With the demise of Chevron deference, another significant deference doctrine warrants renewed examination:  the judicial deference accorded an agency’s findings of facts during an administrative hearing.  Courts have deferred to an agency’s decision when it is supported by substantial evidence by giving one word used twice in the Administrative Procedure Act (“APA”) two opposite meanings; one requires deference in the extreme, the other does not.   Experience has shown that this sort of exercise in linguistic creativity is unworkable.

Defining “Substantial” in “Substantial Evidence.”

The phrase “substantial evidence” is used twice in the APA.  First, in section 7(c), the APA states that at an administrative hearing, the proponent of an “order has the burden of proof” and an order may not be “issued except on consideration of the whole record or those parts thereof cited by a party and supported by and in accordance with the reliable, probative, and substantial evidence.”  Second, in section 10(e), the APA requires that on review, a court must “hold unlawful and set aside” a decision following a hearing which is “unsupported by substantial evidence.”  (Emphasis supplied).   

To most people, the adjective “substantial” has one meaning— “strong,” “solidly built,” “considerable.”   There is nothing ambiguous about it.  The phrase “substantial breakfast” does not mean a morning meal consisting of a single piece of plain, unbuttered toast and black coffee. 

The phrase “substantial evidence” in APA §§ 7(c) and 10(e) sounds like a high hurdle for an agency—or any other party who bears the burden of proof.  The courts, though, have inexplicably re-defined “substantial” in the phrase “substantial evidence” in APA § 10(e) to mean the opposite of its normal meaning— “more than a mere scintilla of evidence” is all that is necessary to support an agency’s factual determinations following an administrative hearing.  In contrast, “substantial evidence” as used in section 7(c) appears to carry its ordinary dictionary meaning. Thus, the agency must present a significant quantity of evidence to support its claim for relief, but only a fraction of that evidence on judicial review.  

This has three effects—one lexical and two legal:  it ignores the word’s plain meaning and instead transforms an otherwise unambiguous word into a contronym—a word with two opposite meanings; it undermines the preponderance of the evidence standard when the burden of proof is on the government; and it extends deference in an era when judicial reverence for agency action is on the wane. 

The saga of the schizophrenic judicial interpretation of “substantial evidence” can be traced, in part, to Consolidated Edison v. National Labor Relations Board, where the Court reviewed an NLRB decision.  Section 10(e) of the National Labor Relations Act of 1935 (“NLRA”) permitting judicial review, stated that “[t]he findings of the Board as to the facts, if supported by evidence, shall be conclusive.”  Consolidated Edison argued that the term “evidence” really meant “substantial evidence.”  The Court agreed, but ultimately held that “[s]ubstantial evidence is more than a mere scintilla. It means such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.”  Left unanswered was how this definition would work when the burden of proof rested with the government.  The Court’s failure to address this question was odd given that section 10(c) of the NLRA expressly imposed the burden on the government to prove its factual allegations by a “preponderance of the testimony.”  Suppose the government sought to prove its case by a preponderance of the testimony where the testimony supporting the agency position was only slightly more than a mere scintilla of evidence, but the petitioner’s evidence was of substantially greater weight.  Would the agency have satisfied the “substantial evidence” test?

The following year, the Court sought to clarify that “substantial evidence” “must be enough to justify, if the trial were to a jury, a refusal to direct a verdict when the conclusion sought to be drawn from it is one of fact for the jury.”   To direct a verdict in a civil case, the trial judge must find that “a reasonable jury would not have a legally sufficient evidentiary basis to find for the party on that issue,” where the party against whom the verdict is directed cannot prevail without “a favorable finding on that issue.”^[1]  This test necessarily incorporates the burden of proof and therefore, would appear to resolve the conundrum of how the preponderance of evidence standard meshes with the “substantial evidence” test on review. The apparent clarity, however, still eluded the courts.  

In perhaps the most intellectually honest “substantial evidence” opinion, Mr. Justice Frankfurter, writing for the Court in Universal Camera Corp. v. NLRB, candidly observed that attempting to craft a meaningful and useful definition for “substantial evidence” was a fool’s errand:  “A formula for judicial review of administrative action may afford grounds for certitude but cannot assure certainty of application.”  The Court stressed a theme that would re-gain currency more than 70 years later in Loper Bright Enter. v. Raimondo.  Justice Frankfurter stated: 

Reviewing courts must be influenced by a feeling that they are not to abdicate the conventional judicial function. Congress has imposed on them responsibility for assuring that the Board keeps within reasonable grounds. That responsibility is not less real because it is limited to enforcing the requirement that evidence appear substantial when viewed, on the record as a whole, by courts invested with the authority and enjoying the prestige of the Courts of Appeals. The Board’s findings are entitled to respect, but they must nonetheless be set aside when the record before a Court of Appeals clearly precludes the Board’s decision from being justified by a fair estimate of the worth of the testimony of witnesses or its informed judgment on matters within its special competence or both.

Justice Frankfurter’s words were overtaken by fuzzier and more agency-friendly opinions.  In Steadman v. Securities and Exchange Comm’n, for instance, Steadman challenged the preponderance-of-the-evidence standard used to determine whether he had violated the anti-fraud provisions of the federal securities laws, arguing instead that the clear-and-convincing evidence standard typically used in fraud cases ought to have been used in his hearing.  In affirming the SEC’s determination, the Court noted that 

[t]he language of § 7(c), therefore requires that the agency decision must be “in accordance with” the weight of the evidence, not simply supported by enough evidence “’to justify, if the trial were to a jury, a refusal to direct a verdict when the conclusion sought to be drawn from it is one of fact for the jury.'” Obviously, weighing evidence has relevance only if the evidence on each side is to be measured against a standard of proof which allocates the risk of error.  .  .  . Section 10(e), by contrast, does not permit the reviewing court to weigh the evidence, but only to determine that there is in the record “`such relevant evidence as a reasonable mind might accept as adequate to support a conclusion[.]'” 

Steadman, 450 U. S. at 99-100 (internal citations omitted, emphasis supplied).

The Steadman Court then made the rather remarkable statement that “Congress intended the words ‘substantial evidence’ to have different meanings in context[,]” i.e., a different meaning in § 7(c) than in § 10(e)(B)(5).  Id. at 100.  Even putting aside the logical and legal canons that words or phrases in the same statute should be given the same meaning and should normally carry their dictionary definitions, the Court’s statement makes no sense.  If, for example, an agency’s determination lacks the quantity of evidence, i.e., substantial evidence, necessary to carry its burden, it does not automatically get a free pass merely because on judicial review a court must give “substantial evidence” a more deferential meaning.  If an agency issued an order that lacked the necessary quantity of evidence required by § 7(c), its order would still be reviewable under § 10(e)(B)(1) (holding unlawful agency action “not in accordance with law”) even if it may escape meaningful review under § 10(e)(B)(5) (supported by substantial evidence).  Steadman appears to be inconsistent with the Court’s decision in Dir., Off. of Workers’ Comp. v. Greenwich Collieries (holding that APA § 7(c) applies to Black Lung adjudications and drawing no distinction between § 7(c) and the standard of review under § 10(e)).    

Finally, in Dickinson v. Zurko, the Court acknowledged that “substantial evidence,” for purposes of judicial review, was less exacting and more agency friendly than the clearly erroneous standard used by appellate courts when assessing the findings of fact of a district court.  Viewed in light of recent structural cases, this also makes no sense.  Where an ALJ’s decision is not reviewed by the agency head, it becomes final agency action subject to judicial review.  But ALJs are at most inferior officers and their decisions deserve less deference than a decision of a principal officer, such as a district judge. Zurko suggests otherwise and may not weather the Court’s recent decisions strictly enforcing the Constitution’s structural limits, especially under the Appointments Clause.

To summarize, “substantial evidence,” as used in APA § 10(e), is now a term of art, whose meaning therefore cannot be ordained from a dictionary or ordinary usage. It means more than a mere scintilla of evidence, enough to avoid a directed verdict if the matter were tried to jury, but potentially not enough under the “clearly erroneous standard” to avoid reversal if the factual findings had been made by a district judge—or even enough to constitute a preponderance of the evidence. And according to the D.C. Circuit, the “substantial evidence” standard and the arbitrary and capricious standard are one and the same.  

Reviewing courts have therefore been left with a hodgepodge of standards requiring them to give one word, “substantial,” as used in the APA, two opposite meanings, to thread the needle between the “directed verdict” and “clearly erroneous” standards, and to give the term “substantial evidence,” in § 10(e) the same meaning as “arbitrary and capricious,” also used in § 10(e).  Reviewing courts, though, are not linguistic contortionists nor should they be expected to give two different terms—substantial evidence and arbitrary and capricious–in the same subsection the same meaning, while giving the same term in the same Act two different meanings, no matter what the D.C. Circuit may say.  

Revisiting Substantial Deference to Agency Findings.

This baggage, accumulated over the past eight decades, has weighed down reviewing courts thwarting their independent judgment in favor of significant deference to the agency.  While the Court in Loper noted that historically, pre-Chevron, courts deferred to an agency’s findings of fact but gave no deference to an agency on questions of law, the Court emphasized that Article III “assigns to the Federal Judiciary the responsibility and power to adjudicate ‘Cases’ and ‘Controversies’—concrete disputes with consequences for the parties involved.”  Loper, slip op. at 7.  And under Article III, there is no reason, for deference purposes, to treat findings of fact and conclusions of law differently.   

With the fall of Chevron in Loper and the more emphatic recognition of the scope of Article III in both Loper and Securities and Exchange Comm’n v. Jarkesy, perhaps it is time to reconsider how a court should apply the “substantial evidence” standard by limiting deference and reverting to the common sense views of Justice Frankfurter that it would be foolhardy to attempt to define the term “substantial evidence.” Instead, courts should assess an agency’s findings of fact against a reasonableness standard while according those findings whatever respect they may be due.  Ironically, in light of Jarkesy, the debate over the meaning of “substantial evidence” may be moot in many settings;  Jarkesy calls into question the constitutionality of most administrative enforcement schemes including those at the NLRB and any that impose or seek to impose civil money penalties, e.g., Health and Human Services (“HHS”) Office for Civil Rights under HIPAA, HHS OIG under Medicare, Medicaid and certain other federal healthcare programs, FDA under the Food, Drug, and Cosmetic Act.

Robert P. Charrow (J.D. Stanford) is the former General Counsel of the U.S. Department of Health and Human Services and retired Principal Shareholder at Greenberg Traurig, LLP; Laura M. Klaus (J.D. Case Western) is the retired Managing Shareholder and retired Principal Shareholder at Greenberg Traurig, LLP.  Ms. Klaus was one of the counsel for the Respondent in Dir., Off. of Workers’ Comp. v. Greenwich Collieries, which is discussed in this writing.

---

^[1]        Fed. R. Civ. P. 50.  A verdict can be granted before a jury has reached its decision.  A judgment notwithstanding the verdict is granted after the jury has reached a verdict; the jury verdict is set aside the verdict in favor of the one reached by the court.  The two are very similar except for the timing within a trial.  Both now are called a “judgment as a matter of law.”