs can seek a different metric through an alternate-earnings appeal. Def.’s Mot. Summ. J.

at 21–23.

Under 5 U.S.C. § 706(2)(A), a reviewing court must set aside agency actions that are

arbitrary or capricious. The touchstone of arbitrary-and-capricious review is reasoned

decisionmaking. Harry T. Edwards, Linda A. Elliott & Marin K. Levy, The Requirement of

Reasoned Decisionmaking: Arbitrary and Capricious Review Under the APA, Federal Standards

of Review Ch. XV (Apr. 2013) (“Federal Standards of Review Ch. XV”) (citing Motor Vehicle

Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43 (1983)). Thus, a

court cannot set aside an agency rule that is “rational, based on consideration of the relevant

factors[,] and within the scope of the authority delegated to the agency by the statute.” State

Farm, 463 U.S. at 42–43. Although this is a deferential standard, it still requires a reviewing

court to take a “hard look” at an agency’s reasoning. See Federal Standards of Review Ch. XV

(quoting Nat’l Lime Ass’n v. EPA, 627 F.2d 416, 451 n.126 (D.C. Cir. 1980)). In the context of

notice-and-comment rulemaking, “[t]he function of the court is to assure that the agency has

given reasoned consideration to all the material facts and issues.” Greater Boston Television

30
Corp. v. FCC, 444 F.2d 841, 851 (D.C. Cir. 1970). This requires the agency to “articulate with

reasonable clarity its reasons for decision, and identify the significance of the crucial facts.” Id.

An agency action is arbitrary and capricious if the agency fails to consider all of the

relevant factors in reaching its decision. See N.Y. Cross Harbor R.R. v. Surface Transp. Bd., 374

F.3d 1177, 1181 (D.C. Cir. 2004). An agency rule is also arbitrary and capricious if the agency

“entirely failed to consider an important aspect of the problem, offered an explanation for its

decision that runs counter to the evidence before [it], or [the explanation] is so implausible that it

could not be ascribed to a difference in view or the product of agency expertise.” State Farm

Mut. Auto. Ins. Co., 463 U.S. at 43. Thus, an “agency must examine the relevant data and

articulate a satisfactory explanation for its action including a rational connection between the

facts found and the choice made.” Int’l Ladies’ Garment Workers’ Union v. Donovan, 722 F.2d

795, 814 (D.C. Cir. 1983) (quotation marks and citation omitted). An agency also must give a

reasoned response explaining how it resolved any significant problems raised during notice and

comment. Id. at 818. When an agency’s reasoning involves a non-obvious, essential factual

assumption, the agency must justify that assumption “notwithstanding a party’s failure to

challenge [it], as part of its affirmative duty to engage in rational decisionmaking.” Am. Mar.

Ass’n v. United States, 766 F.2d 545, 566 n.30 (D.C. Cir. 1985).

With that said, “[a]n agency has discretion to design rules that can be broadly applied,

sacrificing some measure of ‘fit’ for administrability,” but “must justify its failure to take

account of circumstances that appear to warrant different treatment for different parties.’”

Leather Indus. of Am., Inc., 40 F.3d at 403 (quoting Petroleum Commc’ns v. FCC, 22 F.3d 1164,

1172 (D.C. Cir. 1994)); see also Am. Pub. Gas Ass’n v. Fed. Power Comm’n, 567 F.2d 1016,

1046 (D.C. Cir. 1977) (internal quotations and citations omitted) (“Courts cannot fairly demand

31
the perfect at the expense of the achievable.”). Shortcomings and analytical weak points, if

justified or explained in light of practical constraints, do not render a regulation arbitrary or

capricious. See City of Brookings Mun. Tel. Co. v. FCC, 822 F.2d 1153, 1168 (D.C. Cir. 1987).

Indeed, even the “best available data standard leaves room for error, so long as more reliable data

did not exist at the time of the agency decision.” Baystate Med. Ctr. v. Leavitt, 545 F. Supp. 2d

20, 49 (D.D.C. 2008), amended in part, 587 F. Supp. 2d 37 (D.D.C. 2008), judgment entered,

587 F. Supp. 2d 44 (D.D.C. 2008).

The DOE’s overall methodology for determining a program’s average income is arbitrary

and capricious. The DOE certainly had discretion to use SSA data as the presumptive measure

of average income for GE programs. The Court has no reason to believe that SSA data is

anything but precise with respect to reported earnings, and AACS does not challenge the

regulations on the grounds that SSA data is in any way flawed with respect to reported income.

See generally Pl.’s Mot. Summ. J. Nor does it appear categorically flawed to presume that

reported earnings are the same as overall earnings in many cases. Indeed, a court in this district

has already held that the DOE “used the Social Security data . . . only after determining that no

better data existed, . . . only after rejecting other possible sources of data[, including the BLS,] as

inadequate . . ., and only after answering commenters’ concerns regarding the data.” APSCU II,

110 F. Supp. 3d at 195. The D.C. Circuit affirmed that opinion. 640 F. App’x 5. Although the

SSA data is concededly incomplete because it measures reported income instead of total income,

the DOE has discretion to sacrifice some measure of fit for the sake of administrability. See

Leather Indus. of Am., Inc., 40 F.3d at 403; see also Am. Pub. Gas Ass’n, 567 F.2d at 1046; City

of Brookings Mun. Tel. Co., 822 F.2d at 1168; Baystate Med. Ctr. v. Leavitt, 545 F. Supp. 2d at

32
49. And devising formulas for each program based on its own unique mix of self-employment,

cash income, and tip income would quickly become unadministrable.

But wooden use of this SSA data is problematic. As commenters noted during the notice-

and-comment period, graduates of certain programs—including cosmetology programs—

underreport their income with far greater frequency than graduates of other programs. J.A. 797;

see also Civitano Decl. ¶ 28. One commenter, citing a Stanford study, noted that both tip and

self-employment income are underreported by around 60%. J.A. 595, 787; see also Civitano

Decl. ¶ 28; Brodie Decl. at 493. The DOE explicitly acknowledged this problem, noting that tips

make up “about half of earnings in service occupations such as cosmetology,” and that many

people in the industry are self-employed. See 79 Fed. Reg. at 64,955. So, although SSA data is

highly accurate in some respects and is thus a useful starting point for determining average

earnings, it has significant shortcomings—at least for certain occupations.

Pursuant to its obligations to respond to significant problems raised during the notice-and

comment period, see Int’l Ladies’ Garment Workers’ Union, 722 F.2d at 818, the DOE dealt

with the underreporting issue in two ways. First, it relied on the civil and criminal penalties

associated with failing to report income. 79 Fed. Reg. at 64,955–56. Second, it noted that SSA

data is only one of three possible sources of earnings data; programs could also appeal for the

use of state-sponsored data or survey-based data. Id. at 64,956–57.

The first justification does not grapple with the issue in any meaningful way. The DOE

obviously did not demonstrate that the legal disincentives for failing to report meant that the

commenters’ data was incorrect; the DOE itself acknowledged that the problem existed. See 79

Fed. Reg. at 64,955. There is also no basis for the inference that the issue of underreporting

would somehow subside under the new GE regulations. Civil and criminal penalties existed

33
before and after the regulations took effect, and the DOE regulations did not add any additional

deterrent for underreporting. To the extent the DOE believed that the penalties somehow would

correct for underreporting, it was required to explain this assumption explicitly. See Am. Mar.

Ass’n v. United States, 766 F.2d at 566 n.30. The agency thus had no rational basis for

concluding that rates of reporting would change at all after the regulations were to go into effect.

This was not a “satisfactory explanation” for why the DOE chose to adopt the GE regulations

notwithstanding the issue of underreporting.

The DOE’s second response more directly addresses the commenters’ concerns, but

nonetheless comes up short. Although the commenters identified the issue of underreporting,

they did not provide a better data set that the DOE could use to measure overall income, nor did

they suggest a specific adjustment factor that the DOE could use for the various programs that

would inevitably claim underreporting as a problem with their earnings data. APSCU II, 110 F.

Supp. 3d at 195; 79 Fed. Reg. at 64,955–56. Nor did the commenters propose an alternative

calculus to balance fit and administrability. Id. at 64,955–56. The commenter’s difficulties

producing such data is understandable. The problem of underreporting extends across multiple

industries and even across individual entities within those industries. While cosmetology

schools’ graduates engage in, on average, a certain amount of underreporting, other industries

likely also experience different levels of underreporting based on factors like the amount of tips

their graduates earn, how frequently their graduates are self-employed, and the amount of tax-

compliance training their graduates receive. Within these industries, individual schools

experience varying levels of underreporting. A cosmetology school that frequently sends

graduates to nationwide chains or upscale salons (which may have higher percentages of

34
payments made by credit card) likely has lower rates of graduate underreporting than schools

that graduate large numbers of self-employed cosmetologists.

The DOE’s attempt to address these many issues came in the form of an alternate

earnings appeal process, which the DOE cited as its other justification for its use of SSA data in

the face of the underreporting problem. 79 Fed. Reg. at 64,956. On the surface, this justification

seems reasonable. Assuming that every program is capable of mounting an appeal, the data

would yield perfectly tailored D/E rates for individual programs regardless of their levels of

underreporting and the unique characteristics of the occupations in which their particular

graduates are employed. But that assumption is the fly in the DOE’s reasoned-decisionmaking

ointment. As noted above, the DOE “must justify the assumptions essential to its actions,

notwithstanding a party’s failure to challenge those assumptions before the agency, as part of its

affirmative duty to engage in rational decisionmaking.” Am. Mar. Ass’n, 766 F.2d at 567 n.30.

The DOE assumed that programs would be able to effectively use the alternate-earnings appeal

process. See generally 79 Fed. Reg. 64,890. It also assumed that the required response rates—

50 percent for state-sponsored data systems and nearly 100 percent for school-specific surveys—

had some statistical significance. See 79 Fed. Reg. at 64,995–96. But the DOE did not explain

these assumptions.7 See 79 Fed. Reg. at 64,995–96; see generally 79 Fed. Reg. 64,890. The

7
Although the DOE explains why it adopted the NCES’s survey standards by noting that
it expected the NCES, as a national institute devoted to collecting education data, to “balance the
need for reliable data with our intent to provide a meaningful opportunity for appeal,” it never
explained why it imposed the requirement that the survey be given to almost all students, and
indeed that the survey produce the “the alternate earnings of [almost] all the students who
completed the program” during the relevant time period. 79 Fed. Reg. at 65,010. Nor did the
DOE explain why it imposed the 50 percent requirement for state-sponsored surveys. See
generally 79 Fed. Reg. 64,890. Courts routinely accept survey evidence based on the statistical
significance of the survey size. See, e.g., Ethyl Corp. v. EPA, 541 F.2d 1, 51 n.112 (D.C. Cir.
1976); Sexcius v. District of Columbia, 839 F. Supp. 919, 924 n.11 (D.D.C. 1993). The law does
not generally require 100-percent response rates. See Ethyl Corp., 541 F.2d at 51 n.112.

35
DOE did not, for example, suggest that a certain amount of survey compliance ensures a certain

amount of accuracy, let alone weigh data accuracy against feasibility of compliance. See 79 Fed.

Reg. at 64,995–96. And despite not having the burden on this point, AACS has produced

evidence showing that cosmetology schools are simply unable to mount appeals because of the

onerous regulatory prerequisites to doing so. See Varol Decl. ¶¶ 8–9; Rosenberg Decl. ¶¶ 8–9.

In light of the DOE’s affirmative obligation to justify the assumptions underlying its alternate-

earnings appeal process, it has not met its burden of engaging in reasoned decisionmaking.8 See

Am. Mar. Ass’n, 766 F.2d at 566 n.30.

3. The DOE Did Not Act Arbitrarily or Capriciously in
Rejecting the Alternative Identified by AACS

AACS further contends that the DOE failed to adequately consider reasonable

alternatives presented during the notice-and-comment period. See Pl.’s Mot. Summ. J. at 20. It

specifically points to Dr. Bettinger’s suggestion that the DOE could use average-income figures

from the Bureau of Labor Statistics, which derives its data from surveys, rather than from

income reported to the IRS. The DOE responds that it adequately justified its rejection of the

Bureau of Labor Statistics’ data on the grounds that it could not be individualized to particular

programs. Def.’s Mot. Summ. J. at 9–11.

8
AACS suggests that the DOE should be responsible for gathering earnings data for
individual programs instead of forcing those programs to gather their own data in the context of
an appeal. See Pl.’s Mot. Summ. J. at 8–10. Absent some statutory obligation to the contrary,
the agency’s burden is to use the best available data, not the best possible data. See Baystate
Med. Ctr. v. Leavitt, 545 F. Supp. 2d 20, 41 (D.D.C. 2008), amended in part, 587 F. Supp. 2d 37
(D.D.C. 2008) (gathering cases), judgment entered, 587 F. Supp. 2d 44 (D.D.C. 2008); see also
APSCU II, 110 F. Supp. 3d at 195. The best-available-data standard “makes it clear that the
[agency] has no obligation to conduct independent studies.” Sw. Ctr. for Biological Diversity v.
Babbitt, 215 F.3d 58, 60 (D.C. Cir. 2000). Thus, AACS’s argument has no merit.

36
 The arbitrary-and-capricious standard described above also applies to an agency’s

consideration of regulatory alternatives. See Pillai v. Civil Aeronautics Bd., 485 F.2d 1018, 1027

(D.C. Cir. 1973) (finding the “artificial narrowing of options to be arbitrary and capricious”);

accord Int’l Ladies’ Garment Workers’ Union, 722 F.2d at 817. This does not mean that an

agency must consider all possible policy alternatives in reaching its decision. State Farm Mut.

Auto. Ins. Co., 463 U.S. at 51. But the agency must address obviously germane alternatives

proposed by commenters during the notice-and-comment period. See Int’l Ladies’ Garment

Workers’ Union, 722 F.2d at 817–18. The agency must give rational consideration to each of

these alternatives and an “adequate explanation” for each rejection. See id. (citing Action on

Smoking and Health v. Civil Aeronautics Bd., 699 F.2d 1209 (D.C. Cir. 1983), supplemented

by 713 F.2d 795 (D.C. Cir. 1983)).

The DOE adequately explained why it did not use the Bureau of Labor Statistics data as

opposed to SSA data. Indeed, a court in this district has already held that to be the case. APSCU

II, 110 F. Supp. 3d at 195 (holding that the DOE used the best-available data and decided on

SSA data “only after rejecting other possible sources of data as inadequate . . . and only after

answering commenters’ concerns regarding the data”). As noted above, the GE regulations were

created to identify particular schools whose graduates were not making sufficient income to pay

back their student debt. See 20 U.S.C. § 1094; see also APSCU I, 681 F.3d at 435–36. The DOE

reasonably concluded that Bureau of Labor Statistics data would detract from that goal, because

the data would have to be generalized across fields and then applied to individual schools. 79

Fed. Reg. at 64,955–56. The data would be an average of the graduates of all cosmetology

programs, meaning that the lower-performing schools, whose graduates make less money on

37
average, would benefit from higher-earning graduates of other schools, and vice versa. Id. at

64,955.

Plaintiffs suggest that this problem could be alleviated by adding an adjustment factor to

the SSA data for each school. Pl.’s Mot. at 9–10. Thus, schools with graduates who have lower

reported income will have proportionately low levels of unreported income added to their

numbers. Pl.’s Mot. at 9–10. As the DOE concluded in its response to notice and comment, this

does not necessarily solve the problem. This methodology assumes that each program’s

graduates are employed in the industry for which they are trained at proportionate rates. See 79

Fed. Reg. at 64,955–56. Thus, if a cosmetology program’s graduates cannot find work as

cosmetologists and therefore must work in other fields with higher rates of income reporting, that

program would still benefit from the average rate of underreporting in the cosmetology industry

if the adjustment-factor methodology is used. See id. It is also possible that the survey

respondents, who self-identify their careers, have no connection whatsoever to the GE programs,

let alone the programs’ present-day graduates. See id. at 64,956.

Whether the DOE’s decision to reject the Bureau of Labor Statistics approach suggested

by Dr. Bettinger was correct is not at issue here. The Court merely asks whether the decision

was reasoned. Because it was rational, based on a consideration of the need to individualize its

review of schools, and its reasoning was explained, the Court finds that the DOE did not act

arbitrarily and capriciously with respect to the alternative identified by AACS.

C. Remedy

Given that the DOE acted arbitrarily and capriciously with respect to the problem of

underreporting, the Court now must fashion a remedy. In general, courts narrowly tailor

remedies to APA violations. Nat’l Treasury Emps. Union v. Chertoff, 394 F. Supp. 2d 137, 145

38
(D.D.C. 2005), rev’d in part on other grounds, 452 F.3d 839 (D.C. Cir. 2006) (citing Nat’l

Treasury Emps. Union v. Yeutter, 918 F.2d 968, 977 (D.C. Cir. 1990)). Thus, when possible, a

court should strike only offending portions of regulations. See id.; Gulf Oil Corp. v. Brock, 778

F.2d 834, 842 (D.C. Cir. 1985).

To remedy the DOE’s arbitrarily and capriciously rigid use of SSA data to calculate D/E

rates, the Court orders that the DOE may not enforce the numerical survey requirements

currently in effect for alternate earnings appeals against AACS member schools. In effect, this

removes the arbitrary and capricious reasoning behind the otherwise-valid premise that alternate

earnings appeals justify the presumptive use of SSA data. It also avoids upending the entire GE

regulatory scheme and means that AACS member institutions need not secure any specific

amount of survey responses or state-sponsored data to raise an appeal. AACS member schools

will have broader, more feasible options to challenge their D/E rates before they become final,

and the DOE will be able to decide, on a case-by-case basis, what modicum of evidence is

enough to overcome the presumption in favor of using SSA data for each particular program.9

This remedy will also lead to more tailored administrative-appeal records for District Courts to

review in lieu of broad challenges to the overall regulations.

9
As noted above, at least three schools already declined alternate earnings appeals, and
have thus had to provide the required warnings to their students. Cramp Decl. ¶ 3. Because the
available appeals process was arbitrarily and capriciously narrow when they made their choices,
the Court orders the reopening of the appeal period for affected schools. Under 34 C.F.R.
§ 668.406(e)(2), affected schools may remove the warnings until they decide not to appeal or
their appeal fails. They must make their choice by the extended deadline that the DOE set in
March. See Emer. Mot. Extension Time, ECF No. 14.

39
 VI. CONCLUSION

For the foregoing reasons, the Court grants AACS’s motion for summary judgment,

denies the DOE’s motion for summary judgment, and denies AACS’s motion for a preliminary

injunction. An order consistent with this Memorandum Opinion is separately and

contemporaneously issued.

Dated: June 28, 2017 RUDOLPH CONTRERAS
United States District Judge

40