Judicial Deference to Executive Precedent
Author(s): Thomas W. Merrill
Source: The Yale Law Journal, Vol. 101, No. 5, (Mar., 1992), pp. 969-1041
Published by: The Yale Law Journal Company, Inc.
Stable URL: http://www.jstor.org/stable/796961
Accessed: 08/05/2008 16:13
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=ylj.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We enable the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.
http://www.jstor.org
Articles
Judicial Deference to Executive Precedent
Thomas W. Merrillt
In 1984, the SupremeCourt adopted a new frameworkfor determining
when courts should defer to interpretationsof statutesby administrativeagencies. Previousdecisions hadlooked to multiplecontextualfactorsin answering
this question.1Chevron U.S.A., Inc. v. National ResourcesDefense Council,
Inc.2appearedto reject this approachand requirethat federal courts defer to
any reasonableinterpretationby an agency chargedwith administrationof a
statute, provided Congress has not clearly specified a contraryanswer.The
Courtjustified this new generalrule of deferenceby positingthatCongresshas
implicitlydelegatedinterpretativeauthorityto all agencieschargedwith enforcing federal law.3
Chevronis widely regardedas a kind of "counter-Marbury"
for the administrativestate.4Indeed,readfor all it is worth,the decisionwould makeadmint Professor of Law, NorthwesternUniversity School of Law. I served as Deputy Solicitor General,
Departmentof Justice, from 1987 to 1990, during which time I arguedor helped brief a large numberof
the Supreme Court cases involving the "Chevrondoctrine."The analysis of these cases (and all others
pending while I was with the Justice Department)is strictly my own, and not necessarilythe same as that
of the Justice Departmentor its client agencies. I wish to thankJustice Antonin Scalia, the 1991 Howard
J. Trienens Visiting Judicial Fellow at NorthwesternSchool of Law, and my colleagues at Northwestern
for a stimulating discussion at a faculty workshop.Special thanks to Robert Bennett, Steven Calabresi,
CharlotteCrane,Keith Hylton, Paul Larkin,Gary Lawson, MartyRedish, and Peter Schuck for comments
on an earlier draft, and to Melissa McGonigal for her diligent researchassistance.
1. See infra notes 7-29 and accompanyingtext.
2. 467 U.S. 837 (1984).
3. Id. at 843-44, 865-66.
4. Cass R. Sunstein, Law and AdministrationAfterChevron,90 COLUM.
L. REV.2071, 2075 (1990).
969
970
The Yale Law Journal
[Vol. 101: 969
istrative actorsthe primaryinterpretersof federalstatutesand relegate courts
to the largely inertrole of enforcingunambiguousstatutoryterms.This in turn
would have enormousimplicationsfor the overallbalanceof poweramongthe
three branches of government.Executive branch agencies would gain new
power to achieve rapid changes in policy through reinterpretationof their
legislative authority;courts would play a diminishedrole in checking agency
aggrandizementandin protectingrelianceinterestsassociatedwithpastinterpretations;andCongress,unableto rely on the courtsto honorunstatedinstitutional understandings,could react by enacting excruciatinglydetailed statutesor
intensifying the use of oversighthearings.
This Article has two primarytasks.The first, which is the subjectof Parts
I, II, and III, is to determinewhetherChevronin fact operatesin the Supreme
Courtas the "counter-Marbury"
portrayedby commentators.It turnsout that
the Courtdoes not regardChevronas a universaltest for determiningwhen to
deferto executive interpretations:the Chevronframeworkis used in only about
half the cases that the Courtperceivesas presentinga deferencequestion.Nor
have the multiple factorsidentifiedin the pre-Chevronperiod disappeared;to
the contrary,the Courtcontinuesto rely uponthemin manycases, despitetheir
apparentirrelevanceunderChevron.Perhapsmost strikingly,in recentTerms
the applicationof Chevronhasresultedin less deferenceto executiveinterpretations than was the case in the pre-Chevronera. Thus, instead of functioning
as a "counter-Marbury,"
there are signs that Chevronis being transformedby
the Courtinto a new judicial mandate"to say what the law is."5I will argue
that the failure of Chevronto perform as expected can be attributedto the
Court'sreluctanceto embracethe draconianimplicationsof the doctrinefor the
balance of power among the branches,and to practicalproblemsgeneratedby
its all-or-nothingapproachto the deferencequestion.
The second task of the Article is to proposean alternativeto Chevronthat
avoids its theoretical and practical failings and is more congruentwith the
actualpracticeof the SupremeCourt.PartIV sets forthwhatI call the "executive precedentmodel." The root idea is that executive interpretationsshould
be viewed as a form of precedentand integratedinto the process of statutory
constructionin much the same way thatjudicial precedentis integrated.The
model begins with the understandingthat both the executive branchand the
judicial branchhave inherentauthorityto interpretenacted law-and thus to
generate"precedent"construingthatlaw.The modelthenpositsthatthejudicial
power includesnot only the powerto determinewhetherto follow pastjudicial
precedentinterpretingenactedlaw, butalso the powerto determinewhatweight
to give to past executive precedentconstruingthat law. Under the proposed
model,courtswould treatexecutiveprecedentapproximatelythe same way they
regardthe precedentof anothercourtof coordinatejurisdiction:as primafacie
5. Marburyv. Madison, 5 U.S. (1 Cranch) 137, 177 (1803).
1992]
Executive Precedent
971
correct,and as entitledto moreor less deferencedependingon variouscontextual factors(such as the thoroughnesswith which the executive agency considered the issue, how long the executive interpretationhas been followed, and
so forth).
Part V comparesChevron'sdeferencemodel and the executive precedent
model. Chevronrests on a principleof mandatorydeference:courts are compelled to defer to agency interpretationsbecause Congress has directedthem
to defer. The executive precedentmodel, in contrast,rests on a principle of
discretionarydeference:courtsdeferto executiveinterpretationsbecausesound
judicial decisionmakingrequiresthatthey follow the precedentof a coordinate
branch of government. I contend that the discretionarydeference principle
implicit in the executive precedentmodel fits betterwithin the constitutional
andlegal frameworkthatgave rise to the modernadministrativestate,provides
a unifying theory explaining why the traditionalfactorssupposedlybanished
by Chevron are relevant and how they relate to ordinarymodes of judicial
interpretation,and offers better incentives than does Chevronfor principled
decisionmakingby courts, agencies, and Congress.
I. CHEVRON:THE REVOLUTIONON PAPER
The attitudeof courtstowardadministrativeinterpretationsof statuteshas
ranged between two extremes.6At one pole, courts ignore the administrative
view. When operatingin this "independentjudgment"mode, a courtemploys
traditionaltools of statutoryinterpretation-analysisof text, legislativehistory,
and various canons of construction-to arrive at what it regardsas the best
interpretationof the statute,just as it would in a case where the executive
branchhas not spoken previously.At the other pole, courts frame the inquiry
in terms of whetherthe administrativeinterpretationis one that a reasonable
interpretermight embrace. In this "deference"mode, a court implicitly acknowledges thatthe statuteis susceptibleto multiplereadings.The task of the
courtis viewed not as discoveringthe best interpretation,butratheras assuring
that the executive view does not contradictthe statuteand otherwise furthers
legitimate objectives.
Chevronis widely understoodto marka significanttransformationin the
SupremeCourt'sjurisprudenceof deference.This part spells out the support
6. See PittstonStevedoringCorp. v. Dellaventura,544 F.2d35, 49 (2d Cir. 1976) (Friendly,J.) (noting
two extremes in judicial attitudetowardadministrativestatutoryinterpretation),affd sub nom. Northeast
Marine Terminal Co. v. Caputo, 432 U.S. 249 (1977); 5 KENNETHC. DAVIS,ADMINISTRATIVE
LAW
TREATISE
403 (2d ed. 1984) (same);NathanielL. Nathanson,AdministrativeDiscretionin the Interpretation
of Statutes, 3 VAND.L. REV.470 (1950) (distinguishingbetween occasions when courts requireagency
interpretationsto be "correct"and when they requiresuch interpretationsto be merelyreasonable).In recent
literature,the standardterms for the two poles, which I will use here, are "independentjudgment"and
"deference."See, e.g., CynthiaR. Farina,StatutoryInterpretationand the Balance of Power in the Administrative State, 89 COLUM.L. REV.452, 453-54 & n.10 (1989); Henry P. Monaghan, Marburyand the
AdministrativeState, 83 COLUM.L. REV. 1, 6-7 (1983).
972
The Yale Law Journal
[Vol. 101: 969
for this view by brieflyexaminingthe pre-Chevrondecisionallaw andcontrasting it with the frameworkset forthin Chevron.Afterdescribingthe revolutionary potentialof Chevron,I will turnin the next partto a considerationof how
Chevronhas in fact affected the practiceof the SupremeCourt.
A. Pre-Chevron:The MultipleFactors Regime
Prior to 1984, the SupremeCourthad no unifying theory for determining
when to defer to agency interpretationsof statutes.7The approachwas instead
pragmaticandcontextual.One featureof the Court'spracticewas thatdeference
could range over a spectrumfrom "great"to "some"to "little"8(althoughno
attempt was ever made to calibratedifferentdegrees of deference with any
precision). A particularlycommon approachwas to cite the views of those
chargedwith administrationof the statuteas one of severalreasonsfor adopting
a particularconstruction.Thus,the Courtmightembracea particularinterpretation (1) becauseit was supportedby the languageof the text, (2) becauseit was
consistentwith the legislative history,and (3) becauseit was the longstanding
constructionof the administrativeagency.9To be sure, there were also decisions at the polar extremes duringthis era-either ignoring the agency view
or treatingit as virtuallydispositive.?1But in practice,deferenceexisted along
a sliding scale, bridgingthese outer limits.
In addition, in deciding what degree of deference to give an executive
interpretation,the Courtrelied on an eclectic cluster of considerations.1Althoughthere was no explicit rationalelinkingthe various factorstogether,the
overallapproachhad an implicitlogic. The defaultrule was one of independent
judicialjudgment.Deferenceto the agency interpretationwas appropriateonly
if a courtcould identify some factoror factorsthatwould supplyan affirmative
justification for giving special weight to the agency views. Admittedly,the
factors tended to be invoked unevenly. But in this respect, they probably
7. See RobertA. Anthony,WhichAgencyInterpretationsShouldBindCitizensand the Courts?,7 YALE
J. ONREG.1, 6 (1990); Monaghan,supra note 6, at 31.
8. See Battertonv. Francis,432 U.S. 416, 425 n.9 (1977); 5 DAVIs, supra note 6, at 400.
9. For examples of this approach,see Bell v. New Jersey,461 U.S. 773 (1983); Blum v. Bacon, 457
U.S. 132 (1982); United States v. Clark, 454 U.S. 555 (1982); Federal Election Comm'n v. Democratic
SenatorialCampaignComm'n, 454 U.S. 27 (1981).
10. CompareNewport News Shipbuilding& Dry Dock Co. v. EEOC, 462 U.S. 669 (1983) (Court
discusses EEOCguidelineswith no suggestionthatthey areentitledto deference)with Hecklerv. Campbell,
461 U.S. 458, 466 (1983) (Court states that its review is limited to determining whether Secretary's
regulationsare arbitraryor capricious).
11. For useful accounts of the multiple factors employed during this period, see Colin S. Diver,
StatutoryInterpretationin the AdministrativeState, 133 U. PA. L. REV.549, 562 n.95 (1985); ErnestH.
Schopler,Annotation,SupremeCourt'sViewas to Weightand Effect to be Given, on SubsequentJudicial
Construction,to PriorAdministrativeConstructionof Statute,39 L.R.A.2d942 (1975); DavidR. Woodward
& Ronald M. Levin, In Defense of Deference:Judicial Review of AgencyAction, 31 ADMIN.L. REV.329,
332-41 (1979).
1992]
Executive Precedent
973
functionedin a mannernot too differentfromthe way the canons of interpretation operate in statutoryinterpretationcases.12
The pre-Chevrondeferencefactorsmay be classified in various ways. For
presentpurposes,I will groupthem into threecategories:(1) factorsaddressed
to Congress' interpretativeintent(thatis, whetherCongressintendedcourtsto
defer to an agency's interpretationof a statutoryprovision);(2) factors addressed to the attributesof the particularagency decision at issue; and (3)
factorsthoughtto demonstratecongruencebetweenthe outcomereachedby the
agency and congressionalintent regardingthat specific issue.
The first factor focused on Congress' probableinterpretativeintent. The
important distinction was between "legislative rules" and "interpretative
rules."13Legislativerules were the productof a specific delegationof authority from Congress to an administrativeagency to interpreta specific statutory
termor fill in a statutorygap. Interpretative
ruleswereexecutiveinterpretations
not backedby this type of specific delegatedauthority.The SupremeCourton
several occasions suggested that interpretationsin the formercategory were
entitled to great deference, but those falling within the latter category were
entitled only to whateverpersuasiveeffect they might have.'4
A second groupof factorsfocusednot on the agency's authority,butrather
on various attributesof its decision. One factor was whether the issue fell
within an area of agency "expertise."15
The idea was that courts are generalwhereas
are
ists,
agencies
specialists.Specialistsusuallyhave a bettergraspof
technical terms'6or the practicalconsequencesof a decision,'7and thus their
views should be given deferenceby generalists.Anotherimportantfactorwas
the notionthat"longstanding,"
"consistent,"or "uniform"administrativeinterpretations(the terms were used more or less interchangeably)are entitled to
special deference.'8A third factor in this category was that interpretations
12. The dominant view for many years, following Llewellyn, was that the canons are mutually
contradictory,and hence are of little or no value in guidingjudicial decisionmaking.See KarlN. Llewellyn,
Remarkson the Theory of Appellate Decision and the Rules or Canons About How StatutesAre to Be
Construed, 3 VAND.L. REV. 395, 401-06 (1950). More recently, a revisionist trend has set in, and
commentatorshave begun to take a more sympatheticattitudetowardthe canons. See CASSR. SUNSTEIN,
AFTER THE RIGHTSREVOLUTION:RECONCEIVING
THE REGULATORY
STATE 111-92 (1990); William N.
Eskridge,Jr.,Public Valuesin StatutoryInterpretation,137 U. PA. L. REV.1007 (1989); Geoffrey P. Miller,
Pragmatics and the Maximsof Interpretation,1990 WIS. L. REV.1179.
13. See 5 DAVIS, supra note 6, at 421-25; LOUIS JAFFE,JUDICIALCONTROLOF ADMINISTRATIVE
ACTION
564-65 (1965); JAMESM. LANDIS,THEADMINISTRATIVE
PROCESS
146-52 (1938).
14. See, e.g., Heckler v. Campbell,461 U.S. at 466-68 & n.10 (deferringto rule promulgatedunder
grantof generalrulemakingpower);Herwegv. Ray,455 U.S. 265,274-75 (1982) (deferringto interpretation
with "legislative effect"); Batterton v. Francis, 432 U.S. 416, 424-26 & nn.8-9 (1977) (distinguishing
delegated "legislative power"from nondelegated"interpretativepower"and accordinggreat deferenceto
agency interpretationsusing former).
15. See, e.g., AluminumCo. of Am. v. CentralLincolnPeoples' Util. Dist., 467 U.S. 380, 390 (1984).
16. See, e.g., E.I. du Pont de Nemours & Co. v. Train,430 U.S. 112, 134-35 & n.25 (1977).
17. See, e.g., NLRB v. Seven-Up Bottling Co., 344 U.S. 344, 348 (1953); Woodward& Levin, supra
note 11, at 332.
18. See United States v. Clark, 454 U.S. 555, 565 (1982); Haig v. Agee, 453 U.S. 280, 291 (1981);
NLRB v. Bell AerospaceCo., 416 U.S. 267, 275 (1974); Udall v. Tallman,380 U.S. 1, 16 (1965). Compare
974
The Yale Law Journal
[Vol. 101: 969
supportedby a reasonedanalysis were entitledto deference.The most prominent statementto this effect is found in Skidmorev. Swift & Co.,19where the
Courtstatedthatthe weight to be given to an agencyinterpretationwill depend
upon "the thoroughnessevident in its consideration,the validity of its reasoning, its consistencywith earlierandlaterpronouncements,and all those factors
which give it power to persuade, if lacking power to control."20A final
decision-relatedfactor,encounteredless often, was whethermultipleagencies
agreed or disagreed about the correctinterpretationof the statute.21
A third set of factors was designed to measurethe degree to which the
specific outcome reached by an agency was likely to reflect the intent of
Congress.One old idea was thatan executive interpretationis entitledto extra
weight "whenit involves a contemporaneousconstructionof a statuteby the
men charged with the responsibilityof setting its machineryin motion."22
Contemporaneousinterpretationswere thoughtto be especially probativeof
hadthemselvesparticipatcongressionalintent,eitherbecausethe administrators
ed in the draftingprocess23or because such an interpretationwas "itself evidence of assumptions-perhaps unspoken by either the administratorsor
Congress-brought to a regulatoryproblemby all involved in its solution."4
In addition, there was the recurrentnotion that executive interpretationsare
entitled to special deference if they have been ratified in some fashion by
Congress. The notion of what would count as a ratificationwas never very
precise. The paradigmsituationwas when Congress,after being informedof
an agency's construction,reenacteda statutewithout any relevantmodification.25
Standing alone, these factors did not comprise, either individually or
collectively, what could be describedas a coherentdoctrine.No attemptwas
made to connect the various factorstogetheror to explain their relevancein
terms of a model of executive-judicialrelationship.Indeed,my own attempt
to organizethem in functionalcategoriesmay impose a greatersense of order
than the cases themselves warrant.Moreover,there is little evidence that the
factorshad much predictiveor constrainingpower.To take but one example,
Nashville Gas Co. v. Satty,434 U.S. 136, 142 n.4 (1977) (consistentlymaintainedEEOCGuidelinegiven
significant weight by Court) with General Elec. Co. v. Gilbert, 429 U.S. 125, 142-43 (1977) (EEOC
Guideline that conflicts with earlier agency interpretationentitled to little weight).
19. 323 U.S. 134 (1944).
20. Id. at 140; see also Adamo WreckingCo. v. United States, 434 U.S. 275, 287 n.5 (1978) (quoting
Skidmore,323 U.S. at 140); SEC v. Sloan, 436 U.S. 103, 117-18 (1978); InvestmentCo. Inst. v. Camp,
401 U.S. 617, 626-27 (1971).
21. See North Haven Bd. of Educ. v. Bell, 456 U.S. 512 (1982); GeneralElec. Co. v. Gilbert,429 U.S.
at 144-45; see also BERNARD
ADMINISTRATIVE
LAW664-66 (1984) (discussing a case where
SCHWARTZ,
federal agencies adoptedconflicting positions on whethera particularsubstancewas "oleomargarine").
22. Norwegian Nitrogen Co. v. United States, 288 U.S. 294, 315 (1933).
23. See United States v. Moore, 95 U.S. 760, 763 (1877).
24. SEC v. Sloan, 436 U.S. at 126 (Brennan,J., concurring).
25. See, e.g., NLRB v. HendricksCountyRuralElec. Corp., 454 U.S. 170, 177 (1981); NLRB v. Bell
AerospaceCo., 416 U.S. 267, 275 (1974); Red Lion BroadcastingCo. v. FCC, 395 U.S. 367, 381 (1969).
1992]
Executive Precedent
975
in SEC v. Sloan26the Securities and Exchange Commission argued that its
interpretationnot only was entitledto deferencebecausethe interpretationwas
longstandingand consistentbut also becauseit had been ratified by Congress
when its constructionwas cited with approvalin a committeereportat the time
the statute was reenacted.27The Courtrejected these arguments,noting that
the interpretationwas not supportedby a careful analysis of the statutory
language and that evidence of congressionalratificationwas not enough if
"based only upon a few isolated statementsin the thousands of pages of
legislative documents."28
As Sloan suggests, applicationof the variousfactorsin individualcases is
manipulable.Still, it wouldbe presumptuousto dismissthemas emptyrhetoric.
Some factors-such as the importance of longstanding and consistent or
contemporaneousadministrativeconstructions-have been invokedas reasons
for deferring to executive interpretationsfor over 150 years.29Given the
durabilityof these factors,it is plausibleto view them as reflectingdeep-seated
judicial intuitionsabout the kinds of considerationsthat ought to bear on the
decision to defer. If they do not determinethe outcome of cases with logical
certainty,neitherdoes any other "traditionaltool" of statutoryinterpretation.
At least the factors turned the attentionsof courts and litigants-including
administrators-toward relevant considerationsthat presumablyshape the
judicial response.
B. The ChevronFramework
Chevron was an unlikely candidateto produce a landmarkdecision on
deference to executive interpretationsof statutes. The governmenthad not
sought a reconsiderationof the established approachto judicial review of
administrativeinterpretations,30
and only six Justicesparticipatedin the decision. Moreover,the disputedissue could be seen as partof the generaldereguInChevron,the Environmental
latorythrustof the earlyReaganAdministration.
ProtectionAgency (EPA)haddefinedthe term"stationarysource"in the Clean
Air Act to permit owners of pollutingfacilities to treatall emitting devices as
if they were undera single "bubble,"therebyminimizingthe costs of comply26. 436 U.S. 103 (1978).
27. Id. at 117-20.
28. Id. at 121.
29. See, e.g., Brownv. United States, 113 U.S. 568,570-71 (1885) (longstandingandcontemporaneous
construction);United States v. Moore, 95 U.S. 760, 763 (1877) (contemporaneousconstruction);Edward's
Lessee v. Darby, 25 U.S. (12 Wheat.) 206, 210 (1827) (contemporaneousconstruction);United States v.
Vowell, 9 U.S. (5 Cranch) 368, 371 (1809) (longstandingconstruction);see also Annotation,Effect of
Practical or AdministrativeConstructionof a Statuteon SubsequentJudicial Construction,73 L. Ed. 322
(1929) (citing hundredsof state and federal cases discussing deference to executive views).
30. The United States arguedthat the "normalrule"of deferenceshould apply, and that the court of
appeals was "fundamentallyat odds with the most basic principlesof administrativelaw." See Brief for
the National Resources Defense Council at 19, Chevron(No. 82-1005).
976
The Yale Law Journal
[Vol. 101: 969
Just the previous year, in Motor Vehicles
ing with the emissions standard.31
ManufacturersAss'n v. State Farm Mutual,32the Court had reacted with
skepticismto a similarderegulatoryinitiative;thus, it seemed unlikelythatthe
Court would write a broad prodeferenceopinion in response to the EPA's
"bubblepolicy."
Indeed, there is reason to believe that the participatingJustices did not
regardChevronas a departurefrom priorlaw. Justice Stevens' opinion for the
Court generated no concurring or dissenting statementsprotesting or even
remarkingon his approach.And in the year following Chevron,the Court
decided nineteen cases involving deference issues, but applied the Chevron
frameworkonly once.33In time, however, lower courts, agencies, and commentatorsall came to regardthe analysis of the deferencequestion set forth
in Chevronas fundamentallydifferentfrom that of the previous era. Justice
Stevens' opinion contained several features that can only be described as
even if no revolutionwas intendedat the time.
"revolutionary,"34
Chevron'sfirst importantinnovationwas the prescriptionof a procedural
formulafor courtsto follow in determiningwhetherto deferto agencyinterpretations. Each case, the Courtsuggested, shouldproceedin two steps. At step
one, the court would operatein the independentjudgmentmode. It would ask,
using traditionaltools of statutoryconstruction,whetherCongresshad"directly
spoken to the precise question at issue."35If the court concluded that Congress
had a "specific intention"36with respect to the issue at hand, it would adopt
and enforce that answer.But if the court failed to uncoverany such intention,
it would move on to step two, where it would shift into the deferencemode.
Here, the question would be whetherthe agency's position was "a reasonable
of the statute.
interpretation"37
On its face, the two-step formulaseems innocuousenough. Indeed,after
the formlessness of the previous era, it offers the beguiling promise of an
orderlymethodfor resolvinga wide varietyof controversies.Whatwas unclear
at the time, however, was that the two-step inquiry as framed by Chevron
31. 467 U.S. at 840.
32. 463 U.S. 29 (1983).
33. See Chemical Mfrs. Ass'n v. National ResourcesDefense Council, 470 U.S. 116 (1985); see also
Table 1, infra. JusticeStevens' behaviorseems to confirm this point. Earlyin the nextTerm,JusticeStevens
authoredan opinion that was much more consistent with pre-Chevroncase law than with the Chevron
doctrine. ConnecticutDep't of Income Maintenancev. Heckler,471 U.S. 524 (1985). And less than three
weeks before Chevronwas decided, JusticeStevens authoreda dissentingopinion statingthat no deference
was due to an administrativeinterpretationbecause it "has not been a model of consistency."Aluminum
Co. of Am. v. CentralLincolnPeople's Util. Dist., 467 U.S. 380,402-03 n.3 (1984) (Stevens, J., dissenting).
These opinions tend to suggest that Justice Stevens himself regardedChevronas a restatementof existing
law ratherthan a new approach.
34. KennethW. Starr,Judicial Review in the Post-ChevronEra, 3 YALEJ. ONREG.283, 284 (1986).
As Judge Starrcorrectlynotes: "Thisrevolutionaryeffect is not apparentfrom a quick examinationof the
opinion itself." Id. at 284.
35. Chevron,467 U.S. at 842.
36. Id. at 845.
37. Id. at 844.
19921
Executive Precedent
977
would have profoundconsequencesfor the way in which courts approachthe
deference question. There are several reasons for reachingthis conclusion.
First, in contrastto the previous approach,the two-step structuremakes
deferencean all-or-nothingmatter.If the courtresolvesthe questionat step one,
then it exercises purelyindependentjudgmentandgives no considerationto the
executive view. If it resolves the questionat step two, then it appliesa standard
of maximumdeference.In effect, Chevrontransformeda regime that allowed
courts to give agencies deferencealong a sliding scale into a regime with an
on/off switch.
Second, the Court'snew frameworkinvertedthe traditionaldefaultrule.
In the pre-Chevronperiod, deference to executive interpretationsrequired
specialjustification;independentjudgmentwas the defaultrule.UnderChevron,
the court must initially establishwhetherthe issue is suitablefor independent
judicial resolution;if it is not, the court automaticallyshifts into a deferential
mode. As a result,independentjudgmentnow requiresspecialjustification,and
deference is the default rule. If, as the Courtseemed to suggest, the circumstancesjustifying independentjudgmentwere definednarrowly,this inversion
portendeda majortransferof interpretativepower from courts to agencies.
Third,the two-step frameworkhas importantimplicationsfor the kinds of
considerationsthatenterintojudicialdecisionmaking.In particular,the framework appearsto exclude any examinationof the multiple factors historically
relied upon by courts. The most immediatebasis for this conclusion is the
Court's decision to defer to the EPA'sdefinition of "stationarysource"even
thoughit arguablyrepresenteda "sharpbreakwith priorinterpretationsof the
Act."38Thus, the decision appearedto downgradethe frequentlycited factor
stressing the importanceof agency views that were longstandingand consistent.39More importantly,none of the traditionalfactorsfits understep one or
step two of the new framework.They are clearly irrelevantunderstep one,
which focuses entirelyon whathappenedin Congress,not on the agency or its
decision. The factors could conceivably enter into the calculus at step two,
where the court asks if the executive interpretationis "reasonable."But by
"reasonable,"the Courtseemed to meanreasonablein light of the text, history,
and interpretativeconventionsthat govern the interpretationof a statuteby a
court;at least, this was the way the Courtconductedthe reasonablenessinquiry
in Chevron40and subsequentcases.41The question whetheran interpretation
38. Id. at 862.
39. See Antonin Scalia, Judicial Deference to AdministrativeInterpretationsof Law, 1989 DUKEL.J.
511, 517 (under Chevron, "there is no longer any justification for giving 'special' deference to 'longstanding and consistent' agency interpretationsof law"). On the other hand, Chevronitself made passing
reference to one traditionalfactor-agency expertise. See Chevron,467 U.S. at 865.
40. 467 U.S. at 843-45.
41. See, e.g., Pauley v. BethenergyMines, Inc., 111 S. Ct. 2524, 2534-35 (1991); Rust v. Sullivan,
111 S. Ct. 1759, 1769 (1991); United States v. RiversideBayview Homes, Inc., 474 U.S. 121, 131 (1985).
Two decisions from last Termillustratethe problemof whatto do with the traditionalfactorsunderChevron.
In Pauley, 111 S. Ct. 2524, the Courtconsideredtwo traditionalfactors (expertise and longstandingand
978
The Yale Law Journal
[Vol. 101: 969
is reasonablein light of these traditionalnormsof judicial interpretationlikewise provides no place for the various contextualfactorsthat played such an
importantrole in the pre-Chevronera.
In addition to its novel framework,Chevronalso broke new ground by
invoking democratic theory as a basis for requiringdeference to executive
interpretations.Congress, Justice Stevens reasoned,is the ultimate source of
lawmakingauthorityin a democracy."If the intent of Congressis clear, that
is the end of the matter;for the court, as well as the agency, must give effect
to the unambiguouslyexpressedintentof Congress."42
If the intentionsof the
not
are
then
we
are
clear,
however,
primarylawgiver
presentedwith a choice:
who should undertaketo fill in the gap in the understandingof congressional
will, the courtor the agency?Chevrondeclaredthatthe agency is the preferred
gap filler. Judges "are not part of either political branch";they "have no
An agency,on the otherhand,while "notdirectlyaccountable
constituency."43
to the people,"is subjectto the generaloversightand supervisionof the President, who is accountable.Thus, it is fitting that agencies, ratherthan courts,
resolve "thecompetinginterestswhichCongressitself eitherinadvertentlydid
not resolve, or intentionallyleft to be resolvedby the agency chargedwith the
administrationof the statutein light of everydayrealities."44
This new emphasis on democratictheory was importantto the doctrinal
frameworkbecause it suppliedthe justificationfor switching the defaultrule
from independentjudgmentto deference.Underthe pre-Chevronregime, not
every agency decision wouldqualify for deferencein the face of an ambiguous
statute. In order to make deference a general default rule, the Court had to
come up with some universalreasonwhy administrativeinterpretationsshould
be preferredto the judgmentsof Article IIIcourts.Democratictheorysupplied
thejustification:agencydecisionmakingis alwaysmoredemocraticthanjudicial
consistent interpretation)as part of the discussion of whetheror not to defer. In Rust, 111 S. Ct. 1759, on
the other hand, the Court appendeda discussion of traditionalfactors (lack of consistency in the agency
position andthe well-reasonednatureof the decision)to its discussionunderstep two. These recentdecisions
indicate that the Courtcontinuesto believe the traditionalfactorsare relevant,but has no idea what to do
with them under the Chevronframework.
42. 467 U.S. at 842-43.
43. Id. at 865-66.
44. Id. For a recent elaborationof the democratictheory foundationsof Chevron,see LaurenceH.
Silberman, Chevron-The Intersectionof Law & Policy, 58 GEO.WASH.L. REV.821, 822-24 (1990).
Chevron'sdemocratictheory thesis appears to presupposea unitary executive, i.e., an interpretationof
separationof powers that would place all entities engaged in the execution of the law-including the socalled independentregulatoryagencies-under Presidentialcontrol.See Steven Calabresi& Kevin Rhodes,
The StructuralConstitution:UnitaryExecutive,Plural Judiciary, 105 HARV.L. REV.(forthcomingApr.
1992);Geoffrey Miller,IndependentAgencies, 1986 SUP.CT.REV.41. Agencies arenot directlyaccountable
to the people; they can invoke a popularmandateonly if they are accountableto the elected President.
Althoughthe unitaryexecutive constructionwas dealt a setbackin Morrisonv. Olson, 487 U.S. 654 (1988)
(permittingCongress to insulate independentcounsel from removalexcept for good cause), it is far from
clear thatthe matterhas been settled conclusively.See Toubyv. United States, 111S. Ct. 1752, 1757 (1991)
(noting that "[t]he Constitutionvests all executive power in the President,U.S. Const., Art. II, ? 1").
1992]
Executive Precedent
979
decisionmakingbecause all agencies are accountable(to some degree) to the
President,and the Presidentis elected by the people.
Chevron'sheavy reliance on democratictheory left one majorproblem.
Congresspresumablyrankshigheron the democracyscale thando the agencies.
After all, Congress is directly elected by the people; agencies are at most
indirectlyaccountableto the electoratethroughthe President.Yet how do we
know thatCongress,the ultimatedemocratictrumpcard,wantsambiguitiesand
gaps to be resolved by agencies ratherthan by courts?
Chevronsolved this potentialquandaryby adoptingits thirdand probably
most controversialinnovation: a presumptionthat whenever Congress has
delegatedauthorityto an agency to administera statute,it has also delegated
authority to the agency to interpret any ambiguities present in the statute.45
Previous cases, as we have seen, suggested that deferencewas appropriateif
Congress had expressly delegatedto an executive agency the power to define
a particularterm. Chevronin effect adopteda fiction thatassimilatedall cases
involving statutoryambiguitiesor gaps into the expressdelegationor "legislative rule" model.46
In short, the Chevron opinion can be understoodas a prescriptionfor
resolvingquestionsof statutoryinterpretationthrougha series of presumptions
about primary and delegated lawmaking. Congress is presumed to be the
primarylawmakinginstitution,and whereit has spoken,its will must prevail.
But when Congress has not addressedthe precise issue in dispute and has
delegatedauthorityto an administrativeagency,thenwe presumethatinterpretative authorityhas been delegatedto the agency.Onlyif Congressis silentand
has failed to designatean agency to administerthe statutedo we presumethat
power to interpretambiguouslaw has been delegatedto an Article III court.
Given the rhetoricof JusticeStevens' opinion,the most apparentobjective
of this series of presumptionswas to maximize the role of democratically
accountableinstitutionsin the processof legal interpretationand to restrictthe
discretionof unelectedcourts.Conceivably,the formalismof Chevronmay also
45. In Chevron,the Court stated the presumptionin terms of "implicit"delegations:
'The power of an administrativeagency to administera congressionallycreated ... program
necessarily requires the formulationof policy and the making of rules to fill any gap left,
implicitly or explicitly, by Congress.' If Congresshas explicitly left a gap for the agency to fill,
there is an express delegation of authorityto the agency to elucidate a specific provisionof the
statute by regulation.... Sometimes the legislative delegation to an agency on a particular
question is implicit ratherthan explicit.
467 U.S. at 843-44 (quoting Mortonv. Ruiz, 415 U.S. 199, 231 (1974)).
Recent decisions have reaffirmedthat the generalduty to defer identifiedin Chevronis groundedin
a theory of presumeddelegation. See Pauley, 111 S. Ct. at 2534; Martinv. OccupationalSafety & Health
Review Comm., 111S. Ct. 1171, 1175 (1991) (discussingdeferenceto an agency's interpretationof its own
regulations);Adams FruitCo. v. Barrett, 110 S. Ct. 1384 (1990). For an especially clear recognitionthat
Chevronrests on a presumptionthatambiguitiesentaila delegationof interpretativepower,see Scalia,supra
note 39, at 516-17.
46. See Kevin W. Saunders,InterpretativeRules with LegislativeEffect:An Analysis and a Proposal
for Public Participation, 1986 DUKEL.J. 346, 357 (Chevwron
appearsto eliminate the distinctionbetween
legislative rules and intepretativerules).
980
The Yale Law Journal
[Vol. 101: 969
have reflected a desire to streamlinethe deference inquiry by reducing the
multiple variablesof the precedingera to two basic inquiries.In any event, in
the clarity of its proceduralformula,and the implicationsof that formulafor
the balance of power among the branches,Chevrondoes indeed appearto
presage a "revolutionary"transformationof the old order.
II. CHEVRON:
THEREVOLUTION
IN PRACTICE
Each year the SupremeCourtdecides somewherebetweenten and twenty
cases in which it confrontsan issue aboutwhetherto deferto an administrative
interpretationof a statute.This steadystreamof deferencecases has abatedonly
slightly since Chevronwas decided.As a result,thereis now an extensivebody
of decisionallaw within the Courtthatcan be used to gauge Chevron'simpact.
Surprisingly,no one has undertakena comprehensivesurvey of these decisions.47To the contrary,the commentaryon Chevrontends to assume-almost
always based on only a handfulof the Court'spost-Chevrondecisions-either
that the Court has faithfully followed Chevronin succeeding years48or that
the Courthas essentiallydisregardedthe radicalimplicationsof the decision.49
A surveyof the full rangeof the Court'sdecisionsbeforeandafterChevron
discloses a more complex and indeed ratherbewilderingpicture.On the one
hand, Chevron clearly has resulted in a significant shift in the deference
doctrine.On the otherhand,Chevronhas notproducedanythinglike a complete
revolution in the Court's jurisprudence.On the whole, the overall picture
suggests that the judicial understandingthatinformsthe deferencequestionis
probablymore confused today than it has ever been.
A. The IncompleteRevolution
One way to measurethe impactof Chevronis to examine the Court'sown
post-Chevrondeference cases in the aggregate. In order to make such an
assessment,I reviewedall SupremeCourtdecisionsfrom the 1981 Termto the
end of the 1990 Term.The resultsfor the seven post-ChevronTerms(Chevron
was decided near the end of the 1983 Term)are summarizedin Table 1.50
47. The most complete account of post-Chevrondevelopments in the Supreme Court is found in
Sunstein, supra note 4, at 2091-104.
48. See, e.g., RichardJ. Pierce, Ji., Chevronand its Aftermath:Judicial Review of AgencyInterpretations of StatutoryProvisions, 41 VAND.L. REV.301, 302 (1988); Starr,supra note 34, at 291.
49. See, e.g., Linda R. Hirschman,PostmodernJurisprudenceand the Problem of Administrative
Discretion, 82 Nw. U. L. REV.646, 703 (1988).
50. The cases thatcomprisethe dataare summarizedin the tables,infrapp. 981-82, 992, and are listed
in the Appendix, infra.
Executive Precedent
1992]
981
TABLE1. Post-Chevron Terms
A
B
C
E
F
CHEVRON CHEVRON
TOTAL
CASES
TERM
D
INVOLVING
CHEVRON FRAMEWORK:FRAMEWORK:CASESCIING
DEFERENCE AGENCYVIEW FRAMEWORKAGENCYVIEW DECIDEDAT TRADITIONAL
APPLIED
STEPTWO
FACTORS
ACCEPTED
ACCETED
QUESnON
1990
11
1989
14
8 (73%)
8 (57%)
6 (55%)
9 (62%)
3 (33%)
1988
9
1987
14
4 (44%)
9 (64%)
1986
9
5 (55%)
5 (36%)
2 (22%)
1985
14
1984
19
11(78%)
18 (94%)
6 (43%)
1 (5%)
TOTAL1984-90
90
63 (70%)
32 (36%)
4 (66%)
4 (44%)
2 (33%)
2 (25%)
1 (33%)
3 (60%)
1 (50%)
1 (33%)
3 (50%)
1 (50%)
5 (83%)
1 (100%)
4 (66%)
1 (100%)
19 (59%)
14 (44%)
6 (55%)
4 (29%)
6 (66%)
3 (21%)
2 (22%)
6 (43%)
7 (35%)
34 (37%)
Column A reportsthe total numberof cases each Termin which at least
one Justicerecognizedthe presenceof a question(concerningeithera primary
or a subsidiaryissue) about whether to give deference to an administrative
interpretationof a statute.5'ColumnB indicateshow many of these decisions
Column C shows how many of the
accepted the executive interpretation.52
total cases appliedthe two-stepChevronframeworkor its equivalent;53
column
51. Adopting a principlefor includingcases in the survey posed some problems.On the one hand,it
is commonly perceived that there are many cases in which the Court substitutesits judgmentfor that of
the agency and fails even to mentionthe possibilityof deference.See 5 DAVIS,supra note 6, at 403. Ideally,
these cases would be included in the sample in orderto determinethe true incidence of deference.On the
other hand, it would be very difficult to determineall the cases in which the Court could have deferred.
I would have had to read the briefs and records in all cases involving a question of federal statutory
interpretation.Moreover,determiningthe full set of potentialdeferencecases based on some independent
standardwould pose difficult conceptual problems in defining what is a potential deference case. For
instance, would a decision count as a potential deferencecase if it involved an interpretationadoptedfor
the first time in briefs filed in court defending an agency action? Would a decision count as a potential
deference case if it involved an interpretationof a provisionthat the agency is not directly empoweredto
regulate? Given these competing considerations,I compromisedand adopted a selection principle that
included all cases in which at least one Justice mentioned the possibility of deferring to an agency
interpretation.This picks up some of the cases wherethe majoritysubstitutesjudgmentanddoes not mention
deference, but a dissenting or concurringJusticeraises the issue. Thus, it producesa slightly largersample
than would an approachthat looked solely to majorityopinions. But clearly the one-Justiceprincipledoes
not include all potential deference cases (for example, it skips cases where the Court is unanimousand
substitutes its judgment). Consequently,the sample still undercountsthe total number of cases where
deference could have been invoked.
Anotherproblemwas what to do with cases thatpresentslight variationson the questionof deference
to agency interpretationsof statutes,such as cases involvingan agency interpretationof its own regulations,
agency interpretationsof a treaty,or challenges to agency policies on the ground that they were adopted
in an arbitraryand capricious manner(as opposed to being in conflict with a statute).These cases (which
were relatively few in numbercomparedto the statutoryinterpretationcases) were excluded in the interest
of minimizing complexity.
52. "Acceptingthe agency interpretation"is not the same as ruling in favor of the government.In a
few cases the Courtruled against the governmenteven though it acceptedthe agency interpretation.This
might happen,for example, if the SolicitorGeneraldisclaimedthe agency view, but the Courtagreedwith
it. See, e.g., Bob Jones Univ. v. United States, 461 U.S. 574 (1983).
53. I adopted a very broad test for determining whether the Court was "applying"the Chevron
framework.I includedin this category not just the cases thatquoted the two-step procedure,or organized
the discussion aroundthe two-step procedure,but also those that cited Chevronor post-Chevrondecisions,
The Yale Law Journal
982
[Vol. 101: 969
D reportshow many of the cases following the Chevronframeworkaccepted
the executiveinterpretation;
columnE revealsthe numberof Chevroncases that
consideration
to
the
some
reasonablenessof the executive view (step two
gave
of the framework) as opposed to stopping at step one. Finally, column F
indicates how many controllingopinions in the cases rely on one or more of
the traditionalfactorsappliedin the pre-Chevronperiod, such as longstanding
and consistent interpretation,contemporaneousinterpretation,congressional
ratification,and so forth.54In orderto providea basis for comparison,Table
2 reproducesdata for some of these variables (total deference cases, cases
accepting the administrativeinterpretation,and cases applying traditional
factors) for the three pre-ChevronTerms, 1981-83.
TABLE2. Pre-Chevron Terms
A
TERM
B
F
TOTALCASESINVOLVING
CASESCITING
DEFERENCE
FACTORS
QUESTION AGENCYVIEWACCEPTED TRADITIONAL
1983
19
13 (68%)
11 (57%)
1982
15
11 (73%)
11 (73%)
1981
11
10 (90%)
8 (73%)
TOTAL1981-83
45
34 (75%)
30 (66%)
A numberof interestingconclusions can be derived from the figures in
Tables 1 and 2. First, it is clear that Chevronis often ignoredby the Supreme
Court.Althoughthe Chevronopinionpurportsto describea universalstandard
by which to determinewhetherto follow an administrativeinterpretationof a
statute,the two-step frameworkhas been used in only about one-thirdof the
total post-Chevroncases in which one or more Justices recognized that a
deferencequestion was presented.55
AlthoughChevronbegan to be used more
after
the
1987
it
is
still far from the monolithic norm the
Term,
frequently
such as K MartCorp. v. Cartier,Inc., 486 U.S. 281 (1988), in setting forth the standardof review, or that
otherwiseindicatedthatdeferencewould be owed to any agencydecisionprovidedCongresshadnot clearly
answeredthe question. Essentially,I tried to determinewhetherthe authorof the controllingopinion was
thinkingabout Chevronin setting forth the analysis of deference. Although admittedlythis is a highly
subjective standard,I believe it is more reliable than any other yardstick.If anything,it tends to overstate
the influence of Chevron.
54. Column F includes both Chevronframeworkcases and non-Chevronframeworkcases. Most of
the cases in this column either ignore or disclaim reliance on Chevronin favor of the traditionalapproach.
Some, however,purportto rely on the Chevronframework,even if this is (arguably)inconsistentwith also
relying on the traditionalfactors.
55. This undoubtedlyunderstatesthe extent to which Chevronis ignored,because cases where there
is an executive interpretation,but where the Court fails to note even the possibility of deferring to that
interpretation,were excluded from the survey.In otherwords, the most extremeinstancesof "independent
judgment"were ignored. See supra note 51.
1992]
Executive Precedent
983
opinionseems to describe.If we look solely at the 1987-90period,the two-step
frameworkwas still appliedin only half of the deferencedecisions.
Althoughsome of the failureto rely on Chevronmay be attributedto time
as
lag, awarenessof the decisionslowly disseminatedthroughthe administrative
bar,56this cannot accountfor the persistentresistanceto using the framework
in the SupremeCourt. By the end of the 1987 Term,for example, the Court
had appliedthe Chevronframeworkin fifteen differentcases in a wide variety
of areas.57Any Justice who was paying attention to the Court's own work
productshould have been thoroughlyversed in the tenets of Chevronby this
time. Moreover,even if administrativelawyersin specializedareaslike tax and
laborlaw were late in comingto an awarenessof Chevron,the federalagencies
were representedin the SupremeCourt almost exclusively by the Solicitor
General'soffice-a small groupof"generalists"who havebeenvery conscious
of Chevron.58Thus, by 1987 at the latest, the agencies could be expected to
take maximumadvantageof Chevronin the SupremeCourt.Yetin the 1988-90
Terms,the Courtcontinuedto apply the Chevronframeworkin only half the
cases presenting a deference question. The persistentspottiness of Chevron
duringthis period stronglysuggests thatthe Courtin many cases was simply
56. The surveyclearly suggests such a time lag. As previouslynoted,it took some time for the Chevron
doctrine to appearin the Court'sopinions;the 1984 Termmust for all intents and purposes be considered
partof the pre-Chevronera. Moreover,even in yearswhen Chevronis appliedwith some frequency,it tends
to be invoked less often in areas where there is a particularlyrich traditionof pre-Chevronprecedenton
deference.Forexample,in Title VII, labor,tax, social security,andenvironmentalcases, the Court(no doubt
guidedto a degree by the submissionsof the parties)still tends to framethe deferencestandardin the terms
expressedin earlierdecisions specific to these areas,ratherthanin termsof Chevron.See, e.g., CottageSav.
Ass'n v. Commissioner,111 S. Ct. 1503 (1991) (citing previous cases for deference standard;no mention
of Chevron);EEOC v. ArabianAm. Oil Co., 111 S. Ct. 1227, 1235 (1991) (citing pre-Chevroncases for
standardof deference to EEOC); Robertson v. Methow Valley Citizens Council, 490 U.S. 332 (1989)
(Council on EnvironmentalQuality'sinterpretationof NEPA assessed undertraditionalfactors,no citation
to Chevron);Atkins v. Rivera,477 U.S. 154 (1986) (review of Social SecurityAdministrationinterpretation
based on legislative/interpretativedistinctionof pre-Chevronera); MeritorSav. Bank v. Vinson, 477 U.S.
57, 65 (1986) (same);NLRB v. Action Automotive,Inc., 469 U.S. 490,494-96 (1984) (citing PackardMotor
Car Co. v. NLRB, 330 U.S. 485 (1947), and NLRB v. HearstPublications,Inc., 332 U.S. 111 (1944) (deference approachto NLRB decisions).
In contrast,in areasin which the pre-Chevroncase law was less well developed-ERISA andbanking
cases seem to fall in this category-Chevron seems to have made more headwayin the Court. See. e.g.,
Pension Benefit Guar.Corp. v. LTV Corp., 110 S. Ct. 2668, 2676 (1990) (applying Chevronstandardin
ERISA controversy);Mead Corp. v. Tilley, 490 U.S. 714 (1989) (same); Clarkev. SecuritiesIndus.Ass'n,
479 U.S. 388 (1987) (Chevronappliedto interpretationby Comptrollerof Currency);Board of Governors
v. Dimension Fin. Corp., 474 U.S. 361 (1986) (Chevronframeworkappliedto FederalReserve interpretation).
57. K Mart Corp. v. Cartier,Inc., 486 U.S. 281 (1988); EdwardJ. DeBartolo Corp. v. FloridaGulf
Coast Bldg. & Constr.TradesCouncil, 485 U.S. 568 (1988); Bethesda Hosp. Ass'n v. Bowen, 485 U.S.
399 (1988); ETSI Pipeline Projectv. Missouri, 484 U.S. 495 (1988); Honig v. Doe, 484 U.S. 305 (1988);
NLRB v. United Food & Commercial Workers Union, Local 23, 484 U.S. 112 (1987); INS v.
Cardoza-Fonseca,480 U.S. 421 (1987); Clarke v. Securities Indus. Ass'n, 479 U.S. 388 (1987); Japan
Whaling Ass'n v. American Cetacean Soc'y, 478 U.S. 221 (1986); United States Dep't of Transp. v.
ParalyzedVeteransof Am., 477 U.S. 597 (1986); Youngv. CommunityNutritionInst.,476 U.S. 974 (1986);
United States v. City of Fulton, 475 U.S. 657 (1986); Board of Governorsv. Dimension Fin. Corp., 474
U.S. 361 (1986); United States v. RiversideBayview Homes, Inc., 474 U.S. 121 (1985); Chemical Mfrs.
Ass'n v. National Resources Defense Council, 470 U.S. 116 (1985).
58. This was true by the fall of 1987, when I joined the Solicitor General'sOffice.
984
The Yale Law Journal
[Vol. 101: 969
not preparedto abide by the type of analysis dictatedby Chevron-although
apparentlyit was not preparedto abandonit or explain why it ought not to be
controllingeither.
Second,althoughChevronis generallyregardedas directingthatcourtsgive
thereis no discerniblerelationship
greaterdeferenceto executiveinterpretations,
between the applicationof the Chevronframeworkand greateracceptanceof
the executive view. Indeed,cases applyingthe Chevronapproachhave on the
whole producedfeweraffirmancesof executive interpretationsthanthose that
do not follow Chevron.59Althoughthe numberof cases is too small to attribute significance to the precisepercentages,the generalphenomenonis apparent
when we compare the rate of acceptancesin the cases actually applyingthe
Chevron frameworkin the post-Chevronperiod-59% adoptingthe agency
view-with eitherthe overallacceptanceratein the post-Chevronperiod(70%)
or the ratein the pre-Chevronera (75%).Paradoxically,it appearsthatadoption
of the Chevronframeworkhas meant, if anything,a decline in deferenceto
agency views.
The suggestion that Chevronhas had little discernibleinfluence on the
SupremeCourtcontrastssharplywith the only otherpublishedempiricalstudy
of Chevron'simpact. ProfessorsPeter Schuckand Donald Elliot undertooka
survey of Chevron'sinfluence on lower courts and found that deference to
Their findings,howagency interpretationsincreasedsharplyafterChevron.60
ever, are not necessarilyinconsistentwith my datasuggestingno such change
in the SupremeCourt. Lower courts probablytake SupremeCourt opinions
more seriously than does the Courtitself.6'Moreover,one would predictthat
as it became increasinglyevident over time that the SupremeCourtemploys
the Chevronapproachonly sporadically,lower courtswould begin to revertto
their old habits. And indeed, there is evidence in the Schuck and Elliot study
that suggests the "Chevroneffect" in the lower courts may have been only
temporary.62
Third, the tables indicate that the emergence of Chevron has caused a
decline in relianceon the traditionalcontextualfactorsfor determiningwhether
59. This conclusionis to some extent probablya reflectionof my samplingmethodology.In particular,
the non-Chevroncases may contain a disproportionatenumberof affirmances.This is because when the
Courtagrees with an agency construction,it may often throwin deferenceto the administrativeagency as
an additionalfactoror considerationin supportof its judgment.But when it disagreeswith the agency view,
it may not mention the issue of deference.On the other hand, this possibility is mitigatedby the fact that
even if the majorityignoresthe agency interpretation,a dissentingor concurringopinionmay raise the issue
of deference-and these cases are also included in the sample. Moreover,even if the rate of agency
affirmancesis overstatedin the non-Chevroncases, this would not accountfor the relativelylow percentage
of affirmances in the cases applyingthe Chevronframework,where the issue of deference is necessarily
presentedin each instance.
60. Peter H. Schuck & E. Donald Elliott, To the Chevron Station: An EmpiricalStudy of Federal
AdministrativeLaw, 1990 DUKEL.J. 984.
61. Indeed, they are requiredto do so in many circumstances.See Hicks v. Miranda,422 U.S. 332,
343-45 (1975) (lower courts bound by SupremeCourtsummaryaffirmance,even if Court itself is not).
62. Schuck & Elliott, supra note 60, at 1037.
1992]
Executive Precedent
985
deferenceis appropriate.Again, the numbersare too small to attributesignificance to precise percentages,but in the pre-Chevronperiod (1981-83 Terms)
66% of the deference cases cited one or more of these traditionalfactors.In
the post-Chevronperiod,in contrast,the percentageof cases citing one of these
factorsdeclined to 36% overall. Still, it is obvious that the Court'sinvocation
of these factorspersists, with no visible trendpointingtowardtheir complete
extinction.This is not whatone wouldexpect to find if Chevronhadcompletely
transformedthe practice of deciding when to defer.
B. The ContestedRevolution
When we turn from the aggregatedata to an examinationof the Court's
opinions, the picture that emerges again suggests that the Chevronrevolution-although real-is a tenuous one. I will not discuss the many routine
applicationsof Chevron,the even more numerouscases that ignore Chevron,
or even those cases where the majorityand dissent disagree over the proper
application of Chevronto particularcircumstances.More revealing are the
opinionsthatengagein disputesoverwhatmightbe called"Chevronfundamentals." The presence of these disputes is not necessarily surprising.There has
probably never been a watershedSupremeCourt opinion (considerBrown,
Miranda,Mapp, and Roe)63that did not generatesubsequentcontroversyover
the scope and meaning of the original decision. Whatis striking,however,is
the extent to which the Court'sresolutionsof disputesover the meaningof the
Chevron framework rest uncomfortablywith the rationale of the original
decision.
Without a doubt, the most remarkableepisode occurred in INS v.
a case involving interpretationof immigrationlaw, where
Cardoza-Fonseca,64
the Courthas traditionallyshown considerabledeferenceto executive views.65
Specifically, the issue was whetherthe statutorystandardfor awardingasylum
could be construedas requiringthe same showing of likelihoodof persecution
uponreturnto the countryof originas the standardfor withholdingof deportation. The AttorneyGeneral,actingthroughthe Bureauof ImmigrationAppeals,
determinedthatthe two provisionsrequiredthe same showing.A baremajority
of five Justices-in an opinionby JusticeStevens,the authorof Chevron-held
thatthis executive interpretationwas not entitledto deference.The first reason
was entirely consistent with the Chevronframework:"Employingtraditional
tools of statutoryconstruction,we haveconcludedthatCongressdid not intend
63. Brown v. Board of Educ., 347 U.S. 483 (1954); Mirandav. Arizona, 384 U.S. 436 (1966); Mapp
v. Ohio, 367 U.S. 643 (1961); Roe v. Wade,410 U.S. 113 (1973).
64. 480 U.S. 421 (1987).
65. See, e.g., INS v. Rios-Pineda,471 U.S. 444, 452 (1985) (admonishinglower courts not to usurp
authoritygrantedto the AttorneyGeneralin administeringthe immigrationlaws); Hamptonv. Mow Sun
Wong, 426 U.S. 88, 101-02 n.21 (1976) (noting that "power over aliens is of a political characterand
thereforesubject to only a narrowjudicial review").
986
The Yale Law Journal
[Vol. 101: 969
the two standardsto be identical."66In what appearedto be an alternative
holding, however,Justice Stevens statedthatthe Chevronstandardwas in any
event inapplicablebecause the issue for decision was a "pure question of
Deferencewould be appropriate,he suggested,only
statutoryconstruction."67
if the case concernedthe applicationof the standardto particularfacts.68
The notion that deferenceto agency views should turnon the distinction
between pure questionsof law and questionsof law applicationcan be found
in some SupremeCourtdecisions of the 1940's.69But the distinctionhad been
abandonedfor all practicalpurposeslong before Chevronwas decided.70The
suggestionthatwhatChevronhaddone was to restorethis ancientlearningwas
truly astonishing,as JusticeScalia madeclearin an acidic concurrence.In fact,
as Justice Scalia correctlynoted, the issue in Chevron,concerningthe proper
definition of the statutoryterm "source,"was obviously a "purequestion of
law," and hence would be entitled to no deferenceunderthe Court'sdistinction.71Moreover,the dispute about the propermeaning of Chevronhad not
been briefed and argued in Cardoza-Fonseca,nor was it necessary to the
decision. It was thus inappropriatefor the Court to cast doubt on such "an
extremely importantand frequentlycited opinion."72
Cardoza-Fonsecagave rise to a brief period of uncertaintyin the lower
courts over whether Chevronstill applied to pure questions of law.73By the
end of the next Term, however, the Court was again applying the Chevron
doctrine(irregularly,as ever) to questionsof law,andCardoza-Fonsecaquietly
droppedfrom sight.74But if Cardoza-Fonsecahad no lasting impact on the
66. Cardoza-Fonseca,480 U.S. at 446.
67. Id.
68. Id. at 448.
69. ComparePackardMotor Car Co. v. NLRB, 330 U.S. 485, 493 (1947) (no deference to agency
views on "nakedquestion of law") with NLRB v. HearstPublications,Inc., 322 U.S. 111, 130-31 (1944)
(great deference to NLRB's determinationof how statutorystandardappliedto particularfactual circumstances).
70. See, e.g., FederalElectionComm'nv. DemocraticSenatorialCampaignComm.,454 U.S. 27 (1981)
(deferringto agency interpretationof purequestion of law); FordMotorCreditCo. v. Milhollin,444 U.S.
555 (1980) (same); Monaghan,supra note 6, at 29-30.
71. Cardoza-Fonseca,480 U.S. at 455 (Scalia, J., concurringin judgment).
72. Id. at 454 (Scalia, J., concurringin judgment).
73. See NLRB v. FederalLaborRelationsAuth., 834 F.2d 191, 198 (D.C. Cir. 1987) (underCardozaFonseca no deferencedue to agencyinterpretationof "purequestionof law");Union of ConcernedScientists
v. United States Nuclear RegulatoryComm'n, 824 F.2d 108, 113 (D.C. Cir. 1987) (Cardoza-Fonseca
"stronglyindicate[s] that courts are to apply the Chevrontest only in circumstancesin which an agency
is requiredto apply a legal standardto a particularset of facts").But see NLRB v. FLRA, 834 F.2dat 202
(Starr, J., concurring)("I continue to believe that Chevron'stwo-step interpretativeprocess, which was
reaffirmed in Cardoza-Fonseca,provides the polestar that is to guide us."); id. at 203 (D. Ginsburg,J.,
concurring)(sharing Judge Starr's belief).
74. Early in the next term, Justice Scalia stated in a concurringopinion, NLRB v. United Food &
CommercialWorkersUnion, Local 23, 484 U.S. 112, 133-34 (1987) (Scalia, J., concurring),thatCardozaFonseca was not being followed by the Court(althoughthe majorityin that case cited both Chevronand
Cardoza-Fonseca).Shortlythereafter,the Courtrevertedto the Chevronframeworkwithoutany suggestion
thatit was confined to cases of law application.See MeadCorp. v. Tilly, 490 U.S. 714 (1989); PittstonCoal
Group v. Sebben, 488 U.S. 105 (1988); K Mart Corp v. Cartier,Inc., 486 U.S. 281 (1988).
1992]
Executive Precedent
987
Court'sunderstandingof the Chevrondoctrine,the episode neverthelesssuggests that the Court's commitmentto Chevron-at least in its unvarnished
original form-was at best fragile. The fact that a majorityof the Courtwas
prepared(if only temporarily)to recastthat doctrinein a way that would strip
it of most of its significance hardlysuggests that the Courtwas comfortable
with Chevron'sinnovations.
Other disputes over Chevronfundamentalssoon followed. Althoughless
dramaticthan Cardoza-Fonseca,on balancethey too suggestedthatthe Court
was not preparedto accept the full implicationsof the decision.
With respect to one issue-what type of agency decision is entitled to
Chevrondeference-the Courthas been faithfulto Chevron'slogic. In Adams
Fruit Co. v. Barrett,75the Court held that Chevron applies only to agency
decisions that are an exercise of delegatedauthority.Thus, a statementby the
Departmentof Labor about the scope of private judicial remedies available
under a federal statute was entitled to no deferencebecause the agency was
given no authorityto regulatethis matter.Althoughthis decision cuts back on
the potential scope of the Chevrondoctrine,it is consistentwith the language
and rationale of Chevron,with its focus on the delegation of interpretative
authorityimplicit in the delegationof decisional authority.
With respect to other issues, however, the Court has departed from
Chevron'slogic. For example, the Courthas held that Chevrondeferencedoes
not apply to legal interpretationsadvancedfor the first time in appellatebriefs
filed on behalf of an agency.76This conclusionis consistent with established
tenets of administrativelaw.77But unlikethe propositionrecognizedin Adams
Fruit,a general"posthoc rationalizationof counsel"exceptionis not consistent
with Chevron.If Congress has delegatedauthorityto an agency to enforce a
particularstatute,andtherefore(perChevron)hasimpliedlydelegatedinterpretative authorityto the agency,it shouldnot matterif the agency announcesits
interpretationthrougha brief as opposed to, for example, an opinion letter
writtenby the generalcounsel or an informaladjudication(bothof which have
been held to be entitled to Chevrondeference78).Indeed, Justice Scalia has
recognizedin a law review articlethatthe "posthoc rationalization"of counsel
exception is at odds with Chevron.79Whateverthe implicationsof Chevron
75. 494 U.S. 638 (1990). The holdingin AdamsFruitwas foreshadowedby JusticeScalia's concurring
opinion in Crandonv. United States, 494 U.S. 152, 168-84 (1990), where he arguedthat legal opinions of
the Office of Legal Counsel of the Departmentof Justice were not entitled to Chevrondeference because
they dealt with mattersover which that office had no enforcementresponsibility.
76. Bowen v. GeorgetownUniv. Hosp., 488 U.S. 204, 212 (1988).
77. See SEC v. Sloan, 436 U.S. 103, 117-18 (1978); InvestmentCo. Inst. v. Camp,401 U.S. 617, 628
(1971); SEC v. Chenery Corp., 318 U.S. 80 (1943).
78. See PensionBenefit Guar.Corp.v. LTVCorp., 110 S. Ct. 2668 (1990) (Chevrondeferenceto prior
opinion letters); Mead Corp. v. Tilley, 490 U.S. 714 (1989) (same); JapanWhaling Ass'n v. American
CetaceanSoc'y, 478 U.S. 221 (1986) (Chevrondeferenceto CommerceDepartmentcertificationof national
compliance with treaty obligations);Young v. CommunityNutrition Inst., 476 U.S. 974 (1986) (Chevron
deference to FDA notices setting "tolerancelevels").
79. Scalia, supra note 39, at 517-18.
988
The Yale Law Journal
[Vol. 101: 969
might be for this issue, however,it is clear that the Courtis not aboutto start
deferring to governmentlegal briefs.
Anotherissue that has resultedin a potentiallyimportantqualificationof
the Chevronframeworkconcernsthe statusof canonsof statutoryconstruction.
Chevronitself does not mentioncanons, and it is unclearwhere they fit under
the framework.Althoughthe canons clearly qualify as a "traditionaltoo[l] of
it is odd to think of a court concludingon the basis
statutoryconstruction,"80
of a canon that Congress had an intention"on the precise question at issue."
Canons are maxims or rules of thumbthat allow courts to imputeanswersto
interpretativequestionswhen it is not possibleto discernby moredirectmeans
what the legislatureintended. Thus, one would expect that if a court could
interpreta statuteonly by invokinga canon,it would be forcedto acknowledge
that the issue is one to which Congresshas not clearly spoken and that,under
Chevron,it must thereforedefer to a reasonableinterpretationby the administrativeagency. Conceivably,the court would considerthe canons at step two,
in asking whether the agency interpretationis reasonable.81But again, if an
agency interpretationis consistentwith the languageand purposeof a statute,
it is hard to see how it could be condemnedas unreasonablesimply because
a judicial canon would suggest a contraryresult.
The Court first directly confrontedthe issue of canons and Chevronin
EdwardJ. DeBartolo Corp. v. Florida Gulf Coast Building & Construction
Trades Council.82 The Court acknowledged that the relevant provision of the
National Labor Relations Act was ambiguous and that the National Labor
RelationsBoard(NLRB)hadinterpretedthe provisionto prohibitcertaintypes
of handbilling.But the Courtalso thoughtthe NLRB'sconstructionwouldraise
serious constitutionalquestions.Relying on the canon of avoidinginterpretations that raise serious constitutionalquestions,83the Courtin effect held that
the result suggested by Chevronwas trumpedby the result suggested by the
canon.Thus,insteadof integratingthe canoninto step one-on the groundthat
it was indicative of congressionalintent-or step two-on the groundthat it
illuminatedthe question of reasonableness-the Courtused the canon to oust
the Chevronframeworkaltogether.
The implications of this decision for the future of Chevronare hard to
assess. On the one hand, DeBartolo could mean that Chevronitself is just
another canon, as Professor Cass Sunstein has argued.84If true, this would
raise a whole host of unanswered-and probablyunanswerable-questions
aboutwhich canons overrideChevronand which are subordinateto Chevron.
80. ChevronU.S.A., Inc. v. National ResourcesDefense Council, 467 U.S. 837, 843 n.9 (1984).
81. Cf. Note, Chevron and Canons of StatutoryConstruction,58 GEO.WASH.L. REV.829 (1990)
(arguingthat "statutedefining" canons should apply at step one while "statuteapplying"canons should
apply at step two).
82. 485 U.S. 568 (1988).
83. See NLRB v. Catholic Bishop, 440 U.S. 490 (1979).
84. Sunstein, supra note 4, at 2113.
1992]
Executive Precedent
989
On the other hand, in Rust v. Sullivan,85 the Court applied Chevron and re-
fused to follow DeBartoloin circumstanceswherethreeJusticesfoundthatthe
thus establishingthat it raised
agency interpretationwas unconstitutional,86
"seriousconstitutionaldoubts."Rust could mean thatDeBartolo will be confined to its facts.87In any event, in holding that at least one canon trumps
Chevronon at least one occasion, the Courthas furthercalled into doubt its
commitmentto the stricturesof the Chevrondoctrine.88
A thirddevelopmentconcernsagency interpretationsthatare at odds with
prior judicial precedent.Chevronholds that statutoryambiguities are to be
resolved, if possible, by agencies not courts.Thus, if a court has construedan
ambiguousstatute one way (without the benefit of an agency view), and an
agency later has occasion to read the statutedifferently,the logic of Chevron
would seem to indicatethat the court shouldfollow the agency interpretation,
rather than the earlier judicial pronouncement.Indeed, the D.C. Circuit in
Chevronhadinvalidatedthe EPA'sinterpretationof "stationarysource"largely
becauseit was contraryto priorD.C. Circuitprecedent,and the Courtheld that
this did not justify a departurefrom deference.89
Nevertheless, the SupremeCourt,in Maislin Industries,U.S. v. Primary
Steel, Inc.,90emphaticallyrejectedthe idea that an agency might disregarda
previousjudicial constructionof a statute,at least when it is a precedentof the
Supreme Court: "Once we have determineda statute's clear meaning, we
adhereto that determinationunderthe doctrineof stare decisis, and we judge
an agency's later interpretationof the statuteagainst our prior determination
of the statute'smeaning."91It is difficult to quarrelwith this conclusion, for
there would be somethingunsettlingabouta worldin which executive branch
administratorscould "overrule"SupremeCourtdecisions.Again, however,the
85. 111 S. Ct. 1759 (1991).
86. See id. at 1780-86 (Blackmun,J., dissenting,joined by Marshall& Stevens, JJ.).
87. Chief Justice Rehnquist'sopinion for the majoritywas less than illuminatingon the question of
why the constitutionalissues did not raise a "seriousquestion."He simply statedthatpetitioners'constitutional arguments,although"not withoutsome force,"were not enough to "carrythe day."Id. at 1771. The
opinion arguablycan be read as limiting the DeBartolo canon to cases in which the agency interpretation
would be unconstitutional,as opposed to merely raising a "serious question"of constitutionality.
88. Othercases havepresentedquestionsinvolvingthe interactionof Chevronandthe canons,buthave
not shed much light on the issue. In EEOCv. ArabianAm. Oil Co., 111 S. Ct. 1227, 1230 (1991), the Court
relied heavily on the canon thatstatutesarepresumednot to applyoutsidethe territoriallimitsof the United
States, and refusedto defer to the contraryjudgmentof the EEOC.But the Courtdid not applythe Chevron
framework,so the decision does not say anythingaboutthe relationshipbetween Chevronand the canons.
In Michigan Citizens for an Indep. Press v. Thornburgh,868 F.2d 1285, 1292-93 (D.C. Cir. 1989), aff'd,
493 U.S. 38 (1989), the D.C. CircuitreasonedthatChevronsupersedescanons based on substantivepolicy
considerations,such as the canonrequiringthatexceptionsto the antitrustlaws be narrowlyconstrued.The
Court granted certioraribut affirmed by an equally divided vote, and thus wrote no opinion.
89. Chevron U.S.A., Inc. v. National Resources Defense Council, 467 U.S. 837, 841-42 (1984); see
also Mesa VerdeConstr.Co. v. NorthernCal. Dist. Council of Laborers,861 F.2d 1124, 1129-31 (9th Cir.
1988) (en banc) (interpretingChevronto permitagency to change interpretationnotwithstandingSupreme
Court affirmance of prior view).
90. 110 S. Ct. 2759 (1990).
91. Id. at 2768.
990
The Yale Law Journal
[Vol. 101: 969
outcome does not follow from the logic of Chevronand must be counted as
yet anotherqualificationon that doctrine.
In short, the post-Chevrondecisions explicatingthe meaning of Chevron
reveal much that calls into questionthe Court'sfidelity to the doctrine.In one
instance-Cardoza-Fonseca-the Court adopted,albeit only temporarily,an
exception that would swallow the rule. Other decisions reveal a pattern of
qualifyingor compromisingthe doctrinewith otherprinciplessuggestingthat
Chevronrests at best uneasily in the largerbody of administrativelaw.
C. The RevolutionTransformed
Perhaps the most significant post-Chevrondevelopment,however, is a
subtlebutimportantmodificationin the statementof the relevantinquiryat step
one. As we have seen, Chevronformulatedthatinquiryin termsof whetherthe
court could "clearly"discern that Congress"hadan intentionon the precise
question at issue."92If this thresholdrequirementwere faithfully followed,
there is little doubt that it would mark a majorshift of interpretativepower
towardthe executive branch:it is a rarecase wherea courtcan fairly say that
Congressthoughtabout,let alone formulateda clear view on, the preciseissue
in controversy.The "specific intentions"formulationthereforeoperatesas an
engine of judicial deference. By the same token, however, if the threshold
determinationfor independentjudicial resolutionat step one were described
differently-for example, if courts were instructedto ask whetherthe statute
has a general meaning thatresolves the controversy,even if Congresshas not
specifically addressedthe issue at hand-then the balance might shift back
towardindependentjudgment.In short,underthe two-stepChevronframework,
everythingturnson the theory of judicial interpretationadoptedat step one.93
Post-Chevroncases have in fact begun to change the formulationof the
step-oneinquiry.The first sign of changewas whenopinionsbeganto dropany
referenceto "specific intentions"or whetherCongresshad"clearlyspokento"
the issue at hand and instead describedthe thresholdinquirysimply in terms
of whether the statute was "ambiguous"or "unclear."94
Then, beginning in
1988 with the K Mart Corp. v. Cartier,Inc decision,95a more dramaticchange
emerged: the Court began to describe the inquiry at step one in terms of
whetherthe statute has a "plainmeaning."96This rubric,an offspring of the
92. 467 U.S. at 843 n.9.
93. See Anthony,supra note 7, at 18-25.
94. See, e.g., Mobil Oil Exploration& ProducingSoutheastInc. v. United DistributionCos., 111 S.
Ct. 615 (1991); PittstonCoal Groupv. Sebben,488 U.S. 105 (1988); Honig v. Doe, 484 U.S. 305 (1988);
Young v. CommunityNutrition Inst., 476 U.S. 974 (1986).
95. 486 U.S. 281 (1988). The K Mart approachwas anticipatedin Bethesda Hosp. Ass'n v. Bowen,
485 U.S. 399 (1988), Justice Kennedy'sfirst opinion for the Court.
96. See, e.g., Demarestv. Manspeaker,l l S. Ct. 599 (1991) (offering no deferenceto administrative
construction that is contrary to plain language of statute); Sullivan v. Stroop, 110 S. Ct. 2499 (1990)
(upholding agency interpretationas being compelled by plain meaning of statute); Public Employees
1992]
Executive Precedent
991
"new textualism"espoused more generallyby Justices Scalia and Kennedy,97
has not been followed uniformly.Some opinionscontinueto quotethe language
of Chevron about whether Congress has spoken to the precise question at
issue.98The trend,however,has been stronglyaway fromthe originalChevron
formulationof step one.
Whatare the consequencesof substitutingtextualismfor intentionalismat
step one? By itself, such a shift does not necessarilyreducedeference.Indeed,
if the inquiry at step one is formulatedin terms of whetherthe statutorytext
discloses that Congresshas spokento the precisequestionat issue, this results
in even greater deferenceto agency views, becauseCongresshas undoubtedly
"spoken to" fewer issues in text than it has throughsome combinationof
textualand nontextualsources.But at the same time thatthe Courthas dropped
the languageof intentandsubstitutedthe languageof plainmeaning,it has also
droppedthe referenceto "theprecisequestionat issue." In K Mart, for example, Justice Kennedy's opinion stated that in ascertainingwhetherthere is a
plainmeaning,"thecourtmustlook to the particularstatutorylanguageat issue,
as well as the language and design of the statuteas a whole." 99 In this fashion, the "plain meaning" inquiry has tended in practice to devolve into an
inquiry about whether the statute as a whole generates a clearly preferred
meaning.10?
The movement from "specific intention"to "plain meaning"to "plain
meaningconsideringthe design of the statuteas a whole"is butone shortstep
away from "best meaning."In other words, with a shift at step one from the
"specificintentions"rubricto the "plainmeaning"rubric,the Courthasmoved
the thresholdinquirya long way towardthe exercise of independentjudgment.
Justice Scalia has candidlyadmittedthat the use of textualismat step one has
this consequence:
One who finds more often (as I do) that the meaning of a statuteis
apparentfromits text andfromits relationshipwith otherlaws, thereby
finds less often that the triggeringrequirementfor Chevrondeference
exists. It is thus relativelyrarethat Chevronwill requireme to accept
an interpretationwhich, though reasonable,I would not personally
adopt. Contrariwise,one who abhorsa "plainmeaning"rule, and is
willing to permit the apparentmeaning of a statuteto be impeached
RetirementSys. v. Betts, 492 U.S. 158, 171 (1989) ("[N]odeferenceis due to agency interpretationsat odds
with the plain language of the statute itself.").
97. See generallyWilliamN. Eskridge,Jr.,TheNew Textualism,37 UCLA L. REV.621 (1990); Patricia
M. Wald, The Sizzling Sleeper: The Use of LegislativeHistory in ConstruingStatutesin the 1988-89 Term
of the United States SupremeCourt, 39 AM. U. L. REV.277, 281 (1990).
98. See, e.g., Pension Benefit Guar.Corp. v. LTVCorp., 110 S. Ct. 2668 (1990); MeadCorp. v. Tilley,
490 U.S. 714 (1989).
99. 486 U.S. at 291 (emphasisadded).FourJusticesjoined this portionof JusticeKennedy'sopinion,
making it the opinion of the Court.
100. For examples of this trend,see Dole v. United Steelworkers,110 S. Ct. 929 (1990); Sullivan v.
Zebley, 110 S. Ct. 885 (1990); Public Employees RetirementSys. v. Betts, 492 U.S. 158 (1989).
The Yale Law Journal
992
[Vol. 101: 969
by the legislative history,will more frequentlyfind agency-liberating
ambiguity, and will discern a much broaderrange of "reasonable"
interpretationthatthe agency may adoptand to which the courtsmust
pay deference. The frequencywith which Chevronwill require that
judge to accept an interpretationhe thinks wrong is infinitely greater.101
In short, the mergerof the two-step Chevronframeworkand Justice Scalia's
"plainmeaning"approachto statutoryconstruction,if consistentlyfollowed,
would dramaticallytransformChevronfroma deferencedoctrineto a doctrine
of antideference.
Table 3, which collects the results from Table 1 into two time periods,
suggests that the K Mart reformulationmay mark the beginning of such a
transformation.The first period spansthe 1985 and 1986 Terms.Chevronwas
reasonablywell establishedby this time; it had been expressly reaffirmedin
Chemical Manufacturers Ass'n v. National Resources Defense Council102in
the previousyear,andhadnot yet encounteredany significantcontroversy.The
second period covers the 1988 to 1990 Terms-after the Cardoza-Fonseca
episode and afterthe plain meaningreformulationof step one in K Mart.Table
3 indicatesthatthe Chevronframeworkwas usedmorefrequentlyin the second
period than in the first (going from 32% of deferencecases to 52%). This is
perhapswhat one would expect, given the prominenceaccordedChevronby
the controversiesthat eruptedin Cardoza-Fonsecaand othercases. However,
the rate of acceptanceof the agency constructionwent down betweenthe two
periods (from 66% to 50%),as indeeddid the acceptancerate in all deference
cases (from 72% to 59%).
TABLE 3. Post-Chevron Terms in Two Periods
A
TERM
B
C
D
E
F
CASES
TOTAL
CHEVRON CHEVRON
INVOLVING
CHEVRONFRAMEWORK:
FRAMEWORK:
CASESCIrIN
DEFERENCE
AGENCY
VIEWFRAMEWORK
AGENCY
VIEW DECIDED
AT TRADITIONAL
FACTORS
QUESTION ACCEPIED APPLIED ACCEPTED STEPTwo
1988-90
34
20 (59%)
18 (51%)
9 (50%)
5 (30%)
16 (47%)
1985-86
25
18 (72%)
8 (32%)
6 (66%)
5 (62%)
8 (35%)
101. Scalia, supra note 39, at 521. Note that Justice Scalia's explanationfor why textualismwould
resultin less deferenceis quite differentfrom mine. He seems to believe thattextualismleads to determinate
results in most cases and that introducingevidence of intentto "impeach"the text muddiesthings up and
thus requiresdeference. My argumentis that textualismwill answer the "precisequestion"at issue in so
few cases that it leads courts to abandonthe quest for specific congressional answers, thus allowing a
dramaticallyexpandedjudicial role at step one.
102. 470 U.S. 116 (1985).
Executive Precedent
1992]
993
Again, the shift is producedby changesin only a handfulof cases, and the
precisepercentagesare not thatmeaningful.But it is possible thatwhatwe are
seeing here is at least caused in partby the emergenceof the new plain meaning approachto step one associated with K Mart. Some confirmationof this
is supplied by the numberof cases reachingstep two of the Chevronframework. The more expansive the theory of the judicial role is at step one, the
fewer occasions thereare to move on to step two. Althoughthe Courthas not
given much considerationto step two at any time throughoutthe post-Chevron
period,it is interestingto note thatthe frequencydeclinedeven furtherbetween
the earlier and later periods.
III. WHAT'S WRONG WITH CHEVRON?
Chevron'sadoptionof a generaltheoreticalframeworkfor structuringthe
choice betweenindependentjudgmentanddeferencewas an importantadvance
over the formlessnessof the previousera. Unfortunately,evidence is mounting
that the Court picked the wrong framework.I have already described the
primarysymptomof dysfunction:the lack of congruencebetweenChevronand
the actual practice of the SupremeCourt in determiningwhetheror to what
extent to deferto administrativeinterpretationsof statutes.Theoryandpractice
diverge in many ways: the failure to apply Chevronin at least half the cases
in which, by its own terms, it shouldgovern;the continuinguse of traditional
factors of deferencewhich Chevronappearsto renderirrelevant;the creation
of numerousexceptions to Chevronthat do not seem to cohere with the decision's rationale;the developmentof a differentversionof the Chevrondoctrine
that greatly expands the judicial role at step one.
It is possible to arguethatthese manifolddeviationssimply reflect lapses
of judgmenton the partof the Court,and thatthe properresponseis to identify
the "true"version of Chevronand exhortthe Justicesto follow it consistently.
But the divergences between theory and practicecanvassed in Part II are so
pervasivethatit is difficultto attributethe problemsimply to judicialbacksliding. SupremeCourtJusticesarepracticalindividuals,sensitiveto the traditions
of American constitutionalism,and the Court'spersistentrefusal to abide by
the narrowstricturesof Chevronsuggests thattheremust be somethingwrong
with eitherChevron'simplicittheoryof deference,or its practicalimplications,
or both. In this part, I will attemptto spell out why Chevronis incompatible
with fundamentaltenets of American public law, and why the framework
proves to be so unsatisfactoryin practice.
A. ChevronTheory
Chevronraises issues thatgo the heartof our understandingof the judicial
role undera system of separationof powers. In terms of formal separationof
994
The Yale Law Journal
[Vol. 101: 969
powers theory,103
interpretationof law is often said to be the exclusive province of the judiciary.104
This raises the "Marburyproblem":if it is the role
of courts to "to say what the law is,"105then how can courts defer to the
views of another branch of governmenton the meaning of the law?l06In
termsof a functionaltheoryof separationof powers,the purposeof an independentjudiciaryis often describedin termsof its capacityfor checkingarbitrariness and aggrandizementby the other branchesof government.107
This raises
the problemof agency accountability:how can we structurejudicialreview of
agency action so that agencies have enough discretionto implementcomplex
regulatoryprograms, and yet assure that they do not become a tyrannical
"FourthBranch"of government,immunefrom popularcontrol?
One of the strengths of the Chevron doctrine is that it offers, if only
implicitly, answers to the Marburyand agency accountabilityproblems.The
answers it provides,however,are radicallydifferentfrom those that were put
forthin the past, and are difficult to squarewith other,moreenduringcommitments about the properrole of the courts in a system of separatedpowers.
In the early days of modernadministrativelaw, the Marburyand agency
accountabilitydilemmaswereusuallyresolvedby borrowingfromlongstanding
notions about the relationshipbetweenjudges and juries. Courtswould defer
to agency findings of fact, but would decide all question-sof law de novo.108
Under such a division of labor,courts would clearly retain final authorityto
"say what the law is." And by independentlyascertainingthe meaningof the
agency's statutoryauthorityin all cases, courts would provide a powerful
constraintagainst arbitrarinessand aggrandizement.
This solution was short lived, however. The comparativeadvantageof
agencies is not limited to finding facts (or applying the law to facts), but
extends to resolving many questionsof law as well. And if courts decide all
questions of law de novo-even where the meaning of the law is uncertain-then the price of containingagency aggrandizementis very likely to
be judicial aggrandizement.In response to these shortcomings,the Court
abandonedthe judge-jurymodel soon after the AdministrativeProcedureAct
103. On the distinctionbetween formal and functionaltheories of separationof powers, see Rebecca
L. Brown, Separated Powers and OrderedLiberty, 139 U. PA. L. REV. 1513, 1522-31 (1991); Cass R.
Sunstein, ConstitutionalismAfter the New Deal, 101 HARV.L. REV.421, 493-96 (1987); Peter L. Strauss,
Formal and Functional Approaches to Separation-of-PowersQuestions-A Foolish Inconsistency?,72
CORNELLL. REV. 488 (1987).
104. Waymanv. Southard,23 U.S. (10 Wheat.) 1, 46 (1825).
105. Marburyv. Madison, 5 U.S. (1 Cranch) 137, 177 (1803).
106. See Monaghan,supra note 6, at 2; Eric M. Braun,Note, Coring the Seedless Grape:A ReinterL. REV.986, 988-89,
pretationof ChevronU.S.A., Inc. v. National ResourcesDefense Council, 87 COLUM.
994-95 (1987).
107. RichardH. Fallon, Jr.,Of LegislativeCourts,AdministrativeAgencies,and Article111,101 HARV.
L. REV.916, 978 (1988); see also Farina,supra note 6, at 467-99.
108. See Crowell v. Benson, 285 U.S. 22 (1932); JAFFE,supra note 13.
1992]
Executive Precedent
995
was enacted,graduallydevelopingthe multifactoredcontextualapproachthat,
as we have seen, dominatedthe pre-Chevronera.-09
Chevron in effect advances a third solution to the Marburyand agency
accountabilityproblems. The Marburyproblem is resolved by a theory of
congressionallymandateddeference.Courtsreconciletheir duty to "say what
the law is" with the practiceof deferringto agency interpretationsof law by
positing that Congress, in conferringauthorityon an agency to administera
statute, has implicitly directed courts to defer to the agency's legal views.
Accountability is achieved under Chevron by reducing the role of judicial
review and relying instead on Presidentialoversight.
Both halves of the Chevronsolutionareproblematicat best. The mandatory
deference solution to the Marburyproblemrests entirely on the presumption
thatwhen Congressdelegatesthe authorityto administera statuteto an agency,
it wants courts to defer to that agency's interpretationsof law. The evidence
that would supportsuch a presumptionis weak. Congress has never enacted
a statutethat containsa generaldelegationof interpretativeauthorityto agencies. The very practiceof enactingspecific delegationsof interpretativeauthority suggests that Congress understandsthat no such general authorityexists.
Moreover,the one generalstatuteon point, the AdministrativeProcedureAct,
directs reviewing courts to "decide all relevantquestions of law."110If anything, this suggests that Congress contemplatedcourts would always apply
independentjudgmenton questionsof law, reservingdeferencefor administrative findings of fact or determinationsof policy.
The strongestevidence in supportof the Court'spresumptionis the fact
that Congress knows about the practiceof judicial deferenceto agency interpretationsand has not acted to prohibitit."' But in order to establish that
Congress has mandatedthe practiceof deference,the Courtshould be able to
point to more than a debatableinferencefrom congressionalinaction.12Chev109. See supra notes 7-29 and accompanyingtext. I describe how the contextual approachcan be
squaredwith the Marburyand agency accountabilityconcernsat infratext accompanyingnotes 175-76 and
accompanyingtext.
110. 5 U.S.C. ? 706 (1988) (emphasisadded).
111. For a number of years, Senator Dale Bumpers gatheredconsiderablesupport for a proposed
amendmentto the APA thatwould have affirmativelyprohibitedanyjudicialdeferenceto agencyinterpretations of law. See generally James T. O'Reilly, Deference Makes a Difference:A Studyof Impacts of the
Bumpers Judicial Review Amendment,49 U. CIN. L. REV.739 (1980). Failure to enact the Bumpers
Amendment may suggest congressional ratificationof the deference doctrine. Drawing inferences from
unenactedlegislation is always hazardous,however,and here more so than usual. Although the Bumpers
Amendmentwas never enacted, it drew consistent (and overwhelming)support-hardly suggesting that
Congress is in fact strongly committed to the deference doctrine. Instead,Congress may have failed to
approvethe amendmentout of inertia, because of timing problems,or because the impact of deferenceis
too diffuse to generatethe concentratedpolitical supportneeded to producelegislative action. See Farina,
supra note 6, at 473-74.
112. A recent decision dealing with judicial deference to an agency's interpretationsof its own
regulationssought to explain the practice of deference in terms of a presumptionof congressionalintent
based on the agency's "historicalfamiliarity"with the issue and its "policymakingexpertise."Martinv.
OccupationalSafety & Health Review Comm'n, 111 S. Ct. 1171, 1177 (1991). In Martinthe argumentin
favor of the presumptionof intent is based on policy argumentsin favor of deferring to executive views,
996
The Yale Law Journal
[Vol. 101: 969
ronitself, in definingthejudicialrole in the interpretationof statutes,suggested
that courts have no authorityto impose decisionalrules that cannotbe traced
to an authoritativejudgment of Congress. Yet the Court could point to no
statuteindicatingthat Congresshas requiredthatagencies, ratherthan courts,
A decisionalframeworkthatrests
interpretambiguitiesand gaps in statutes.'13
on two foundations,one of which contradictsthe other,is difficult to regard
as sound.
Chevron'ssolutionto the agencyaccountabilityproblemis also unsatisfactory. Unlike previousdiscussionsof the accountabilityproblem,which tended
to assume that popular control comes about only through the election of
representativeswho pass statutes that are then enforced by courts, Chevron
perceives a dual channelof popularcontrol:one operatingthroughthe election
of representativeswho pass statutes;the other through the election of the
President who directs the agents who implementthose statutes. The Court
sought to forge a formulathat would allow both channelsof popularcontrol
to operate by limiting courts to the enforcementof unambiguouslegislative
directives,leaving all discretionarydecisions to be disciplinedby Presidential
oversight.
The Court'sperceptionthatthereis a dual channelof controlis an important insight."4But Presidentialoversighthas inherentlimitations.Many administrative entities-including the "independent"regulatoryagencies and
"legislative"or Article I courts-enjoy variousdegreesof statutoryimmunity
from direct Presidentialcontrol."1Several prominentseparationof powers
decisions handed down since Chevronhave legitimized these immunities,116
diluting the power of the Presidentto assureoverall directionof those agents
who administer the law. But even without the Court's sanction for these
immunities,it is simplyunrealistic,given the vastnessof the federalbureaucracy, to expect that the Presidentor his principal lieutenants can effectively
monitor the policymaking activities of all federal agencies.117Nor does it
the assumption being that if the Court finds these policies persuasive then Congress must too. Again,
however, the Court cited no direct evidence that Congress has in any context actually endorsed these
policies.
113. See Michael Hertz, Textualismand Taboo: Interpretationand Deferencefor Justice Scalia, 12
L. REV. 1663, 1665-68 (1991).
CARDOZO
114. See Harold Bruff, Legislative Formality,AdministrativeRationality,63 TEX.L. REV.207, 217
(1984); Lloyd Cutler& David Johnson,Regulationand the Political Process, 84 YALEL.J. 1395, 1410-12
(1975); RichardJ. Pierce, Jr.,TheRole of the Judiciaryin Implementingan Agency Theoryof Government,
64 N.Y.U. L. REv. 1239 (1989).
115. See Miller, supra note 44; Peter L. Strauss,The Place of Agencies in Government:Separation
L. REV.573 (1984).
of Powers and the Fourth Branch, 84 COLUM.
116. See Freytag v. Commissioner, 111 S. Ct. 2631 (1991) (suggesting that it is constitutionally
permissiblefor Congress to create an Article I courtoutside both the executive and the judicial branches);
Mistretta v. United States, 488 U.S. 361 (1989) (upholding statute creating a commission exercising
rulemakingpower as partof thejudicial branch);Morrisonv. Olson, 487 U.S. 654 (1988) (upholdingstatute
creating independentprosecutorinsulatedfrom removal by the Presidentor Attorney Generalexcept for
good cause).
117. See Strauss, supra note 115.
1992]
Executive Precedent
997
seem wise or appropriateto leave controlof agency behaviorto congressional
In the end, the primaryprotectionagainst arbitraryor
oversight hearings.118
action
aggrandizing
by agencies must remainthe fundamentalconstitutional
limitation on all executive action-that it "comportwith the terms set in
And the only effective institutionalmechanism for
legislative directives."119
this
is judicial review.
constraint
preserving
To be sure, Chevrondoes not eliminate all judicial enforcementof legal
limitations on executive action. But by restrictingcourts to enforcementof
"specificintentions"or "clearandunambiguous"
statutorydirectives,it seriousweakens
the
while offering no adequate
check
on
abuses
ly
primary
agency
alternativein its place. Of particularconcern here is the matterof enforcing
boundarylimitations.Overthe years,the SupremeCourthaspermittedincreasingly broaddelegationsof discretionaryauthorityto agencies, but only on the
understandingthat the exercise of this delegatedauthoritywould be subjectto
To the extent that broaddelegationsare often
independentjudicial review.120
Chevron
undermines
this understandingby suggestingthat courts
ambiguous,
must defer to an agency's interpretationof the scope of its own authority.'21
As ProfessorCynthiaFarinaputs it:
[A key assumptionof Chevron-] that Congress may give agencies
primaryresponsibilitynot only for makingpolicy within the limits of
their organic statutes,but also for defining those limits wheneverthe
text andsurrounding
legislativematerialsareambiguous-is fundamentally incongruouswith the constitutionalcourse by which the Court
came to reconcile agencies and separationof powers.122
A possible solution to this particulardifficulty would be to carve out yet
anotherexception to Chevron,this time for decisions that implicatethe scope
of the agency's own jurisdiction.In fact, three Justicesendorsedjust such an
exception in Mississippi Power & Light Co. v. Mississippi ex rel. Moore.'23
The difficulty with the proposedexception,as JusticeScalia pointedout in his
concurringopinion, is that virtually any decision about the meaning of the
118. See Richard J. Lazarus, The Neglected Question of Congressional Oversight of EPA: Quis
CustodietCustodes(WhoShall Watchthe WatchersThemselves)?,LAW& CONTEMP.
PROBS.205 (Autumn
1991) (describing multiple difficulties associated with congressionaloversightof EPA).
119. HaroldJ. Krent,Separatingthe Strandsin Separationof Powers Controversies,74 VA. L. REV.
1253, 1256 (1988).
120. See Touby v. United States, 111 S. Ct. 1752, 1757-58 (1991), where the Court rejected a
nondelegationdoctrinechallenge to regulationsnot subjectto judicialreview, butonly afterconcludingthat
the regulationscould be reviewedin individualenforcementactions.JusticeMarshall'sconcurringopinion
expressly statedthatthe availabilityof judicial review was critical to the conclusionthatthe delegationwas
constitutional.Id. at 1758 (Marshall,J., concurring);see also Skinnerv. Mid-AmericaPipeline Co., 490
U.S. 212, 218-19 (1989); Yakus v. United States, 321 U.S. 414, 426 (1944).
121. Farina, supra note 6, at 487-88; Cass R. Sunstein,InterpretingStatutesin the RegulatoryState,
103 HARV.L. REV.407, 446 (1989).
122. Farina, supra note 6, at 487-88.
123. 487 U.S. 354, 386-87 (1988) (Brennan,J., dissenting,joined by Marshall& Blackmun,JJ.).
The Yale Law Journal
998
[Vol. 101: 969
substantivetermsof a statutecan be viewed as eitherexpandingor contracting
an agency's "jurisdiction."'24
The questionat issue in Chevronitself, concernthe
definition
of
a
ing
proper
"stationarysource"of air pollution,has implications for the scope of the EPA'sregulatorypower. So a "jurisdictionalquestions"exception to Chevronwould eitherswallow the rule or lead to arbitrary
decisions based on explication of a notoriousformalism("jurisdiction").
In sum, Chevronseeks to resolve the centraltheoreticalproblemsof the
modernadministrativestateby adoptinga dubiousfiction of delegatedauthority
and by reducingthe role of the courts to a point that threatensto undermine
the principal constitutionalconstrainton agency misbehavior.Given these
failings, it is small wonder that the Court often seems wary of the Chevron
doctrine, applying it inconsistentlyat best.
B. ChevronPractice
The Court'sgeneralfailureto abideby the teachingsof Chevron-whether
acts
of omissionor commission-also suggeststhatthe frameworkfrequentby
ly generates unwantedresults. There are a numberof reasons why Chevron
does not functionwell as a guide to determiningthe relativeweightto be given
to judicial and agency views.
1. The SequentialInquiry
Perhapsthe most basic problemstems from the fact thatChevroncasts the
relevantdeterminantsin a sequentialhierarchy:considerationof judicialcompetence first, considerationof administrativecompetencesecond. By sequencing
the inquiry in this fashion, Chevronalmost guaranteesthat in every case the
independentviews of the judiciarywill be given either too much or too little
weight, and concomitantly,that the views of the agency will be given either
too little or too much deference.If the issue is resolved at step one, then the
court gives no considerationto the views of the executive and decides the
matterindependently.If the issue is resolvedat step two, then the courtregards
the agency view as dispositive,unlessit can say thatit is unreasonable.In other
words, Chevron inevitably generates "one-sided"decisions. In those cases
where the Courtfeels that one-sidednessis appropriate,Chevronwill do fine.
But if the Courtperceives that a more refinedapproachis called for, Chevron
is a source of awkwardness.The predominantsolution in the latter situation
appearsto be to ignore Chevron.
For example, Chevronseems to rule out the previouslycommon practice
of using executive interpretationsas a "comfortfactor"to confirm a judicial
124. Id. at 381 (Scalia, J., concurring).
1992]
Executive Precedent
999
If all cases
interpretationbased primarilyon the text and legislative history.125
fell into neat piles-unambiguous statutesin one pile, ambiguousstatutesin
the other-this would entail no loss. Realistically,however, there is a broad
gray areawherethe existence of an administrativeinterpretationmighthelp tip
the balance.Surelyit makesno sense to forbidcourtsto drawuponthe consensus of the two branchesin assuringthe public that the correctresult has been
reachedin such a case.
Similarly,it was possible underthe pre-Chevronregime to look to executive interpretationsto disconfirma conclusionbasedon an independentanalysis
of text and history.Thus, even if the courtfound thatjudicialprecedentor the
text suggested one answer,it mightfind this tentativejudgmentoutweighedif,
say, a contraryadministrativeinterpretationhadbeen consistentlyfollowed for
many years, and this executive view had been ratified by a subsequentCongress.126Chevron'ssequentialinquiryrendersit impossible to reach this kind
of result.Althoughallowingadministrative
to overcomeindepeninterpretations
dent analysis in this fashionis morecontroversial,it is really not thatdifferent
fromthe commonpracticeof allowingjudicialprecedentto overcomea conclusion drawn from an independentexaminationof text and history.127
Indeed,
allowing especially strongexecutive interpretationsto dominatejudicialreadings provides a way of injecting a dynamic or evolutionary element into
statutoryinterpretation-in a way that is far more consistentwith democratic
theory than permittingcourts to do the updating.128
The other half of the one-sidednessproblem,of course, is that Chevron
eliminatesany significantjudicialinputat step two. In practice,this has proved
to be less of a problem because a court that is convinced that judicial input
would be desirablehas ways of avoiding the conclusion that the case should
be decidedat step two: it can "rig"the inquiryat step one, eitherby overstating
the evidence that Congress had a specific intentionor, more likely today,by
finding that the meaning of the statuteis "plain."Thereare several examples
125. See supra text accompanyingnote 9.
126. See, e.g., GuardiansAss'n v. Civil Serv.Comm'n,463 U.S. 582, 591-93 (White, J.); id. at 619-20
(Marshall,J., dissenting)(statingthatlongstandingagencyinterpretationratifiedby Congressshouldprevail
over contraryinterpretationsuggested by priorSupremeCourtopinion);Bob Jones Univ. v. United States,
461 U.S. 574, 599 (1983) (relying on agency interpretationratified by Congress to bolster conclusion
supportedonly weakly by statutorytext and history);see also John P. Dwyer, The Pathology of Symbolic
Legislation,17 ECOLOGY
L.Q. 233,284-315 (1990) (arguingthatcourtsshoulddeferto agencyinterpretation
mitigating the extreme consequencesof "symbolic"environmentallaws).
127. See Pattersonv. McLean CreditUnion, 491 U.S. 164, 173 (1989).
128. Judicialupdatingnot only violates normsof democraticlegitimacy,but is also inconsistentwith
the concept of judicial "interpretation"as conventionallyunderstood.See Steven D. Smith, Law Without
Mind,88 MICH.L. REV.104 (1989) (criticizingT. AlexanderAleinikoff, UpdatingStatutoryInterpretation,
87 MICH.L. REV.20 (1987)); see also William N. Eskridge,Jr.,DynamicStatutoryInterpretation,135 U.
PA. L. REV.1479 (1987) (advocatinginterpretationof statutesin light of evolving societal, political, and
legal context).
1000
The Yale Law Journal
[Vol. 101: 969
Alternativeamong recently decidedcases that I would put in this category.129
ly, the courtcan simply ignorethe Chevronframework,following the Supreme
Court'smost common avoidancedevice. But both techniquesrest on misapplication of supposedly controllinglaw and representan unhappysolution to a
problemthat should not exist. It would seem far betterto devise a methodfor
courts to articulatetheir misgivings directly.
2. The Dividing Line
A second set of problemsis generatedby the methodfor determiningthe
dividingline betweenthe two steps in the sequentialinquiry.As we have seen,
the balance is determinedby the theory of interpretationadoptedat step one.
As originallyformulated,Chevrondescribedthe judicialinquiryin sucha way
that it would almost certainlyproduceroutine deferenceto agency views.
Chevron'stheory of the judicial role at step one is the statutoryanalogue
of what in constitutionallaw has been called "strictintentionalism."As Dean
PaulBresthas put it, "Strictintentionalismrequiresthe interpreterto determine
how the adopterswould have applieda provisionto a given situation,and to
The problemwith such a theory of interpretationis
apply it accordingly."130
thatno matterhow precisethe meaningof the text, unanticipatedquestionswill
invariablyarise about how the text is to apply in differentcircumstances.13
129. See, e.g., Dole v. United Steelworkers,494 U.S. 26 (1990) (rejectingOffice of Managementand
Budget's construction of the PaperworkReductionAct largely on the basis of structuralargumentsand
canons of construction);Sullivan v. Zebley, 493 U.S. 521 (1990) (rejectingthe Secretaryof Health and
Human Service's interpretationof the Social Security Act based on inferencesdrawn from reading two
provisionstogetherand equitableconsiderations);PublicEmployeesRetirementSys. v. Betts, 492 U.S. 158
(1989) (findingan agencyinterpretationof the Age Discriminationin EmploymentAct contraryto the "plain
meaning"of the statute,even thoughdeterminingthe precise meaning of the Act was a "somewhatmore
difficult task").
130. Paul Brest, The MisconceivedQuestfor the Original Understanding,60 B.U. L. REV.204, 222
(1980). Such a theoryof interpretationis vulnerableto attackfor all the reasonstextualistslike JusticeScalia
attackintentionalismgenerally:Congressenacts texts, not intentions,see, e.g., Pennsylvaniav. Union Gas
Co., 491 U.S. 1, 29 (1989) (Scalia, J., concurringin part and dissenting in part);In re Sinclair,870 E2d
1340 (7th Cir. 1989) (Easterbrook,J.) (holding that where conflict exists, languageof statuteprevailsover
USINGANDMISUSING
LEGISLATIVE
A REEVALUATION
HISTORY:
legislative history);U.S. DEP'TOFJUSTICE,
HISTORY
OFLEGISLATIVE
IN STATUTORY
INTERPRETATION
(1989); reconstructinghistorical intentionsis
very difficult given the limited informationwe have available; see, e.g., Wisconsin Pub. Intervenorv.
Mortier,111 S. Ct. 2476, 2488 (1991) (Scalia, J., concurring)(statingthat committeereportsare unreliable
as a "genuineindicatorof congressionalintent");and heavyrelianceon legislative historyleads to strategic
manipulationby legislative insiders, see, e.g., Blanchardv. Bergeron,489 U.S. 87, 97 (1989) (Scalia, J.
concurring) (suggesting that references to district court decisions were inserted in committee report by
staffers "to influence judicial construction").But it is not my purpose here to take sides in the debate
between intentionalism(whetherof the strict or generalvariety) and textualism.The only point is thatthe
particularbrandof intentionalismendorsedby Chevron-strict intentionalism-results in a very small role
for courts at step one.
131. See REEDDICKERSON,
THEINTERPRETATION
ANDAPPLICATION
OF STATUTES
(1975); Frank
Easterbrook,The Role of OriginalIntentin StatutoryConstruction,11 HARV.J.L. & PUB.POL'Y59, 62-63
(1988). The point has been recognized in a variety of contexts. See, e.g., Robert Bennett, Objectivityin
ConstitutionalLaw, 132 U. PA. L. REV.445 (1984); MarthaField, Sourcesof Law: The Scope of Federal
CommonLaw, 99 HARV.L. REV.881, 942-45 (1986).
1992]
Executive Precedent
1001
In the familiar example,'32if the statute says "no vehicles are permittedin
the park,"does this languageapplyto bicycles?Or,if the legislatureanticipates
this application,and provides that "no motor vehicles are permitted in the
park,"does this version apply to motorizedwheel chairs?Given the limits of
human imagination,it is virtually impossiblefor the legislatureto formulate
a specific intentionabouthow a provisionshouldbe appliedto all or even most
situations.The process of application,especially over time, will always outrun
the understandingof even the most farsightedlegislature.For this reason, to
ask the interpreterto "ascertai[n]thatCongresshadan intentionon the precise
question at issue," as Chevrondemands,'33is to ask a question that will yield
an affirmativeanswer-at least an honest affirmativeanswer-in only a tiny
numberof cases.134
Underthe Chevronframework,thereis only one way to expandthe small
judicial role contemplatedby the specific intentionstheory-reformulate the
natureof the inquiryat step one. The cure,unfortunately,may be as bad as the
disease. As we have seen, recent decisions have begun to drop the specific
intentions requirementin favor of a "plainmeaning"formulation.35But this
leaves unansweredseveralcriticalquestionsaboutwhatkindsof interpretative
aids courtsmay consult at step one. Mustthe courtfind the statuteplain on its
face? Or may it be plain after consulting extrinsic sources, like dictionary
definitions, definitions contained(explicitlyor implicitly)in otherstatutes,or
considerationsof the structureof the act? In its more latitudinarianversions,
where the court looks to considerationsof structureand purposein deciding
whetherthe meaning is "plain,"the plain meaning inquirytends to converge
with a "clearlypreferredmeaning"approach.136If this happens,then Chevron
would suffer from the exact opposite of the problemthatplaguedthe original
formulation:if step one does not completely swallow step two, at least it will
have dramaticallyexpanded-perhaps too far-in the opposite direction.
On other occasions the Court has not committed to either a "specific
intentions"method,or to a textualistmethod,buthas statedthe relevantinquiry
simply in terms of whether the statute is "ambiguous"or "unclear."137This
formulationleaves even more questionsunanswered.To what extent may the
courtresolve the meaningof the statuteat step one basedon inferencesdrawn
from legislative history or considerationsof statutorypurpose? May it use
canons of constructionbasedon linguisticusage?Canonsbasedon substantive
132. See. e.g., H.L.A. HART,THECONCEPT
OFLAW(1961); Brest, supra note 130, at 209-10.
133. ChevronU.S.A., Inc. v. National ResourcesDefense Council, 467 U.S. 837, 843 n.9 (1984).
134. The Court has recognized this point outside the Chevroncontext. See Moskal v. United States,
111 S. Ct. 461, 467 (1990) ('This Courthas never requiredthat every permissibleapplicationof a statute
be expressly referredto in its legislative history.");Diamond v. Chakrabarty,447 U.S. 303, 315 (1980)
("ThisCourtfrequentlyhas observedthata statuteis not to be confinedto the 'particularapplication[s]...
contemplatedby the legislators.'") (quoting Barr v. United States, 324 U.S. 83, 90 (1945)).
135. See supra text accompanyingnote 99.
136. See supra text accompanyingnote 100.
137. See cases cited supra note 94.
1002
The Yale Law Journal
[Vol. 101: 969
policy considerations?Given the wide range of possible answers to these
questions,the Chevronframeworkbecomes at best unpredictable,and at worst
a vehicle for complete substitutionof judicialjudgmentfor the views of the
agency.
In short, the reformulationof the judicial inquiryat step one will have at
least one of two consequences.First, it will result in a considerablesacrifice
of the formality of the original Chevronstructure,and with it, much of the
simplicityand predictabilitythatmadethe two-stepprocedureseem attractive.
Moreproblematically,if the reformulationresultsin a greatlyenlargedjudicial
role at step one-as recent cases suggest may be happening-then it would
have the paradoxicaleffect of elevating the courtsratherthan agencies to the
role of primarypolicymaker.The originalrationalefor Chevron-the need to
have statutorygaps filled by the most democraticallyaccountabledecisionmaker-would be turnedon its head.
Thereare severalreasonswhy we shouldview the potentialtransformation
of Chevroninto a doctrineof antideferencewith alarm.The practiceof deferring to executive interpretationsof statutesperformsmany valuablefunctions:
it allows policy to be made by actorswho are politically accountable;it draws
upon the specialized knowledge of administrators;it injects an element of
flexibility into statutoryinterpretation;and it helps assurenationallyuniform
constructions.138
As statutes become increasinglycomplex, and courts tend
to be increasinglyformalistic in their approachto statutoryinterpretation,a
robustmeasureof deferenceto agencyinterpretations
may be, if anything,more
than
In
ever
if
before.
courts
do
the
fact,
imperative
gap filling at step one
underthe "plainmeaning"nostrumby applyingdictionarydefinitions,rulesof
grammar,and canons of construction,then the contentof nationalpolicy will
be determinedby courts without any considerationof the substantivevalues
at issue in the policy disputes-either those that animatedCongressor those
articulatedby the agency chargedwith administrationof the statute.
Finally, there is a more fundamentaldifficulty with the focus on judicial
interpretationat step one. The Chevronframeworkmakesthe decisionto defer
turn on some conception,either explicit or implicit, of what kind of judicial
interpretationis legitimate. But there is no a priori reason why the sphereof
deferenceshould always be the exact mirrorimage of whatevera majorityof
the Court believes on any given day is permissible in the name of judicial
interpretation.Surely there are other variables-such as the degree of the
agency's expertise and the existence of reliance interests implicated by the
138. See generally Diver, supra note 11, at 585-92 (discussing a variety of utilitarianargumentsfor
deference to administrativeinterpretations,includingagency expertise and flexibility); RichardJ. Pierce,
TheRole of Constitutionaland Political Theotyin AdministrativeLaw, 64 TEX.L. REV.469, 520-21 (1985)
(arguingthat deferenceto agency interpretationspromotespolitical accountability);Silberman,supra note
44 (deferenceserves ends of democratictheoryand uniformity);PeterL. Strauss,OneHundredFifty Cases
Per Year: Some Implicationsof the SupremeCourt'sLimitedResourcesfor Judicial Review of Agency
Action, 87 COLUM.L. REV. 1093 (1987) (deferencerequiredto assure uniformityin interpretation).
1992]
Executive Precedent
1003
agency's interpretation-that also ought to have some bearingon the decision
to defer.By makingthe transitionfromindependentjudgmentto deferenceturn
solely on the amenabilityof the issue to independentjudicial resolution,the
Chevronframeworkexcludes these considerations.
C. A New and ImprovedChevron?
Could the Courtimproveon the presentstate of affairs by tinkeringwith
the Chevronframework?No doubtsome beneficial reformsare possible. For
example, the Courtcouldreformulatethe inquiryat step one in termsof degree
of judicialcertitudeaboutstatutorymeaning,ratherthanin termsthatimplicate
Thus, it could ask whetherthe statuteadmitsof only
interpretativemethod.139
"one possible meaning,"ratherthan posing the question in terms of whether
Congress had a "specific intention"on the point or whetherthe text of the
statuteis "plain."Such a reformmight offer a "compromise"position on the
judicial role that would provide a more robustcheck on agency misbehavior
thanthe originalChevronwithoutgoing to the oppositeextremeof encouraging
judicial substitutionof judgment.
But no matter how much tinkeringis done with the formulationof the
inquiry at step one, a numberof serious and irremediableproblems would
remain.The frameworkwould still provideno logical role for the traditional
deferencefactors;it would still generateone-sideddecisions;and it would still
requirethe deference decision to be based exclusively on a considerationof
judicial competenceto exerciseindependentjudgmentto the exclusionof other
factors. Moreover,it is unclear that any formulationof the properjudicial
inquiry at step one can-over a wide range of cases-satisfactorily reconcile
the desire to drawuponthe comparativeadvantagesof agencies as interpreters
with the desire thatcourtscontrolagencyarbitrarinessand aggrandizement.In
light of the manifold problemsassociatedwith the Chevrondoctrine,and the
persistentgulf between theory and practice,it is time to considerwhetheran
alternativeframeworkmight do better.
IV. THE EXECUTIVE
PRECEDENT
MODEL
In this part, I offer an alternativeto Chevron-the executive precedent
model. The basic idea assimilates the deference doctrine into the general
judicial practiceof following precedent.The differenceis thatthe "precedent"
the courts would be asked to follow is generatedby a different branch of
government. Just as courts generally adhere to interpretationsof statutes
139. See JAFFE,
supra note 13, at 572 (suggestingthat the relevantinquiryis whether"thejudges are
themselves convinced that [a] certain reading, or application,of the statute is the correct-or the only
faithful-reading or application");Sunstein,supra note 4, at 2092 ("If the court has a firm conviction that
the agency interpretationviolates the statute,that interpretationmust fail.").
1004
The Yale Law Journal
[Vol. 101: 969
embodiedin judicial precedents,so they wouldgenerallyfollow the interpretations of statutesembodiedin prior decisions by executive branchagencies.
A. ExecutiveInterpretationof Law
The executiveprecedentmodelbeginswith the premisethatentitiescharged
with the execution of the law, includingthe President,executive departments,
and independent regulatory agencies, have an inherent power to interpret
enactedlaws. Eventhoughthe Presidentandthe entitiesthatassistthe President
in the execution of law have no inherentpower to "makelaw,"'40once Congress has delegated authorityto executive actors under law, the executive
agencies must determinewhat that law means, and need not await a further
delegation of interpretativeauthorityfrom Congress to do so.
The constitutionalbasisfor an inherentexecutivepowerto interpretthe law
is straightforward.The Constitutionexpresslygrantsthe President"executive
power,"and directs the Presidentto see that the laws are "faithfullyexecuted."141The conferral of these powers would seem to presuppose that the
Presidentandthose who serve underhis directionhavethe capacityto ascertain
the meaningof the law.Indeed,law interpretation
is an inevitableandnecessary
of
the
of
the
constitutional
functionsof the executive
byproduct
performance
branch.142
as
courts
Just
must interpretthe law in orderto resolve cases and
controversiesthat arise within their jurisdiction,so executive officials must
interpretthe law in orderto promulgateregulations,bringenforcementactions,
instructemployees how to carry out programs,or perform any of the other
myriadtasksentrustedto agencies.In fact, becauseonly a fractionof executive
actionsend up in court,administrativeactorsengage in law interpretationwith
greater frequencyand over a wider range of cases than courts do.'43If only
the courts had the capacityto interpretlaw, our system of governmentcould
not continue to function.
Moreover,althoughexecutive interpretationsoccur in a variety of "formats,"'44they share much in common with judicial precedent.145Executive
interpretationsare usuallywritten,aregenerallyaccompaniedby statementsof
140. See YoungstownSheet & Tube Co. v. Sawyer, 343 U.S. 579, 585 (1952) (steel seizure case);
Thomas W. Merrill, The Judicial Prerogative, 12 PACEL. REV.(forthcoming 1992).
141. U.S. CONST.art. H, ?? 1, 3.
142. See Bowsher v. Synar, 478 U.S. 714, 733 (1986) ("Interpretinga law enacted by Congress to
implementthe legislative mandateis the very essence of 'execution' of the law.");United States v. Nixon,
418 U.S. 683, 703 (1974) ("In the performanceof assigned constitutional duties each branch of the
Governmentmust initially interpretthe Constitution,and the interpretationof its powers by any branchis
due great respect from the others.");FrankH. Easterbrook,PresidentialReview,40 CASEW. RES.L. REV.
905 (1989-90).
143. See, e.g., JERRY
L. MASHAW,
BUREAUCRATIC
JUSTICE
186 (1983) (judicial review touches less
than one percent of the Social Security disabilitycaseload).
144. See Anthony,supra note 7, at 7-14.
145. See FrederickSchauer,Precedent, 39 STAN.L. REV.571, 572 (1987) (noting that reliance on
precedentnot limited to courts of law but "is part of life in general").
1992]
Executive Precedent
1005
reasons,'46and often respondto argumentsby affected entities supportingor
opposingthe interpretation.Furthermore,the writtenstatementsof reasonsthat
accompanyexecutive interpretationsserve many of the same functionsas do
the written opinions that accompanyjudicial decisions. They establish the
thatthe executiveconstruction
legitimacyof the interpretation
by demonstrating
has a sound basis in law and policy.147
They clarify the scope of the interpretation so that similar issues that arise in the futurecan be quickly resolved
without having to "reinventthe wheel."'48They offer instructionsto subordinate employees for applyingthe interpretationin analogous but nonidentical
circumstances.Finally,they explain to membersof the public the position the
governmenthas taken so that they can make appropriateadjustmentsin their
behaviorand plan for the future.149
Thus, it seems highly appropriateto speak
of executive interpretationsas a separate system of precedent-executive
precedent.150
B. The Judicial Systemof Following Precedent
Given that both the executive branch and the judicial branch generate
interpretationsof statutesthatcan be regardedas "precedent,"whatis the basis
for the judicial practice of deferring to executive precedent?The executive
146. In this respect at least, it is more naturalto speak of executive branchdecisions as "precedent"
than it would be to speak of congressional legislation as precedent.Administrativeagencies, like courts,
typically vote to approveofficial explanationsgiven for their actions, such as agency adjudicatoryorders
or Federal Register notices. Congress, in contrast, usually does not incorporatestatementsexplaining
particularrules set forth in statutes.Explanationsmay be found in legislative history,but Congress does
not as a whole vote to approvelegislative history,nor is it sent to the Presidentfor his signatureor veto.
Still, the Court has on occasion treatedcongressional legislation as "precedent"supportinga particular
interpretationof the Constitution.See Field v. Clark, 143 U.S. 649, 690-91 (1892) (treatingprior statutes
delegatingdiscretionto executive as precedentsupportingthe constitutionalityof delegation).See generally
James B. Thayer,The Origin and Scope of the AmericanDoctrineof ConstitutionalLaw,7 HARV.L. REV.
129 (1893) (arguing that courts should defer to legislative interpretationsof ambiguous constitutional
provisions).
147. See Henry P. Monaghan,StareDecisis and ConstitutionalAdjudication,88 COLUM.
L. REV.723,
749-55 (1988).
148. See RICHARD
THEJUDICIAL
DECISION
72-73 (1961); Schauer,supra note 145,
WASSERSTROM,
at 599.
149. See MELVIN
A. EISENBERG,
THENATURE
OFTHECOMMON
LAW10-12 (1988) (discussing the
requirementthat judicial decisions be "replicable"by private parties and their advisors);Schauer,supra
note 145, at 597-98; WASSERSTROM,
supra note 148, at 66-69.
150. Indeed,certaintypes of administrativedecisions-adjudications-have legal consequenceslargely
indistinguishablefrom judicial decisions. Executive agencies can enter ordersfor the paymentof money
(damages) and cease and desist orders(injunctions).See CommodityFuturesTradingComm'n v. Schor,
478 U.S. 833 (1986) (agency may adjudicatecommon law claim for money damages);AmericanAirlines
v. North Am. Airlines, 351 U.S. 79 (1956) (Civil AeronauticsBoard may properlyissue cease and desist
order).Failure to abide by an agency ordercan result in sanctions for contempt. Regal KnitwearCo. v.
NLRB, 324 U.S. 9 (1945). And courts "have long favored applicationof the common-lawdoctrines of
collateral estoppel (as to issues) and res judicata (as to claims) to those determinationsof administrative
bodies that have attained finality." Astoria Fed. Sav. & Loan Ass'n v. Solimino, 111 S. Ct. 2166, 2169
(1991). Although I would not confine the notion of executive precedent to statements of reasons in
adjudicatorydecisions, the parallel to judicial precedenthere is especially striking.
1006
The Yale Law Journal
[Vol. 101: 969
precedentmodel locates thatpracticein the courts'inherentpowersto develop
We know that the "judicial
principlesfor resolving cases and controversies.151
in
III
the
Article
of following precedescribed
practice
power"
encompasses
dent, even though this practice is not expressly authorizedin any source of
positive law.152Indeed, the norm of following precedentis ubiquitousin the
Anglo-Americanlegal system.Theexecutiveprecedentmodelwouldassimilate
the judicial practiceof deferringto executive interpretationsinto this system
of following precedent.Courtswould be deemedto have inherentauthorityto
develop guidelinesfor deferringto executiveprecedentsin appropriatecircumstances.
This inherentauthority,like otherproductsof judicialself-governance(such
as common law rules of procedure153),
would be subject to congressional
modification.154
Thus, if Congresswere expresslyto instructcourtsto deferor
not to defer to particular agency interpretations,these instructions would
supersedethe norms of following precedent.But absent express instructions
fromCongressaboutthe allocationof authority,courtswoulddeveloptheirown
standardsfor determiningwhen to defer to priorexecutive interpretationsof
law.
In determiningwhereexecutiveprecedentmightfit into the judicialsystem
of following precedent,it is useful to considerbriefly the variousways courts
treat different types of judicial precedent.None of these understandingsis
compelled by any statutoryor constitutionalprovision. Rather, they have
emerged over the years througha gradualprocess of judicial self-definition.
Decisions of superiorcourtsarethe mostpowerfulformof precedent.They
are regardedas legally binding on lower courts.155A court's own previous
151. See generally Chambersv. NASCO, Inc., 111 S. Ct. 2123, 2132 (1991) ("It has long been
understoodthat '[c]ertain implied powers must necessarilyresult to our Courts of justice from the nature
of their institution,'powers 'which cannot be dispensedwith in a Court,becausethey are necessaryto the
exercise of all others."') (quoting United States v. Hudson, 11 U.S. (7 Cranch)32, 34 (1812)).
152. Therecan be no doubtthatthe Framersof the Constitutionunderstoodthatthe normof following
No. 78, at 471
precedent was an integral element of the functioning of courts. See THEFEDERALIST
(AlexanderHamilton)(J. Cooke ed., 1961) ("Toavoidan arbitrarydiscretionin the courts,it is indispensable
that they should be bounddown by strictrules and precedents.").For example, Blackstone,a familiarlegal
source to the Framers,pronouncedthat precedentmust be followed "unless flatly absurdor unjust."1
WILLIAM
COMMENTARIES
ONTHELAWSOFENGLAND
70 (Facsimile ed. 1979).
BLACKSTONE,
153. See Sibbachv. Wilson & Co., 312 U.S. 1, 9-10 (1941); Waymanv. Southard,23 U.S. (10 Wheat.)
1, 43 (1825).
154. In the contextof constitutionaladjudication,ProfessorMonaghanhasquestionedwhetherCongress
would have the authorityto direct the Courtto disregardthe principle of stare decisis. Monaghan,supra
note 147, at 754-55. But given Congress' acknowledgedpower to legislate standardsof review of agency
action underdelegated statutoryauthority,there can be little doubt that Congress could direct the courts
to decide questions of statutoryinterpretationde novo, or (short of constitutionalviolation) to deem an
agency's interpretationauthoritative-or anythingin between.
155. See, e.g., ThurstonMotorLines v. JordanK. Rand,Ltd.,460 U.S. 533, 535 (1983); Huttov. Davis,
454 U.S. 370, 375 (1982). It is not plausible to view this norm as simply grounded in an empirical
generalizationthat a lower court will get reversedby a superiorcourt if it fails to follow superiorcourt
precedent.The U.S. SupremeCourt,for example,does not begin to have the institutionalcapacityto review
all decisions of the federalcourtsof appealsto ensurethatthey remainfaithfulto SupremeCourtprecedent.
See Peter Strauss, One Hundred Fifty Cases Per Year, 87 COLUM.L. REV. 1093 (1987). The same is
1992]
Executive Precedent
1007
decisions are not as binding as decisions of superior courts. But their own
opinions are also accordedmore than deference;underthe doctrine of stare
decisis, they are regardedas enjoying a very strong presumptionof correctness.156 In practice, this means they will be followed unless they are overruled, and generally speaking they will be overruledonly if there is a very
strongjustification for doing so.'57
A third type of precedentis made up of prior decisions of tribunalsof
coordinatejurisdiction.The most familiarexample is when one federal court
of appealsconfrontsan issue thathas been decidedpreviouslyby anothercourt
of appeals. Here the judicial attitudeis more uncertain.English courts and
nineteenth-centuryAmericancourtsdevelopedthe doctrineof the "controlling
decision," whereby the precedentof a court of coordinatejurisdiction was
regardedas binding as a matterof comity.'58After the creationof the federal
courts of appeals in 1891, and the reconceptualizationof the role of the SupremeCourtas being devotedlargelyto the resolutionof conflicts betweenthe
courtsof appeals,the attitudeof the federalcourtsof appealsrelaxedconsiderably.159Some courts came to speak of a duty of following the precedentof
another circuit unless it could be said to be "clearly erroneous."'60Others
stressed that such precedentis not "binding,"and implied that intercircuit
precedentwould be followed only to the extent it is "persuasive"-suggesting
that it is entitled to no deferenceat all.161Perhapsthe most accuratesynthesis
of the contemporaryunderstandingis thatone courtof appealswill "give most
respectful considerationto the decisions of the other courts of appeals,"162
probablytrue in many of the large state court systems.
156. California v. FERC, 110 S. Ct. 2024, 2028-30 (1990); Pattersonv. McLean Credit Union, 491
U.S. 164, 172 (1989); Square D Co. v. Niagara FrontierTariff Bureau, Inc., 476 U.S. 409, 424 (1986)
(referringto "the strong presumptionof continuedvalidity that adheresin the judicial interpretationof a
statute").
157. Patterson,491 U.S. at 172. It has even been arguedthatin the statutorycontext, the presumption
of correctness should be absolute. See LawrenceC. Marshall, "Let Congress Do It": The Case for an
AbsoluteRule of StatutoryStare Decisis, 88 MICH.L. REV.177 (1989).
158. See Alan D. Vestal, RelitigationBy Federal Agencies: Conflict, Concurrenceand Synthesisof
Judicial Policies, 55 N.C. L. REV.123, 129-40 (1977).
159. Id. at 140-66. Professor Vestal concluded his comprehensive 1977 survey by noting that "the
doctrine of controllingdecision may not have the vitality it once had,"although he thoughtmany of the
departuresfrom the doctrine were unexplained,and"the underlyingprinciplesthat have been the basis for
the doctrine of controllingdecision have lost none of theirvitality."Id. at 166. For an interestingargument
in supportof applyinga rule of staredecisis to intercircuitprecedent-based largelyon the desirableeffects
on the SupremeCourt's workload-see WalterV. Schaefer,ReducingCircuitConflicts,69 A.B.A. J. 452
(1983); Walter V. Schaefer,Reliance on the Law of the Circuit-A Requiem, 1985 DUKEL.J. 690.
160. See, e.g., Homan v. United States, 279 F.2d 767, 773 (8th Cir.) ("We have in a long line of
opinionsdeclaredthat,on an unsettledquestionof federallaw, while a decision by anotherCourtof Appeals
is not compulsively bindingupon us, we will, in the interestof judicial uniformity,accept it as persuasive
and follow it, unless we are clearly convinced that it is wrong."(citationsomitted)),cert. denied, 364 U.S.
866 (1960).
161. See, e.g., City Storesv. LernerShops,410 F.2d 1010, 1014(D.C. Cir. 1969) ("Decisionsof district
courts and other courts of appeals are, of course, not binding on us and are looked to only for their
persuasive effect.").
162. Colby v. J.C. Penney Co., 811 F.2d 1119, 1123 (7th Cir. 1987) (Posner,J.).
1008
The Yale Law Journal
[Vol. 101: 969
or there
especially where the prior interpretationis a "reasoneddecision"163
In effect, a legal interpretationof another
is an "accumulationof authority."164
circuit is regardedas prima facie correct,and is given more or less deference
dependingon contextualfactorssuch as the extent of considerationgiven the
issue by the other circuit and how long its precedenthas been followed.
Finally, there are precedentsof inferiorjudicial tribunals.A court will
nearly always take note of the view of the tribunalwhose decision is under
review; it is regardedat the very least as an interesting datum. Generally
speaking,however,the superiorcourtwill resolve all questionsof law de novo;
the views of the inferior court will be followed only to the extent they are
regardedas persuasive.165The decisions of inferior tribunalsother than the
one underreview are noted only sporadically.166
Thus,the generalunderstandis
that
lower
court
is
not
entitled
to any measureof deference
ing
precedent
the
tribunal.
by
superior
Which of the four attitudestoward differenttypes of judicial precedent
should provide the relevantbenchmarkfor determiningthe attitudeof courts
toward executive precedent?We can quickly dismiss the possibilities that
executive precedentshould be treatedlike the decisions of superiorcourtsor
like a court's own prior decisions. Part of our received understandingof the
"judicialpower"is thatcourtshave final say in articulatingthe meaningof the
law in cases broughtbefore them.167It would seem to follow from this that
courtsare not legally boundby executiveinterpretationsof statutes.Forsimilar
reasons,it would seem to be inconsistentwith ourunderstandingof the judicial
power to posit thatcourtsare boundby executiveprecedentsthe way they must
follow their own precedents.
The two most plausiblemodels for executive precedentare the treatment
of decisions by courtsof coordinatejurisdictionand the treatmentof decisions
by inferiorcourts. For a variety of reasons,I think that the treatmentof decisions by courts of coordinatejurisdictionprovidesthe best analogy. Some of
these reasonswill unfold only as the full implicationsof the model are spelled
163. Aldens, Inc. v. Miller, 610 F.2d 538, 541 (8th Cir. 1979), cert. denied, 446 U.S. 919 (1980).
164. 1B JAMESW. MOOREET AL., MOORE'SFEDERALPRACTICE? 0.402, at 15 (2d ed. 1984).
165. See generally Salve Regina College v. Russell, 111 S. Ct. 1217 (1991) (federal court of appeals
should not defer to district court's interpretationof state law in diversity suits).
166. Occasionallyone will find the suggestionthata widespreadconsensusamonglower tribunalsover
a long period of time is entitled to some deference.Reves v. Ernst & Young, 494 U.S. 56, 74-76 (1990)
(Stevens, J., concurring) (Court should defer to unanimousconsensus among courts of appeals about
meaning of Securities Act exemption);United States v. Ryan, 284 U.S. 167 (1931) (Courtshould hesitate
to set aside uniformconstructiongiven to statuteby lower courtsfor more than 60 years). And from time
to time, courts will rely on argumentsor insights containedin particularlower court decisions, especially
if the judge is widely respected. See, e.g., InterstateCommerceComm'n v. Brotherhoodof Locomotive
Eng'rs, 482 U.S. 270, 293 n.8 (1987) (Stevens, J., concurring)("JudgeFriendly'sdiscussionon the general
issue of denials of petitions to reopen continues to merit our respect.").
167. See United States v. Nixon, 418 U.S. 683, 705 (1974) (rejectingclaim of unreviewableexecutive
authorityto determinewhen executive communicationsare legally privileged);Marburyv. Madison,5 U.S.
(1 Cranch) 137, 170 (1803) (courts have power to compel executive officers to perform legal duty as
determinedby court).
1992]
Executive Precedent
1009
out. One overarchingreasonfor selecting the precedentof courtsof coordinate
jurisdictionas the properlodestar,however,bears emphasisat the outset.
In comparingthe situationof superiorcourts and inferior courts as law
interpreters,the comparativeadvantageresidesalmostentirelywith the superior
court.'68Thus, there is no reason for a superiorcourt to defer to the judgments of a trial court on questions of law. But when we comparecourts and
agencies, there are strengths and weaknesses on both sides of the ledger.
Executiveinterpretershavegreaterexpertiseon mattersthatarehighlytechnical
or complex;they have morefamiliaritywith the overallstructureof a statutory
program,and with the policies followed underthose programs;and they are
more accountableto the public. On the other hand, courts are more insulated
from political pressuresthan agencies; their membersare more likely to be
selected for theirlegal abilitiesthanare agencyheads;they may be able to hire
better law clerks; and they may have more time to do research and write
opinions, if only because they are exempt from the statutorydeadlinesoften
imposed on agencies.
The point is not thatagenciesare always betterinterpretersthancourtsare
or vice versa, but simply that there are advantagesand disadvantageson both
sides. This suggests that, at least as a first approximation,the treatmentof
precedentof courtsof parallelauthorityappearsto supplythe best benchmark
for establishingthe judicial attitudetowardexecutive precedent.'69Of course,
this is just an analogy. There are a numberof aspects in which the treatment
of executive precedentshould differ from that of intercircuitprecedent.The
analogy is useful, however,for it demonstratesthatit is possible to groundthe
deference doctrine in the practice of following precedent.As we shall see,
moreover,the treatmentof the precedentof courtsof parallelauthorityappears
168. See Salve Regina College, 111 S. Ct. at 1221; Dan T. Coenen,ToDefer or Not to Defer: A Study
of Federal Circuit Court Deference to District CourtRulings on State Law, 73 MINN.L. REV.899, 923
(1989).
169. If courts should treatthe interpretationsof executive agencies like the precedentof a tribunalof
coordinate jurisdiction, then should agencies treat the opinions of courts the same way, i.e., as only
presumptivelycorrectandentitledto moreor less weightdependingon contextualfactors?This is a variation
on the problem of agency "nonacquiescence"in judicial interpretationsof law. See Samuel Estreicher&
RichardRevesz, Nonacquiescenceby Federal AdministrativeAgencies, 98 YALEL.J. 679 (1989); Vestal,
supra note 158. My answer would be that, as a matterof theory,agencies must give judicial precedentno
more deference than courts must give executive precedent.Courtsrenderjudgmentsthat bind the parties
before them (includingagencies) and suchjudgmentsmust of coursebe obeyed. They also issue statements
of reason-opinions-that help readers(includingagencies)predicthow they will resolve futurecontroversies. But they do not issue edicts that bind all the world, as the Constitution,a statute, or an agency
regulationdoes. See John Harrison,TheRole of the Legislativeand ExecutiveBranchesin Interpretingthe
L. REV.371 (1988); Edwin Meese, III, The Law of the Constitution,61 TUL.
Constitution,73 CORNELL
L. REV.979 (1987). In theory, therefore,agencies could regardjudicial precedentthe same way courts
regardagency precedent.
At a practical level, however, the fact that courts review agencies ratherthan vice versa may well
dictate a more accommodatingattitudeby agencies. Thus, agencies may wish to "acquiesce"in judicial
opinions for futurecases even if those opinions are not bindingas a matterof law, and even if the agency,
as a matter of interbranchprecedent-following,is not convinced by the court's analysis.
1010
The Yale Law Journal
[Vol. 101: 969
to sharemany of the main featuresof the administrativedeferencedoctrineas
it has been historicallypracticedand understood.
C. The Model Stated
If we view the deferencedoctrineas partof the system of precedent,then
the decision whether or to what extent to defer to executive interpretations
would in each case entail a three-partinquiry:(1) Is therean executive precedent?(2) How strongis thatprecedent?(3) Given the strengthof the executive
precedent,does an independentjudicialexaminationof the questionof statutory
interpretationcompel a differentresult?
1. Is Therean ExecutivePrecedent?
Not every interpretationof law by an executive agency shouldbe regarded
as a "precedent,"at least for purposes of determiningthe judicial response.
Here, I would suggest that the relevantdistinctionis betweenan interpretation
initially advancedwhen the agency is acting as an independentdecisionmaker
and an interpretationfirst advancedwhen the agency is acting as a party to
judicial proceedings.Insofaras the agency adoptsan interpretationwhen it is
wearing its independentdecisionmakerhat, its views should be regardedas
"precedent."Includedin this categorywould be any interpretationembodied
in a decision by an agency thatis legally bindingon actorsoutsidethe agency
itself, such as the promulgationof legislativerules, the issuanceof administrative orders,or the awardof licensesandgrants.In addition,any legal interpretation embodied in an official statement of agency position that predates a
particularjudicial controversywould count as precedent. Examples would
includeinterpretativerules, policy statements,agency enforcementguidelines,
and previous opinion letters.
When an agency initially advancesan interpretationwhile it is wearingits
judicial litigant hat, however, the interpretationshould not be regardedas
precedent.This is not because the agency loses any comparativeadvantageit
may have as an interpreterwhen it changeshats. Instead,judicial normsabout
impartialadjudicationprecludegiving the interpretationprecedentialweight.
When there are two parties before a court contestingthe meaning of the law,
it is generally understoodthat the court should decide the matterimpartially
and independently,without giving one party's views greater weight simply
because of that party's status.170
Thus, insofar as the agency view is not embodied in a previous authoritativeexpression of its views, the agency view
shouldbe treatedas no differentfromthe positionof any otherparty.Examples
170. See Fishgold v. Sullivan Drydock & RepairCorp., 154 F.2d 785, 789 (2d Cir.), aff'd, 328 U.S.
275 (1946); Anthony,supra note 7, at 60 ("Itwould exceed the boundsof fair play to allow an institutionally self-interestedadvocacyposition, which 'may properlycarry a bias,' to controlthe judicial outcome.").
1992]
Executive Precedent
1011
of agency action that fall into this categoryand thus shouldnot be regardedas
precedentwould includean interpretationimplicitin the decision of a prosecutor to file criminal charges, an interpretationreflected in the decision of an
agency to institute judicial enforcement proceedings, or an interpretation
advancedfor the first time in legal briefs filed in courtdefendinga previously
taken agency action.
2. How StrongIs the ExecutivePrecedent?
Once it is determinedthatthereis an executiveprecedenton point,the next
step would be to ascertain how "strong"it is. This inquiry, which would
necessarily be fairly judgmental,would entail two distinct components.
The first componentturns on the level in the executive hierarchyfrom
which the intepretationemanates.Under the executive precedentmodel, the
authorityof executive agencies to interpretlaw derives not from a delegation
of interpretativeauthorityfrom Congress,but from the President'sexecutive
power and constitutionalobligationto see thatthe laws are faithfullyexecuted.
Thus,the more directlyaccountablethe interpreteris to the President,the more
the decision would partakeof the President'sconstitutionalauthorityand the
an
stronger it would be regarded as executive precedent. To illustrate:171
interpretationcontainedin a letter from a regionaloffice of the Departmentof
Health and Human Services (HHS) would not be as strong as one contained
in a "programinstruction"adopted by the general counsel of HHS, which
would not be as strong as one reflected in a policy directive issued by the
Secretaryof HHS, which would not be as strong as an interpretationdirectly
approvedby the President. This hierarchyof authority,of course, directly
parallelsthe understandingof the strengthof precedentin thejudicialhierarchy,
wherethe SupremeCourtpartakesof the "judicialpower"by directgrantfrom
the Constitution,andlowercourtspossess thejudicialpoweronly in accordance
with their position in relationto the SupremeCourt.172
The second componentof the strengthof executive precedentis derived
by analogy to the treatmentof precedentof courts of coordinatejurisdiction
within the system of judicial precedent.Considersome of the factors that a
courtmightlook to in determiningwhetherto follow the precedentof a judicial
tribunal of coordinate authority:How long has the precedentof the other
tribunalbeen followed? Has it been criticizedby laterdecisions, or is it inconsistent with anotherprecedentof equal status?Has the public come to rely on
171. The illustrationis derived from Miller v. Youakim,440 U.S. 125, 144 n.25 (1979).
172. Unlike the Chevrondoctrine,the executiveprecedentmodeldoes notrequirethattherebe a unitary
executive branch.See supra note 44. Everythingelse being equal, however,the greaterthe "independence"
of an executive entity from Presidentialoversightand control, the weakerits precedent.Thus, a statutory
interpretationadopted by the Securities and ExchangeCommission-an "independent"regulatoryagency-would be entitled to less deference than would an interpretationadoptedby a Cabinetofficer such as
the Attorney General or the Secretaryof HHS, both of whom are removableby the Presidentat will.
1012
The Yale Law Journal
[Vol. 101: 969
the interpretation?Does the interpretationconcern a matteras to which the
othertribunalhas particularfamiliarityor expertise?Wasthe issue fully briefed
and arguedin the othertribunal,and is the othertribunal'sdecision supported
of the othertribunal
by a thoroughandreasonedopinion?Has the interpretation
been ratified by subsequentcongressional action? As developed more fully
below, these factors bear an uncannyresemblanceto the contextual factors
courts developed in the pre-Chevronera in orderto determinewhether,or to
what extent, to defer to executive interpretations.The analogy to intercircuit
precedenthelps to sharpen our understandingof why many of these factors
should be relevant and which ones are most importantin particularsettings.
3. Does an IndependentJudicial JudgmentCompela DifferentResult?
Once the courthas establishedthatthereis an executive precedentand has
determined (roughly) how strong that precedent is, then it must make an
independentinquiryto determinewhetherthereis sufficientreasonto overcome
the presumptionof correctnessthatsucha precedentenjoys.In undertakingthis
inquiry, the court could call upon any and all of the "traditionaltools of
that would be relevant in interpretingthe statute
statutoryinterpretation"173
de novo. The conclusionsreachedthroughthis kindof analysis-ranging from
a judgmentthatthe statutecan haveonly one possible meaningto a conclusion
that it is utterlyopaque-would then be consideredin light of the conclusions
reachedaboutthe strengthof the executiveprecedent.A strongprecedent-such
as an interpretationadopted by the Secretary of HHS in a well reasoned
decisionconsistentlymaintainedovera long periodof time-would be rejected
only if the courtwere firmly convincedit was wrong.A weak precedent-such
as a recent letter from a regionaloffice of HHS assertingthe interpretationin
a conclusoryfashion-would be disregardedif the courthada fair doubtabout
its correctness.
This is an admittedlyimprecisedecisional formula.But there are simply
too many relevantvariablesto reducethe inquiryto a simple algorithm.Nor
is it particularlysurprisingthatthis shouldbe the case. Afterall, the same must
be said of the role of judicial precedentin statutoryinterpretation:no single
formula can capturethe process by which courts determinethe competing
claims of priorjudicialprecedentanda de novojudicialexaminationof statutory languageand legislative intent.The only differencewhereexecutive precedent is involved is thatthe precedentcomes from a differentbranchof government.174
173. ChevronU.S.A., Inc. v. National ResourcesDefense Council, 467 U.S. 837, 843 n.9 (1984).
174. An alternativebasis for a model not groundedin presumeddelegationis suggestedby the doctrine
of practical construction:administrativedecisions could be seen as a form of "conduct"that is evidence
of the practicalconstructionof a statuteby the partychargedwith its implementation.See Annotation,Effect
of practical or administrativeconstructionof a statuteon subsequentjudicial construction,73 L. Ed. 322
(1929). The doctrine of practicalconstructionis often encounteredin cases construingtreaties,see Eastern
1992]
Executive Precedent
1013
V. THE MODELSCOMPARED
The proposedexecutive precedentmodel sharesmany of the strengthsof
the Chevron doctrine in that it strives to encourage courts to defer to the
judgments of political actors who have greater public accountabilityand
specialized knowledge.Yet, I will contend,it avoids Chevron'smajortheoretical and practicalfailings. Moreover,the model is not radical,unless perhaps
one assumes (contraryto fact) that the full-blown Chevrondoctrine actually
defines the statusquo. Indeed,a majorpart of the justificationfor the model
is that it coheresbetterwith both existing and historicalpracticethan does the
Chevronformulation.Finally, the executive precedentidea has consequences
for judicial, administrative,and congressionalbehaviorthat, on balance, are
superiorto the effects of Chevron.
A. OvercomingChevron'sTheoreticaland Practical Problems
Any theory of deferencemust come to terms with the Marburyproblem
and the problemof agency accountability.We can now see thatthere are three
general strategies for accomplishingthese ends.'75
The first-which prevailedfleetinglywhenmodernadministrativelaw was
taking form-is to allow courts to review all questions of law de novo. This
solution totally eliminates the Marburyproblem and provides a powerful
potentialcheck on agency abuse.But it deprivesus of any comparativeadvantage agencies may have as interpreters,and it gives the courts too much
discretionarypower over the formulationof policy.
The second strategy-that of Chevron-is grounded in the notion of
mandatorydeference.Courtsseek to remainfaithfulto theirdutyto "saywhat
the law is" while deferringto executive interpretationsby positing that Congress has mandatedthatthey defer.Agency accountabilityis securedby having
courtsenforceclearstatutorydirectivesandby relyingon Presidentialoversight.
As we have seen,'76this strategy, as pursued by Chevron, falls victim to
several difficulties, includingthe need to adopta doubtfulfiction of delegated
interpretativeauthority,and to abandona good partof the only effective check
on the abuse of power by agencies.
Airlines, Inc. v. Floyd, 111 S. Ct. 1489, 1499 (1991); Air France v. Saks, 470 U.S. 392, 403 (1985), and
traces its origins to the doctrine of practicalconstructionin the law of contracts.See SAMUEL
WILLISTON,
A TREATISE
ON THELAWOFCONTRACTS
? 623 (3d ed. 1961). As suggested by these origins, however,
the doctrineultimatelyrestson the idea of mutualassent-if bothpartiesto an agreementconcurin a certain
practicalimplementation,thenthis suggeststhatbothpartieshave assentedto this construction.The absence
of anythingresemblinga conventional bilateralagreementin the agency setting may make this particular
analogy problematicas a basis for a model of deference to administrativedecisions.
175. See Stephen Breyer,Judicial Review of Questionsof Law and Policy, 38 ADMIN.L. REV.363,
368-69 (1986).
176. See supra text accompanyingnotes 7-29.
1014
The Yale Law Journal
[Vol. 101: 969
The third strategy-that pursuedby the executive precedentmodel-is
groundedin discretionarydeference.Courtsremainfaithful to the duty to say
what the law is, but in exercising this duty they conclude as a matterof selfgovernancethat the best guide to the meaningof the law is often to follow the
prior interpretationsof a coordinatebranchof government.At the same time,
courts hold agencies accountableby upholdingonly the exercise of sound
executive discretion:they affirmdecisionsthatpresenta combinationof strong
precedentand/orcongruencewith congressionalintent, but they reversedecisions thatpresenta combinationof weakprecedentand/ortensionwith congressional intent.This approachthereforeallows courtsto continueto act as a check
on abuse of delegated power.
1. TheoreticalFailings
By groundingthe practiceof deferencein normsof judicial comity rather
than legislative compulsion,the executive precedentmodel avoids Chevron's
most glaring weakness:the need to posit a fictitious delegationof power from
Congress to executive agencies. Instead,the practiceof deferenceis derived
from two relatively uncontroversialconstitutionalpropositions:(1) executive
entities have an inherentpower to interpretthe laws they are chargedwith
enforcing, and (2) the courts have an inherentpower to develop norms for
followingprecedentthatcan be extendedto encompassthe practiceof deferring
to executive precedent.
Because the executive precedentmodel grounds deference in inherent
powers,it is unaffectedby the ambiguityoverCongress'trueintentionsregarding the allocation of interpretativeauthority.Chevronfails in the face of this
uncertaintybecause ambiguitydoes not add up to a mandatorydirective to
defer.But if courtshave inherentpowerto decide when to defer,an ambiguous
congressional attitude simply means Congress has done nothing that would
requirecourts to desist from their existing practiceof deference.
In addition,the model harmonizesbetterwith establishedunderstandings
of constitutionalstructure.At a formallevel, Chevronrecognizesan omnipotent
Congress,which is envisionedas not only establishingsubstantivelegal norms,
butalso as directingthe allocationof interpretativeauthoritybetweenexecutive
agencies and courts. But in practice,because of its fiction that Congresshas
delegatedthe resolutionof all ambiguitiesto the agencieschargedwith enforcement of a statute, Chevronsanctions the exercise of vast executive powers
uncheckedby meaningfuljudicial review.
The executive precedentmodel, like Chevron,posits that Congress has
ultimatepower over the adoptionof substantivelegal norms.But the allocation
of interpretativeauthoritybetweenagencies and courts-at least for purposes
of judicial review-is determined by the courts themselves, absent some
supersedinginstructionfromCongress.By groundingthe practiceof deference
1992]
Executive Precedent
1015
in the discretionarypowers of courts, the executive precedentmodel permits
courtsto continueto performan importantcheckingfunctionin reviewingthe
exercise of delegated power by agencies. In particular,courts would not be
requiredto identify a clear statutoryviolation before they could decline to
follow executiveprecedent.Otherabusesof agencypower-including decisions
thatupset settledexpectations,decisionsthatreflectagency bias, and decisions
that seek to expand agency power beyond the sphere contemplatedbut not
expresslyspelledoutby Congress-would also be subjectto judicialcorrection.
2. Practical Failings
The executive precedentmodel, with its greaterflexibility and openness
to a wider range of potentiallyrelevantconsiderations,also avoids the major
practicalfailings of Chevron.By retainingflexibilityaboutthe relativeweights
to be given executive precedent and independentjudgment, the executive
precedentmodel avoids the extremes of Chevron,whereby the views of the
executive are given eithertoo little or too muchweight in virtuallyevery case.
In particular,the model ensuresthatthe executive branchview will always be
considered,ratherthan become relegatedto a subset of cases in which courts
do not feel confident aboutexercisingindependentjudgment.At a minimum,
therefore,the executiveprecedentmodelguaranteesthatthe entitywith the best
graspof practicalconsequences-the executive agency-at least gets its views
on the table in every case.
The executiveprecedentmodelalso sidestepsthe needto adoptany particular theory of judicial interpretationin orderto fix the point at which the court
will defer to executivejudgment.The model can be weddedto any of the rival
methods of interpretation.It is based on a contextual assessment of the
"strength"of the executive precedenttogetherwith an independentjudicial
assessment of statutorymeaning, with no a priori specification of a single
legitimate method for ascertainingmeaning.
Finally,the model is open to a wide rangeof contextualfactorsthatcourts
continue to regard as important.Under Chevron,the court is instructedto
decide whetherto defer after looking at only one variable:whetherthe court
is entitled to exercise independentjudgment. Other variables related to the
agency's authorityand its performanceare ignored. In effect, the decision
whether to defer is made with blinders on. Under the executive precedent
model, in contrast,the courtlooks at both sides of the equation:measuringfirst
the strengthof the executive precedentand then consideringthis measurement
against an independentassessmentof the clarity of the statute.
1016
The Yale Law Journal
[Vol. 101: 969
B. Congruencewith Practice
Perhapseven more impressiveis the capacity of the executive precedent
model to explainvariousfeaturesof the deferencedoctrine,both the traditional
pre-Chevronunderstandingand the recent exceptions recognized under the
Chevronregime.
1. The Status of Pre-ChevronFactors Under the Executive Precedent
Model
Quite a numberof the featuresof the deferencedoctrineas it existed in the
pre-Chevronera-features thatwere apparentlybanishedby Chevron,butseem
to persistin the actualpracticeof the Court-suddenly becomeexplicableonce
we view the practiceof deferenceas a form of following precedent.In particular, most of the deference factors have direct analogues in the practice that
courtsobservewhendecidingwhetherto followjudicialprecedent.And the few
factorsthatlack directanalogueshave a commonsensebasis andcan be readily
accommodatedto the executive precedentmodel.
(a) DeferenceAs a Sliding Scale. Pre-Chevronpractice,as we have seen,
treateddeference as existing along a sliding scale,177whereas Chevrontends
to make deferencean all-or-nothingproposition.Viewing deferenceas a form
of following precedentmakessense of the traditionalapproach.Wherejudicial
precedentis concerned,courts typically do not resolve questions of statutory
interpretationsolely on the basis of precedentfrom courtsof coordinatejurisdiction. Some nonbindingprecedentsare consideredweightierthanothers,and
considerationsdrawnfrom the precedentof othercourtsare typicallyweighed
Simiagainstthe court'sindependentanalysis of the meaningof the statute.178
under
the
executive
deference
to
executive
views
would
model,
larly,
precedent
fall along a continuum,dependingon the strengthof the executive precedent
and the degree of confidence the court has in its independentview of the
statute'smeaning.
(b) Express Delegations. Pre-Chevroncases also distinguishedbetween
"legislative rules" and "interpretativerules." This construct has no direct
analogue in the realm of judicial precedent,no doubt because it is rare for
177. See supra text accompanyingnote 8.
178. See Colby v. J.C. Penney,Inc., 811 F.2d 1119, 1123 (7th Cir. 1987) (circuit court does not give
automaticdeference to decisions of other circuits, but recognizes that "withinreason, the parties to cases
before us are entitled to our independentjudgment.");HaberleCrystal SpringsBrewing Co. v. Clarke,30
F.2d 219, 222 (2d Cir. 1929), rev'd on other grounds, 280 U.S. 384 (1930) ("Much as we respect the
considered decisions of other circuits, we conceive that our duty requires us to form an independent
judgmentin cases of first impressionin our own court,and forbidsus blindlyto follow othercircuits,when
our minds are not persuadedby the argumentsadvanced.").
1992]
Executive Precedent
1017
Nevertheless,
Congress expressly to delegate rulemakingpower to courts.179
the basic insight of the legislative/interpretativerule distinctionis sound and
shouldbe adoptedto the executiveprecedentframework.Any expressdirective
from Congressaboutthe allocationof interpretativeauthoritywould supersede
the internalnorms.But such directivesare rare.Shortof an express allocation
of interpretativeroles, a specific congressionalgrant of authorityto agencies
in a particularareashouldbe viewed as a congressionaljudgmentof the superior competenceof the agencyto developlegal normsin the areaunderconsideration,andthus shouldalso resultin greaterdeferenceto agencyinterpretations.
I would, however, reject one feature of the pre-Chevronapproachto
legislative rules. Pre-Chevroncase law tendedto speakin all-or-nothingterms.
Eitheran agency interpretationwas a "legislativerule"entitledto greatdeference, or it was an "interpretativerule" entitled only to whateverpersuasive
effect it might have.'80Unfortunately,this dichotomytended to break down
in practice. Statutes granting general substantive rulemaking authority to
agencies,ratherthanspecific authorityto definea particularterm,wereparticuI would avoid crudedichotomiesby makingthe specificilarly problematic.81'
ty of any expresscongressionaldelegationof regulatoryauthorityone contextual factor to be consideredin determiningthe strengthof the executive precedent.The specificity of the grantfromCongress,however,wouldnot necessarily override or negate other considerations.
(c) AgencyExpertise.Pre-Chevroncase law variedthe degreeof deference
accordingto whetherthe issue was one thatimplicatedagency"expertise."This
idea is encounteredless often in the realm of judicial precedent,but it is not
unheardof. For example, courtsof appealswill occasionallygive extradeference to decisionsof othercircuitsthathavegreaterfamiliaritywith the substantive body of law in question.'82And courtsquite commonlydefer to interpre-
179. The principalexception is the Rules EnablingAct, which authorizesthe SupremeCourtto adopt
rules of civil procedureand evidence. 28 U.S.C. ? 2072 (1988). This delegation,however,concernsmatters
of judicial self-governance,over which courts would presumablyhave inherentauthorityin the absenceof
a legislative delegation. See ThomasMerrill, The CommonLaw Powers of Federal Courts, 52 U. CHI.L.
REV. 1, 46-47 (1985). Anotherarguableexception is the SentencingCommission,on which three federal
judges sit. In Mistrettav. United States, 488 U.S. 361 (1989), the SupremeCourtheld that it was constitutionally permissible for Congress to constitute the SentencingCommissionas part of the judicial branch,
even though it exercises substantiverulemakingauthority.
180. See Battertonv. Francis,432 U.S. 416, 424-25 & n.9 (1977); 1 DAVIS,supra note 6, at 421-22.
181. Some pre-Chevroncases suggested that interpretationsadoptedpursuantto general rulemaking
authoritywere entitled to less weight than interpretationsmore specifically authorizedby Congress.See,
e.g., United States v. Vogel FertilizerCo., 455 U.S. 16 (1982); Rowan Cos., Inc. v. United States, 452 U.S.
247,253 (1981). Othercases suggestedthatinterpretationsadoptedpursuantto generalrulemakingauthority
were no differentfrom specific authorizations.See, e.g., FCCv. WNCN ListenersGuild, 450 U.S. 582, 594
(1981).
182. See City of Westfield v. Federal Power Comm'n, 551 F.2d 468 (1st Cir. 1977) (views of D.C.
Circuit, which reviews great bulk of Federal Power Commission orders, carry great weight when case
involving similar issue comes before other circuit);Watsonv. Allen, 254 F.2d 342, 347 (D.C. Cir. 1958)
("substantialweight" given to views of Court of Customs and PatentAppeals "as to technical or highly
specialized patent matters").
1018
The Yale Law Journal
[Vol. 101: 969
tations of state law or procedural law by other courts that have greater experience with these localized legal regimes.183
The notion that specialized competence warrants deference naturally plays
a more prominent role in the executive precedent context because of the
specialized nature of administrative agencies. Common sense suggests that the
views of those with specialized knowledge about a subject are generally more
valuable than the views of those without such knowledge. Because agencies
are specialists, it follows that additional deference should be given to their
judgments within the area of their specialization. Thus, pre-Chevron case law
accorded extra deference to an agency's interpretations of the specific statute
it was charged with administering, but did not give any special deference to
other types of agency interpretations, such as constructions of common law
terms, constitutional principles, or prior judicial precedents,184 or to interpretations of statutes the agency did not ordinarily administer.'85The executive
precedent model would carry this general distinction forward, treating the
presence or absence of specialized agency knowledge as one of the factors to
be considered in determining the strength of agency precedent.
(d) Longstanding Interpretations. Pre-Chevron cases put great stress on
whether an agency interpretation was longstanding and consistent. This has a
direct analogue in practice regarding judicial precedent. Here too, the longevity
of the precedent is clearly an important factor in determining how much weight
it will be given.'86 The case for giving greater weight to old precedents is
based primarily on two considerations. First, the fact that an interpretation has
survived for a long period of time is some evidence that it is sound, in the
sense that it is workable.'87 Unworkable interpretations generate discontent
that may result in congressional overruling, or modification or repudiation by
the court that initially offered the interpretation. The fact that none of this has
occurred is circumstantial evidence that the interpretation does not impose
undue costs on regulated entities or frustrate the basic objectives of the proponents of the legislation.'88 Second, longstanding interpretations give rise to
183. See Bishop v. Wood, 426 U.S. 341, 345-46 n. 10 (1976) (SupremeCourtdefers to interpretation
of state law adoptedby lower federal courts that are "familiarwith the intricaciesand trendsof local law
and practice");PanduitCorp. v. All States Plastic Mfg. Co., 744 F.2d 1564, 1574-75 (Fed. Cir. 1984)
(FederalCircuit adopts policy of following law of regionalcircuits with respect to proceduralquestions).
184. See, e.g., Piper v. Chris-CraftIndus., Inc., 430 U.S. 1, 41 n.27 (1977); Texas Gas Transmission
Corp. v. Shell Oil Co., 363 U.S. 263, 268-70 (1960); SEC v. CheneryCorp., 318 U.S. 80, 89 (1943).
185. See, e.g., NLRB v. Bildisco & Bildisco, 465 U.S. 513, 529 n.9 (1984); Alaska S.S. Co. v. United
States, 290 U.S. 256, 261-64 (1933).
186. See, e.g., Californiav. FERC, 110 S. Ct. 2024, 2029 (1990) (referringto "thedeferencethis Court
must accord to long-standing and well-entrencheddecisions, especially those interpretingstatutes that
underliecomplex regulatoryregimes");SquareD Co. v. NiagaraFrontierTariff Bureau,Inc., 476 U.S. 409,
421-22, 424 (1986) (refusing to overrule precedentfollowed for six decades notwithstandingextensive
changes in statutoryframeworkand relateddoctrine).
187. See Commissionerv. Fink, 483 U.S. 89, 102-03 (1987) (Stevens, J., dissenting).
188. Similar points have been advancedin the literatureseeking to explain why common law rules
tend to be efficient. See William M. Landes & RichardA. Posner,Legal Precedent:A Theoreticaland
EmpiricalAnalysis, 19 J. LAW& ECON.249 (1976); George L. Priest, The CommonLaw Process and the
1992]
Executive Precedent
1019
relianceintereststhatwarrantjudicialprotection.One of the primaryjustifications for the practiceof followingjudicialprecedentsis to promotestabilityin
the legal system by protectingthese reliance interests.'89Of course, the presumption in favor of old precedentsis not irrefutable.Generally speaking,
however,the longer and more consistentlya precedenthas been followed, the
higherthe burdenis on the partyseekingto overcomethe precedentto explain
why it is infirm.'90
Given these judicial understandings,it is not surprisingthat the duration
of an executive interpretationis the most frequentlyencounteredfactorin the
pre-Chevroncase law (and for that matterin the post-Chevroncases as well).
The rationalefor giving weight to longstandingagency precedentis identical
to that for giving weight to judicial precedent:age is indicativeof soundness
(in the sense of workability),andolderinterpretations
generaterelianceinterests
It
is
also
true
that
deservingprotection.'9l
longstandingagency precedent,like
longstandingjudicialprecedent,can be overruled.Courtscan accommodatethe
need for change by imposing a higher burdenof explanationon an agency
reversingits own longstandingprecedentthan otherwisewould be the case.192
(e) Well-ReasonedDecisions. Beginning with Skidmorev. Swift & Co.,193
the SupremeCourt'spre-Chevroncases also gave greaterdeferenceto "wellreasoned" agency decisions. In determiningwhether to follow nonbinding
precedentsin the judicial context, such as decisions of courts of coordinate
jurisdiction, courts frequentlyconsider how persuasive the reasoning of the
other courtis: whetherit has consideredall relevantarguments,supportedthe
Selection of Efficient Rules, 6 J. LEGALSTUD.65 (1977); Paul H. Ruben, Why is the CommonLaw
Efficient?, 6 J. LEGALSTUD.51 (1977).
189. See United States v. Mason, 412 U.S. 391 (1973) (stare decisis requiresthat people rely on the
Court's decisions and that they should not be penalized for such reliance);cf. Payne v. Tennessee, 111 S.
Ct. 2597, 2610 (1991) ("[C]onsiderationsin favor of stare decisis are at their acme in cases involving
propertyand contractrights, where reliance interestsare involved.").
190. See Fink, 483 U.S. at 89, 103 (Stevens, J., dissenting);WestinghouseElec. & Mfg. Co. v. Formica
Insulation Co., 266 U.S. 342 (1924) (Court will not lightly disturbrule settled by 45 years of judicial
constructionwith respect to rights arising underpatent law).
191. The need to protectreliance interestshas long been recognized as an importantreason to defer
to longstandingexecutive interpretations.As the Court observed in one case:
[G]overnmentis a practicalaffair intendedfor practicalmen. Both officers, law-makersand citizens
naturallyadjust themselves to any long-continuedaction of the Executive Department-on the presumptionthat unauthorizedacts would not have been allowed to be so often repeatedas to crystallize
into a regularpractice.Thatpresumptionis not reasoningin a circle but the basis of a wise and quieting
rule that in determiningthe meaning of a statute or the existence of a power, weight shall be given
to the usage itself-even when the validity of the practice is the subject of investigation.
United States v. Midwest Oil Co., 236 U.S. 459, 472-73 (1915); see also Brewsterv. Gage, 280 U.S. 327,
336 (1930); Logan v. Davis, 233 U.S. 613, 627 (1914); United States v. Burlington& Mo. RiverR.R., 98
U.S. 334, 341 (1879).
192. See American Hosp. Ass'n v. NLRB, 111 S. Ct. 1539, 1546 (1991) ("Giventhe extensive notice
and commentrulemakingconductedby the Board,its carefulanalysis of the commentsthatit received,and
its well-reasonedjustificationfor the new rule, we would not be troubledeven if therewere inconsistencies
between the currentrule and priorNLRB pronouncements.");SEC v. CheneryCorp., 318 U.S. 80 (1943).
See generally STEPHEN
G. BREYER
& RICHARD
B. STEWART,
ADMINISTRATIVE
LAWANDREGULATORY
POLICY
416-26 (2d ed. 1985).
193. 323 U.S. 134 (1944).
1020
The Yale Law Journal
[Vol. 101: 969
conclusion with a thorough explanation, and so forth.19 This is perfectly
sensible. Because the precedentof a coordinatetribunalis not binding,it must
compete with the court's independentanalysis of the issue. Althoughthereis
a general bias in favor of uniformity,and thus the precedentwill be regarded
as prima facie correct,a poorlyreasonedprecedentobviously commandsless
respect and requiresthatthe secondcourtdig moredeeply in orderto ascertain
for itself what the correct answer should be.
The same patternis followed in the executive precedentcontext. Here, the
pre-Chevroncase law could, if anything, have done more to emphasize the
importanceof careful agency reasoning.Whereagency factfindingand policy
judgmentsare concerned,courtshavecome to insist thatagenciestake a "hard
look" at the relevant variables and options.195There is no reason why the
preferencefor rigor should not extend to legal interpretations.
(f) InteragencyAgreement.Pre-Chevroncase law reducedthe degree of
deferencewheretwo agencieswerein disagreementaboutthe properinterpretation of a statute.This practicealso makes sense if we think of deferenceas a
form of following precedent.Wherea court faces conflicting precedentsfrom
differentjurisdictions,it is muchless likely to defer to one of those interpretations than it would where the precedentfrom other jurisdictions is unanimous.196Here too, we see how the executive precedentmodel can rationalize
pre-Chevronpracticeand integrateit with generalnormsof judicial self-governance.
A particularlyvenerablepre-Chevron
(g) Contemporaneous
Interpretations.
factor laid stress on whetherthe agency interpretationwas contemporaneous
with the enactment of the statute. Because courts are generally not active
participantsin the processof formulatinglegislation,the views of a contemporary court are not necessarily probativeof the thinking of the draftersof a
statute.Thereis, however,a moredirectanaloguein constitutionallaw: the idea
thatcourtsshouldgive deferenceto the views of the first Congresswith respect
to the meaning of the Constitution.197
The views of the first Congress are
to
be
because
so many of the membersof that
thought
especially revealing
were
active
in
the
Congress
participants
framingandratificationof the Consti194. See, e.g., Richardsv. Local 134, Int'l Bhd. of Elec. Workers,790 F.2d 633, 636 (7th Cir. 1986);
City Stores Co. v. Lerner Shops of D.C., Inc., 410 F.2d 1010, 1014 (D.C. Cir. 1969); see also Oneida
County Fair Bd. v. Smylie, 386 P.2d 374, 391 (Idaho 1963) (state supremecourt will select and apply
decisions from other jurisdictions that reflect soundestreasoning.)
195. The general tenets of the "hardlook" doctrine have been said to be: (1) agencies must offer
detailedexplanationsfor theirdecisions;(2) they mustexplaindeparturesfrom pastpractices;(3) they must
allow effective participationby affectedinterests;and(4) they mustgive considerationto possiblealternative
measures.Cass R. Sunstein, Deregulationand the Hard Look Doctrine, 1983 SUP.CT. REV.181-82. For
examples of "hardlook" review in the SupremeCourt,see Bowen v. AmericanHosp. Ass'n, 476 U.S. 610,
627 (1986); Motor Vehicles Mfrs. Ass'n v. State Farm MutualAuto Ins. Co., 463 U.S. 29, 41-44 (1983);
Baltimore Gas & Elec. Co. v. National ResourcesDefense Council, 462 U.S. 87, 105-06 (1983).
196. See Vestal, supra note 158, at 163-65.
197. See Bowsher v. Synar, 478 U.S. 714, 723-24 (1986); Marsh v. Chambers,463 U.S. 783, 790
(1983).
1992]
Executive Precedent
1021
tution,and thus presumablyhadspecial insightinto the originalunderstanding.
In effect, the decisions of the first Congressare viewed as one form of legislative branch"precedent"entitled to deferenceby the courts.
constructionfactorin administrative
The rationalefor the contemporaneous
law is similar to that in the constitutionalcontext. It has been observedthat
agency officials are often "thedraftsmenof the laws they are afterwardscalled
Even if agency actorsdid not participatedirectlyin the
upon to interpret."198
formulationof the legislation, their contemporaneousinterpretationcan be
regarded as indicative of the prevailing views about policy, given that the
administrativeagency often shares the same assumptionsthat animate the
Like its constitutionalcousin, the administrativedoctrine has
legislature.199
somewhatlimited value, and goes moreto the judicialdiscernmentof congressional intent than to ascertainingthe "strength"of the executive precedent.
Moreover,I suspectthatmanyof the referencesto contemporaryconstructions,
especially in the earlier cases, are synonyms for longstandingconstruction.
Nevertheless,even when used in its correctsignification,there is no harmin
incorporatingit as one factorin the decisional matrix,as long as its rationale
is clearly perceived.
(h) RatifiedInterpretations.Finally,pre-Chevroncases often placedweight
on evidence thatCongresshadratifiedthe executiveconstruction.The ideathat
interpretationsare entitled to greaterdeferenceif they have been ratified by
Congressalso has a preciseparallelin the worldof judicialprecedent.Congress
is presumedto be aware of the way statuteshave been interpretedby courts.
If one court adoptsa particularinterpretation,and Congresslaterreenactsthe
statute without substantively changing the interpretedlanguage, then it is
presumedthat Congress has approvedthe interpretation.20The doctrinecan
be explained by a kind of "squeakywheel" theory of the legislative process
similar to that which underlies(in part)the longstandinginterpretationfactor.
If the initial interpretationgeneratesdiscontent-in the sense that it imposes
large costs on particularinterest groups-then we can presume that those
groups will bring it to the attentionof Congress;if Congress agrees, it will
amendthe statuteto overturnthe interpretation.By the same token,if Congress
reenacts a statuteafterjudicial interpretationwithoutchangingthe language,
thatis some evidence thatthe interpretationhas notgeneratedgreatdiscontent.
Because the doctrinemakes some ratherheroic assumptions,it should be
used with caution. One problemrelates to the assumptionthat Congress is
awareof judicial interpretationsof statutes.This may be contraryto fact if the
interpretationaffects large and diffuse intereststhathave difficulty organizing
198. United States v. Moore, 95 U.S. 760, 763 (1888).
199. See, e.g., SEC v. Sloan, 436 U.S. 103, 126 (1978) (Brennan,J., concurring).
200. See, e.g., Pierce v. Underwood,487 U.S. 552, 566-68 (1988); Lorillardv. Pons, 434 U.S. 575,
580 (1978). See generally William N. Eskridge,Jr., InterpretingLegislative Inaction, 87 MICH.L. REV.
67, 70-84 (1988).
1022
The Yale Law Journal
[Vol. 101: 969
for legislative action.201
Anotherproblemis presentedby those cases that find
congressionalratificationbased on the failure of Congress to enact proposed
legislation. Legislative inaction is always inherentlyambiguous and could
reflect nothingmore thana crowdedlegislative agenda.But as long as ratification is limited to cases where at least the relevant committees have been
informed of an interpretation,and Congressreenactsthe relevant legislation
without change, it is probablya useful doctrine.
The administrativeratification doctrine-a staple of pre-Chevroncase
law-has exactly the same logic and limitations as the judicial ratification
doctrine. Administrativeinterpretationscan be just as importantto interest
andwe can expectthatthe interpretations
that
groupsas judicialinterpretations,
make well-organizedinterest groups unhappywill come to the attentionof
Congress. Thus, subject to the caveats noted about the judicial ratification
version, congressionalratificationis a permissiblefactor for courts to use in
ascertainingthe weight of executive precedent.
2. The Post-ChevronExceptionsUnder the ExecutivePrecedentModel
The executive precedentmodel also makes sense of some of the disputes
over Chevronfundamentalsthathaveeruptedin recentyears. In each case, the
model suggests that there are sound reasons to deviate from the logic of
Chevron.
(a) InterpretationsBeyond Agency Authority. In Adams Fruit Co. v.
the Courtheld that agency interpretationsare entitled to no deferBarrett,202
ence if they concern a topic that the agency has not been empoweredto regulate. In effect, Adams Fruit carries Chevron'spresumeddelegationtheory to
its ultimateconclusion.Since the delegationof regulatoryauthoritygives rise
to the duty to defer, where there is no delegatedpower there is no deference.
Pre-Chevroncase law was more refined. When an agency spoke abouta
matteras to which it had no power to regulate,its views were still entitledto
considerationby courts,providedthey boreotherindiciathatwouldentitlethem
to respect. This was the holding of Skidmore.203
The executive precedent
model would adhereto this pre-Chevronunderstanding,ratherthan to Adams
Fruit. If an agency's interpretationqualifies as executive precedentand has
other attributesthat suggest it is entitledto deference,there is no reasonwhy
it shouldnot be given appropriateweight as a form of precedent.For example,
agencies often have specializedknowledgeaboutthe operationsof a particular
industry and the way it is affected by a statutoryscheme. If the agency has
issued an interpretativeregulationor policy statementthatprovidesa reasoned
201. See MANCUR
ACTION
OLSON,THELOGICOFCOLLECTIVE
(1971); Bruce A. Ackerman,Beyond
Carolene Products,98 HARV.L. REV.713 (1985).
202. 494 U.S. 638, 649-50 (1990).
203. Skidmorev. Swift & Co., 323 U.S. 134, 139-40 (1944).
1992]
Executive Precedent
1023
analysis of the way the statuteshould be read, its views should be entitledto
some weight, even if does not have day-to-dayregulatoryauthorityover the
matter in contention. Adams Fruit is the wooden product of a conceptual
scheme that all too often forecloses potentiallyvaluablelines of inquiry,and
it should be rejected.
(b) Post hoc Rationalizations.In contrast,the Court'sdecision to exclude
"post hoc rationalizations"of counsel from Chevrondeference04is not consistent with the logic of Chevron.As long as the question concerns a matter
delegated to the agency, Chevronsuggests it should not matterhow or when
the agency announcesits interpretation.
The "posthoc rationalization"exception,however,conformsto pre-Chevron case law,205and can be explained by the executive precedentmodel. As
noted above, post hoc rationalizationsare simply not "precedents."They are
interpretationsfirst advancedby an agency in its capacityas a party litigant,
not in its capacity as an independentdecisionmaker.Thus, any deferenceto
agency views in this context would offend ideals of judicial impartiality.
(c) Canons of Construction.The Court has also suggested that certain
canonsof construction-most notably,the canonthatinterpretations
of doubtful
constitutionalityare to be avoided-may trumpthe Chevrondoctrine.20Chevron itself supplies no rationalefor such a holding.Underthe executive precedentmodel, in contrast,courtsmay consultany traditionaltool of interpretation
in arriving at an assessmentof the clarity of statutorymeaning. Canonsthat
serve as defaultrules in the absenceof more directevidence of congressional
intent should not play a majorpartin this exercise. If thereis no bettermeans
of resolving a questionof interpretationthan calling upon such a canon, then,
as a general rule, the executive interpretationshould not be set aside. On the
otherhand,canons groundedin considerationsof constitutionallaw and structure207
may supply a strongerbasis for overcomingexecutive precedent.
(d) Judicial Precedent.In Maislin Industries,U.S. v. PrimarySteel, Inc.,
the Courtheld that an agency cannot change its mind aboutthe meaningof a
statuteif by doing so it "casts doubton [the Court's]prior interpretation"of
the statute.208
This conclusion remains a mystery underChevron,but makes
perfect sense underthe executive precedentmodel. It is useful here to return
to the judicial analogy. Althoughone court of appeals shouldtreata decision
of anothercourtof appealsas primafacie correct,the precedentof the second
204. Bowen v. GeorgetownUniv. Hosp., 488 U.S. 204, 212 (1988) (quoting BurlingtonTruckLines
v. United States, 371 U.S. 156, 168 (1968)).
205. See, e.g., SEC v. Sloan, 436 U.S. 103, 117-18 (1978); InvestmentCo. Inst. v. Camp, 401 U.S.
617, 628 (1971); SEC v. CheneryCorp., 318 U.S. 80, 87-88 (1943).
206. EdwardJ. DeBartolo Corp. v. FloridaGulf Coast Bldg. & Constr.TradesCouncil, 485 U.S. 568,
574-75 (1988).
207. See, e.g., Gregoryv. Ashcroft, 111 S. Ct. 2395, 2403-06 (1991) (adoptingcanon that statutewill
not be construed to interfere with traditionalstate governmentalfunctions absent clear congressional
statementindicating intent that it be so applied).
208. Maislin Indus., U.S. v. PrimarySteel, Inc., 110 S. Ct. 2759, 2770 (1990).
1024
The Yale Law Journal
[Vol. 101: 969
court-however well reasonedandconsistentlyfollowed-cannot overcomethe
force of the court'sown precedent.209
Staredecisis prevailsover the deference
to
of
other
circuits.
given precedents
Similarly, an executive precedentcannot outweigh a directly applicable
judicial precedent.Of course, the executive view may providean occasion for
the courts to reconsidertheir own precedent,just as contrarydecisions from
other circuits may often trigger en banc reconsiderationby courts of appeals.210But such a change at best constitutesan appeal to the discretionof
the court to overrule its prior precedentunderthe criteria of stare decisis; it
does not impose an obligationon the courts to change their views or even to
engage in a reconsideration.Thus,the executiveprecedentrendersthe outcome
in Maislin perfectly explicable.
(e) Decisions ImplicatingAgency Jurisdiction.In Mississippi Power &
Light Co. v. Mississippiex rel. Moore,21JusticesBrennanand Scalia engaged
in an inconclusive debate over whether an exception to Chevronshould be
createdfor interpretationsthatimplicatethe limits of an agency'sjurisdiction.
This issue is particularlyill suited to sweeping, either/or solutions. A rule
precludingdeference on issues that implicate an agency's jurisdictionwould
accomplish indirectly what Cardoza-Fonsecasought to do directly:it would
eliminate any deferenceon "pure"questions of statutoryinterpretation.This
is becausealmostany purelylegal questionwill haveimplicationsfor the scope
of an agency's authority.At the same time, courts have long operatedwith a
kind of unwrittenunderstandingthat deferenceto agency views is not appropriate, at least not to the same degree, where the agency view is likely to be
influencedby institutionalbias.212
For example, courtsdo not defer to agency
views on whetheragency decisions are judiciallyreviewable,whetheragency
action falls within the termsof the FederalTortsClaimsAct, whetheragencies
are liable to pay attorney'sfees, or whetheragency documentsmust be disclosed underthe Freedomof InformationAct.213For similarreasons, at least
a lesser degree of deference may be warrantedwhen an agency decides to
regulatea new area:the decision to regulatemay be motivatedby designs for
agency aggrandizementratherthan by a disinterestedassessmentof statutory
authorityand appropriatepolicy.
The executive precedentmodel permits courts to give effect to concerns
aboutinstitutionalbias in determiningwhetherdeferenceis appropriate.In the
209. E.g., FederalSav. & LoanIns. Corp.v. Bonfanti,826 F.2d 1391, 1394 n.3 (5th Cir. 1987), vacated
sub nom. Zohdi v. Federal Sav. & Loan Ins. Corp., 490 U.S. 1001 (1989).
210. See FED.R. APP.P. 35.
211. 487 U.S. 354 (1988).
212. See Braun,supra note 106, at 1005-07; Sunstein,supra note 4, at 2101.
213. See, e.g., United States v. Fausto, 484 U.S. 439 (1988) (judicial review); Pierce v. Underwood,
487 U.S. 552 (1988) (attorney'sfees); CIA v. Sims, 471 U.S. 159 (1985) (Freedomof InformationAct);
United States v. S.A. Empresade Viacao Aerea Rio Grandense(Varig Airlines), 467 U.S. 797 (1984)
(FederalTorts Claims Act).
1992]
Executive Precedent
1025
most egregious cases of agency bias-for example, wherethe issue is whether
an agency decision is subjectto judicial review-there may well be no executive precedentsince the issue arises only when the agency is engagedin its role
as a judicial litigant. In the moredifficult case wherean agency has concluded
thatits statutorymandatepermitsa new industryor activity to be regulated(or
deregulated),the executiveprecedentmodelpermits-as is appropliate-a more
modulatedresponse.The concernwith possible agency bias justifies reducing
the degree of deference without disregardingagency views and arguments
altogether.In contrast,the rigid Chevrondoctrinedemandsthat courts either
defer or not defer,permittinga courtto disregardan agency's conclusionsonly
if it can identify a clear statutorymandatethat has been violated. Again, the
more flexible approachbetter serves the separationof powers concerns furthered by the practice of independentjudicial review.
All in all, the executive precedentmodel makes sense of a great deal of
learningpreviouslyregardedas an unconnectedhodgepodgeof factorswith no
unifying theme or rationale.The persistenceof these traditionalfactorsin the
face of an official doctrinethat is hostile to their survival suggests that there
is value in this conventionalwisdom. The congruencebetween the executive
precedentmodel and observationsof conventionalpracticeprovidespowerful
supportfor the model's validity.
C. Consequencesfor Judicial and Agency Performance
Finally, it is importantto give some considerationto the systemic effects
that the executive precedentmodel might have relative to the Chevronmodel.
Of course, in undertakingany such inquiry,one must bear in mind that the
Chevronmodel operatesquite differentlyin practicethan it does on paper.
1. Consequencesfor Courts
Proponentsof the Chevrondoctrineare attractedalmostexclusively by the
prospect that its formal or rulelike qualities will confine judicial discretion.
Restrictingjudicial discretionis thoughtto be desirablein this contextfor both
instrinsic and extrinsic reasons.
The intrinsic reason is that restrictionson judicial discretionare thought
to be necessary if we are to speak of courts as being bound by "law."214
Whateverthe theoreticalor practicalfailings of Chevron,or its discontinuities
with past (andpresent)judicialpractice,it at least has the appearanceof a legal
rule ratherthan an "all things considered"standard.Justice Scalia, probably
214. See Antonin Scalia, The Rule of Law as a Law of Rules, 56 U. CHI.L. REv. 1175 (1989). For
a general discussion, see FrederickSchauer,Formalism,97 YALEL.J. 509 (1988).
1026
The Yale Law Journal
[Vol. 101: 969
the foremost championof the Chevrondoctrine,215
is clearly attractedto this
featureof Chevron.Fromthis perspective,the executiveprecedentmodel is not
sufficiently rulelike: it importsso many factorsthat the practiceof deference
would degenerateinto an ad hoc inquirythat would allow courtsto reachany
result they wanted to. Moreover,with so many factorsfrom which to choose,
effective review or oversight by the SupremeCourt would be impossible.216
While thereis clearly some meritto this criticism,I would offer severalpoints
by way of response.
First, although my proposedexecutive precedentmodel is clearly more
open texturedthan Chevron,I do not think that it can fairly be describedas
"lawless." Unlike the pre-Chevronregime that it resembles, the executive
precedent model endorses neither ad hockery nor an open-ended "list of
factors"approach.Instead,it offers a theory thatexplains the respectiveroles
of the executive and judicial branches,tells us which deference factors are
relevant and why, and provides an account,in general terms at least, of how
the courtis to considerthe executive view relativeto its own assessmentof the
interpretativequestion.The modelmaybe complex-too complexto be reduced
to a simple Chevron-likeformula-but it is not "unprincipled."Nor can it be
said thatthe modelprescribesan ad hoc balancingtest. Suchtests requirecourts
to engage in a "head-to-headcomparison"of competing interests.217The
executive precedentmodel, in contrast,is groundedin analogical reasoning,
whereby various factorsbecome relevantin determiningwhethera particular
case approximatesa particularparadigm.In this sense, it is simply a variant
on what is perhapsthe most universaltool of legal reasoning.218
Second, to a considerableextent, the case for legal formalism is much
strongerwhererules of primarybehaviorare concernedthanit is when we are
dealing with rules of legal method.Whencourtsadopt"all things considered"
tests for the governanceof primarybehavior,serious questionscan be raised
about whether similarly situatedpersons are in fact treatedalike.219But it is
hardto see how the equalitynormis offended,at least to anythingapproaching
the same degree, when an issue of statutoryinterpretationis resolved in one
case using one set of interpretativetools and a different issue of statutory
interpretationis resolved in anothercase using other interpretativetools. As
long as the outcome of each exercise of statutoryinterpretationis applied
215. See NLRB v. United Food & CommercialWorkersUnion, 484 U.S. 112, 133-34 (1987) (Scalia,
J., concurring)(applaudingthe Courtfor abandoningthe limiting approachof Cardoza-Fonseca);Scalia,
supra note 39.
216. See Scalia, supra note 214, at 1178-82 (criticizing multifactortests for transformingquestions
of law into questions of fact and for insulatinglower courts from effective appellatereview).
217. T. Alexander Aleinikoff, ConstitutionalLaw in the Age of Balancing, 96 YALEL.J. 943, 945
(1987).
218. See EDWARD
H. LEVI,AN INTRODUCTION
TOLEGALREASONING
1-2 (1949).
219. Scalia, supra note 214. at 1178.
1992]
Executive Precedent
1027
consistently,the individualsprimarilyaffectedwill generallyperceivethatthey
are being treatedequally.
Third, the claim that Chevronallows the SupremeCourt to control the
behaviorof lower courts, whereasthe executive precedentmodel would not,
cannot be based on the Chevrondoctrineas we currentlyknow it-where the
two-step frameworkis ignoredas often as it is followed, and when followed,
is applied in several differentversions. But even if the Courtwere to adhere
to a "pure"version of Chevron,there would still be ample opportunitiesfor
manipulation,most prominentlythroughspuriousidentificationsof "specific
intentions"or "plain"statutorymeanings.Thus,even thoughChevronis more
formalisticthanthe executive precedentmodel, it too cannotconstrainwillfulness in the lower courts. To be sure, there is no basis for believing that the
executive precedentmodel would impose greater constraintson lower courts
than would a rigorously and consistentlyapplied Chevron.But I do think it
likely thatthe decision to deferunderthe executive precedentmodel would be
made in a more candid manner,with more revealingreasonsgiven in support
of judicial outcomes. The relevant choice, therefore, may not be between
constrainedand unconstrainedjudicial decisions, but between more or less
judicial candor.
Finally, I seriously doubt whetherit would ever be possible to decide all
deference questions without being drawn into some type of contextual or
multivariateinquiry,call it "pragmatism"
or "practicalreasoning"or whatever.220As Judge Stephen Breyer has observed, "thereare too many different
types of circumstances,includingdifferentstatutes,differentkinds of application, differentsubstantiveregulatoryor administrativeproblems,and different
legal posturesin which cases arrive, to allow 'proper'judicial attitudesabout
The
questions of law to be reducedto any single simple verbal formula."221
fact that the Courtis now creatingvarious ad hoc exceptions and alternative
formulationsof the Chevron doctrine suggests the beginnings of a process
whereby the apparentsimplicity of Chevronis tempered with a variety of
qualifyingrules. Whenwe startwith the two-stepChevronframeworkandthen
create these exceptions, we end up with a crazy patchworkthat lacks any
internalcoherence.Theexecutiveprecedentmodel,by admittinga widervariety
of variablesin a more contextualizedrelationshipat the outset, can avoid the
need for this kind of improvisedstructure.
The extrinsic reasonfor wantingto constrainjudicial discretionby formal
rules is that this would enhancethe ability of agencies to change the law. It is
no accidentthatmanyof the principaldefendersof Chevron-including Justice
220. See RICHARD
THEPROBLEMS
A. POSNER,
OFJURISPRUDENCE
286-309 (1990) (urgingpragmatic
approachto statutoryinterpretation);William N. Eskridge,Jr.& Philip P. Frickey,StatutoryInterpretation
as Practical Reasoning, 42 STAN.L. REV.319 (1990).
221. Breyer,supra note 175, at 373.
1028
The Yale Law Journal
[Vol. 101: 969
Scalia and Judges Starrand Silberman222-allserved in the first Reagan Administration, when an aggressively conservative executive branch sought
widespreadchange in the law and encounteredresistancefrom both Congress
and the judiciary.These defendersendorseChevronas a tool for translatingthe
electoral mandateof the Presidentinto legal reform,therebyovercomingthe
resistance of the other two branchesof government.
This rationale for Chevron,however,is obviously limited to a particular
set of historicalcircumstances-circumstancesthat have alreadybeen at least
in partsupersededby the appointmentof increasingnumbersof federaljudges
by PresidentsReaganandBush. Whatis neededis a formulathatstrikesa more
enduringbalancebetweenexecutive, legislative,andjudicialperspectives,and
between the forces of change and stability.
The executiveprecedentmodelenvisionsa largerrole for thejudiciarythan
does the original version of Chevron.However,as previouslyindicated,this
largerrole is necessary if courts are to performtheir traditionalfunction of
On the other hand, I do not think
checking administrativeabuses of power.223
that the discretionarynatureof the executive precedentmodel would cause
courts to engage in wholesale substitutionof judgmentfor agency interpretations. Otherrules groundedin judicial self-governance,most prominentlythe
understandingthat courts must follow precedentsof superiorcourts and the
doctrineof staredecisis, exert a powerfulinfluence over the mannerin which
courts resolve legal controversies. There is no reason why the deference
doctrine, understoodas a form of respect given to legal interpretationsby
anotherbranchof governmentof coordinateconstitutionalstatus, should not
also exert an importantinfluence over judicial behavior.
To be sure, by stressing the importanceof reliance factors such as longstandingandconsistentagencyinterpretation,contemporaneousinterpretation,
and congressional ratification, the executive precedentmodel is also more
resistant to rapid change in law than is Chevron.But again, I think this is
healthy.Chevronis a doctrinefor Jacobeans:the resultsof a single Presidential
election are a sufficient cause for wholesale modificationsin the law, with no
offsetting incentive for stabilityor protectionof relianceinterests.Surely,it is
ratherodd for Whiggishconservativeslike JusticeScalia-who speakglowingly
of legal formalism because it promotespredictabilityin the law,224and who
eloquently defend a strict separationof powers in order to prevent undue
concentrationsof power that pose a threatto liberty and private rights225-to
endorse Jacobeanmeasures.
222. See Scalia, supra note 39; Silberman,supranote44; Starr,supra note 34; see also JudicialReview
of AdministrativeAction in a ConservativeEra, 39 ADMIN.L. REV.353 (1987) (paneldiscussion including
Judge Starr and RichardWillard,a Justice Departmentofficial in the Reagan Administration).
223. See supra text accompanyingnotes 117-22.
224. Scalia, supra note 214, at 1179.
225. Morrison v. Olson, 487 U.S. 654, 697 (1988) (Scalia, J., dissenting).
1992]
Executive Precedent
1029
The executive precedent model would strike a better balance between
stabilityand change.By giving greaterdeferenceto longstanding,contemporaneous, and ratified agency interpretations,it would introducea bias in favor
of stability,and would give a measureof protectionto the relianceintereststhat
But this model would not rule out
these interpretationshave generated.226
reasons
for
that
the
change,provided
changeareclearlyarticulatedandthe new
with
is
consistent
the
interpretation
statutoryframework.Thus, it would not
"freeze"the law or precludelegal developmentin responseto changedperceptions of policy, whethercaused by Presidentialelections or otherwise.
2. Consequencesfor Agencies
When we turn from the consequencesfor judicial behaviorto the effects
on agency behavior,the balance sheet wholly favors the executive precedent
model. The Chevronframeworkessentially providesno incentives for agency
behavior.It focuses at both steps one and two almost exclusively on variables
relevant only to judicial functioning.The executive precedentmodel, in contrast,pays equal attentionto the caliberand qualityof the agency's precedent,
and by making the weight of the agency's precedentrelevant to the decision
to defer,providesincentivesfor agenciesto engage in certaintypes of behavior
ratherthan others.Three incentives,in particular,would be introducedby the
executive precedentapproach.
The first is the incentive to maintain a consistent position over time.
Consistency is desirableinsofar as it promotesequalityof treatmentbetween
similarlysituatedparties,protectsrelianceinterests,andrendersadministrative
actionmorepredictable,therebyallowingprivatepartiesto engagein meaningful planning.227
Consistency may also reduce "rentseeking" in the form of
of
expenditures resources designed to influence agencies to change their
interpretationsof the law.228
On the other hand, a rigid rule of "first in time, first in right" would
eliminate one of the advantages of using agencies to implement law: the
possibility of modifying policy over time as experience accumulates and
conditionsand publicperceptionschange.And a rule makingthe first interpretation controllingmight induce excessive expendituresto influence the initial
constructionof the law. The executive precedentmodel strikes an appropriate
226. See Monaghan,supra note 147, at 751-52 (practice of judicial adherenceto precedentfosters
conservative values).
227. It is no coincidence that these are some of the same values served by formal rules of law. See
Scalia, supra note 39. For traditionaladministrativelaw statementsendorsinggreateragency consistency,
& STEWART,
see, e.g., BREYER
supra note 192; 4 DAVIS,supra note 6, ? 20:11.
228. This point was suggested to me by my colleague Keith Hylton. See Keith Hylton, Doctrinal
Efficiency and LaborLaw (Aug. 1991) (unpublishedmanuscripton file with author)(discussingpossible
rent-seekingresponses to frequentchanges in National LaborRelations Board doctrine).
1030
The Yale Law Journal
[Vol. 101: 969
balance between these concernsby demandingeither agency consistency or a
persuasive explanationfor change as a condition for deference.
The second relevantincentive is the attemptto strive for harmonyamong
differentagencies. It is always difficultto maintaincoherencein oursprawling
bureaucraticstate. Variousmechanismsfor centralizedcontrolexist, including
and
preclearanceof regulationsby the Office of Managementand Budget229
of
the
Solicitor
General.230
But
the
preclearance judicial appealsby
centripetal
forces tend always to overpowerthe centrifugal.The idea thatagencyinterpretations that conflict with other agency interpretationswill lose a measureof
judicial deference adds anothersmall incentive for internalcoherence.
The third is the incentive to providereasonedexplanationsfor statutory
thatagency factfindingand policymaking
interpretations.It is "anomalous"231
is today subject to searchingreview underthe "hardlook" doctrine, which
requiresa carefulconsiderationof relevantfactorsand an answerto prominent
UnderChevron,however,agency interpretaobjections by affected parties.232
tions of law are subjectto automaticdeferenceunless the court finds that the
matterhas been clearlyresolvedby Congress.Paradoxically,therefore,official
doctrine requires more searchingjudicial review on questions of fact and
policy-where agenciesare traditionallythoughtto enjoya comparativeadvantange-than on questionsof law,wherethe traditionalview was thatcourtsheld
the edge.233
One of the promisingfeaturesof the executive precedentmodel is that it
would tend to incorporatesome of the featuresof "hardlook"review into the
process of consideringwhetherto give deferenceto executive interpretations
of statutes.The strengthof an agency precedentwould to a significantextent
be a functionof how carefullythe agencyconsideredthe questionof interpretation, respondedto objections,and offeredreasonsin supportof the interpretation adopted.In effect, an agency that has given a "hardlook" at a question
of interpretationwill be entitled to greaterdeferencethan an agency that has
given only cursoryconsiderationto the issue.
3. Consequencesfor Congress
Finally, it is worth speculating briefly about what sorts of incentives a
rigorously enforced Chevron would create for Congress. It is unlikely that
Congresswould standidly by in responseto a majorrealignmentin the division
of powers that enhanced the executive branch's ability to institute policy
229. As requiredby Exec. OrderNo. 12,291, 3 C.F.R. 127 (1981), reprintedin 5 U.S.C. ? 601 (1982).
See ChristopherC. DeMuth& Douglas H. Ginsburg,WhiteHouse Reviewof AgencyRulemaking,99 HARV.
L. REV. 1075 (1986).
230. See 28 C.F.R. ? 0.20(b) (1991).
231. Breyer,supra note 175, at 397.
232. See sources cited supra note 195.
233. See Diver, supra note 11, at 574.
1992]
Executive Precedent
1031
changesandminimizedthe role of the courtsin checkingadministrativeabuses.
One would expect some combinationof threeresponsesfrom Congress,none
of which in my view would be desirable.
congressionalresponsewouldbe to enact
Perhapsthe most straightforward
If
longer and more detailed statutes.234 the only things courts will enforce are
specific directives,then Congresswould try to supplyas many specific directives as possible. While some mighthail this developmentas a healthyreasserthere is
tion of the "legislative power" by its constitutionalinstrument,235
reason to believe that an upsurgein the specificity of legislation would lead
There can
to bad policy236and more influence by private interestgroups.237
be little doubt that it would add to the complexity-and incomprehensibility-of the legal system.
Anotherpossiblecongressionalresponsewouldbe to seek greaterinfluence
over administrativeagencies.This mighttake the formof directrestrictionson
Presidentialauthorityover agency heads. Morelikely, it would manifestitself
in the formof moreintensiveoversighthearingsandcontactsbetweencongresThe result, again, would be an erosion
sional staff and agency personnel.238
of coherence and greaterinterestgroup influence.239
Third, Congress might attemptto overrule Chevrondirectly,perhapsby
As previouslyindicated,
enacting some form of the BumpersAmendment.240
I think that Congresshas the constitutionalpower to directcourtsto adandon
the Chevronapproach.But such a responsewould very likely take the form of
an overreaction-as the BumpersAmendmentsurelywouldbe-and at the very
least would create a troubling precedent for congressional interventionin
matterstraditionallyleft to evolutionaryjudicial development.
Of course, the chance of these responsesoccurringwill be reducedif the
Courtcontinues to create exceptions to the Chevrondoctrineand to ignore it
altogether.Nevertheless,therecan be little doubtthat the executive precedent
approach-which is more consistent with historic practice and promises to
resultin a morebalancedjudicialrole-would provideeven greaterassurances
against undesirablecongressionalreactions.
234. Hertz, supra note 113, at 1675.
235. See, e.g., JOHNH. ELY,DEMOCRACY
AND DISTRUST131-33 (1980) (criticizing excessive
delegation of legislative authorityon democratictheory grounds);J. Skelly Wright,BeyondDiscretionary
Justice, 81 YALEL.J. 575, 584-86 (1972) (urgingrevival of nondelegationdoctrine).
236. See BRUCEA. ACKERMAN
& WILLIAM
T. HASSLER,
CLEANCOAL/DIRTY
AIR(1981) (arguing
that highly detailed provisions in 1977 Clean Air Act Amendmentsimpeded EPA's ability to develop
effective policy).
237. See JerryL. Mashaw,Prodelegation:WhyAdministratorsShouldMakePolitical Decisions, 1 J.L.
ECON.& ORGANIZATION
81 (1985) (public choice theory suggests that interestgroups will have greater
influence in legislative arena than in administrativearena).
238. See Lazarus,supra note 118.
239. See HaroldH. Bruff,LegislativeFormality,Administrative
Rationality,63 TEX.L. REV.207 (1984)
(discussing interest group influence exerted throughcongressionaloversight process).
240. See supra note 111.
1032
The Yale Law Journal
[Vol. 101: 969
VI. CONCLUSION
From the perspective of this Article, the long controversyover judicial
deferenceto administrativeinterpretationsof statutescan be seen as a rivalry
between two competing models: the mandatorydeference model and the
discretionary deference model. The rivalry has been largely unconscious
becausethe assumptionsof the mandatorydeferencemodel are often unstated,
and the theoreticalbasis for the discretionarydeferencemodel has neverbeen
previously articulated.Still, we can see these ideas strugglingfor dominance
in judicial practice.
In the pre-Chevronera, the two models coexisted uneasily.Courtsapplied
the mandatorydeferencemodelto interpretations
backedby expressdelegations
of regulatoryauthorityfromCongress.But in the absenceof an expressdelegation, they appliedsomethingthatin retrospectlooks very muchlike the discretionary deferencemodel-examining variouscontextualfactorsthat could be
appliedjust as easily to a precedentof a court of coordinatejurisdictionas to
the constructionof an executive branchagency. The Chevronera on paper
representsa rejectionof the discretionaryapproachand an embraceof a pure
mandatoryregime.Now, however,the realmof mandatorydeferencehas vastly
expandedto include a presumptionof delegationin all cases where a federal
statute is ambiguous or unclear.As we have seen, however, in practice the
discretionaryapproachhas lived on, in the shadowsof Chevron,and in considerable tension with its expandeddelegationtheory.
Chevron's many failings are largely those of the mandatorydeference
model that it incorporates.Althoughthe idea of expressdelegationis straightforwardenough,the conceptof an implieddelegationof interpretativeauthority
is difficultto characterizeas a congressionalcommand.The focus on delegation
also makes the decision about whetherto defer subordinateto the resolution
of a questionof judicialinterpretation:
Did Congress(constructively)delegate,
or did it not? In this fashion, considerationof the executive view becomes
ensnaredin controversiesover the legitimate scope of judicial interpretation.
Finally, the idea of delegationrequiresthat deferencebe viewed in either/or
terms:eitherCongressdelegatedauthority,or it did not; no intermediatepossibilities make sense. The either/ornatureof the mandatorydeference theory
based on delegated power pushes the doctrinetowardextremes of either too
little or too much deference.
I have argued that a better solution to the uneasy mixture of the
pre-Chevronperiod would have been to move to a pure discretionaryregime,
whatI havecalled the executiveprecedentmodel.The idea of delegationwould
not disappearin such a regime. It would simply revertto its original scope of
express delegation, and it would become one of several factors that courts
would examine in determiningthe "strength"of an executive precedent.Becausethe practiceof deferenceunderthe executiveprecedentmodelis grounded
1992]
Executive Precedent
1033
in a conception of the judicial power ratherthan in a fiction of delegation,it
avoids the major theoreticalfailings of the Chevrondoctrine. In addition,it
allows courtsto considervariousattributesof the executive decisionthatseem
undeniablyrelevant to the question of how much weight it should be given:
such as whetherthe agency's interpretationis supportedby carefulreasoning,
whetherits interpretationhas generatedrelianceintereststhatshouldbe protected, andwhetherits constructionhas been approvedby Congress.The executive
precedentmodel also permitsa flexible responseto the questionof deference,
allowingcourtsto give variousdegreesof deferenceto executiveinterpretations,
dependingon the strengthof the contextualfactorsinvolved and on the court's
conviction about the clarity of the statuteas an original matter.
Like its close cousin, the doctrine of judicial precedent,the model of
executive precedentdoes not admitof easy or pat answers.But this does not
mean it has nothing to say about what variablesare relevant to its operation
and why they are significant. In this respect,one of the significantbenefits of
the executive precedentmodel is that it might teach us somethingabout the
practiceof followingjudicialprecedent,of whichlegal academicshaveunfortunately had far too little to say.
[Vol. 101: 969
The Yale Law Journal
1034
APPENDIX
Agency Deference Cases in the Supreme Court
1981-1990 Terms
Case
Chevron
Agency
Step 1
Interpretion Framework or
Accepted? Followed? Step 2
Traditional
Factors
1990 TERM
1. Pauley v. Bethenergy Mines,
Inc., 111 S. Ct. 2524 (1991)
Y
Y
2. Gregory v. Ashcroft,
N
N
N/A
None
3. LittonFin. PrintingDiv.
v. NLRB, 111 S. Ct. 2215 (1991)
Y
N
N/A
Expertise
4. Rustv. Sullivan,
111 S. Ct. 1759 (1991)
Y
Y
S2
Longstanding
Well-reasoned
5. AmericanHosp. Ass'n v.
NLRB, 111 S. Ct. 1539 (1991)
Y
Y
S1
Well-reasoned
6. CottageSav. Ass'n v.
Commissioner,111 S. Ct. 1503 (1991)
Y
N
N/A
Longstanding
Ratification
7. Norfolk& WesternRy. Co. v. American
TrainDispatchersAss'n,
111 S. Ct 1156 (1991)
Y
Y
S1
None
8. International
Union,UAWv. Johnson
Controls,Inc., 111 S. Ct. 1196 (1991)
Y
N
N/A
None
9. EEOCv. ArabianAm. Oil Co.,
111 S. Ct. 1227 (1991)
N
N
N/A
Legis./Interp.
S2
Expertise
Longstanding
111 S. Ct. 2395 (1991)
Longstanding
Contemporaneous
Well-reasoned
10. Demarest v. Manspeaker,
111 S. Ct. 599 (1991)
N
Y
S1
Ratification
11. Mobil Oil Exploration & Producing
Southeast Inc. v. United Dist. Cos.,
Y
Y
S1
None
N
Y
S1
None
Y
Y
S2
None
111 S. Ct. 615 (1991)
1989 TERM
12. Maislin Indus., U.S. v.
Primary Steel, Inc.,
110 S. Ct. 2759 (1990)
13. PensionBenefit Guar.Corp.v. LTV
Corp., 110 S. Ct. 2668 (1990)
1992]
1035
Executive Precedent
14. Sullivanv. Stroop,
110 S. Ct. 2499 (1990)
Y
Y
S1
None
15. Fort StewartSch. v. FederalLabor
RelationsAuth.,495 U.S. 641 (1990)
Y
Y
S1
None
16. Davis v. United States,
495 U.S. 472 (1990)
Y
N
N/A
Longstanding
Contemporaneous
Ratification
17. Departmentof Treasuryv. Federal
LaborRelationsAuth.,494 U.S. 922 (1990)
N
Y
S1
None
18. NLRBv. CurtinMathesonScientific,
Inc., 494 U.S. 775 (1990)
Y
N
N/A
Longstanding
19. AdamsFruitCo. v. Barrett,
494 U.S. 638 (1990)
N
Y
S1
None
20. Crandonv. United States,
494 U.S. 152 (1990)
N
N
N/A
None
21. Sullivanv. Everhart,
494 U.S. 26 (1990)
Y
Y
S2
None
22. Dole v. UnitedSteelworkers,
494 U.S. 26 (1990)
N
Y
S1
None
23. Sullivanv. Zebley,
493 U.S. 521 (1990)
N
Y
S1
Ratification
24. United Statesv. GoodyearTire&
RubberCo., 493 U.S. 132 (1990)
Y
Y
N/A
None
25. Chesapeake& Oh. Ry. Co.
v. Schwalb,493 U.S. 40 (1990)
y
N
N/A
Longstanding
26. PublicEmployeesRetirementSys.
v. Betts, 492 U.S. 158 (1989)
N
Y
S1
Contemporaneous
27. PublicCitizenv. UnitedStatesDept.
of Justice,491 U.S. 440 (1989)
N
N
N/A
Contemporaneous
28. MeadCorp.v. Tilley,
490 U.S. 714 (1989)
Y
Y
S2
Longstanding
29. Robertsonv. MethowValleyCitizens
Council,490 U.S. 332 (1989)
Y
N
N/A
Longstanding
30. Marshv. OregonNaturalResources
Council,490 U.S. 360 (1989)
Y
N
N/A
None
31. Massachusettsv. Morash,
490 U.S. 107 (1989)
y
N
N/A
Longstanding
32. Coit IndependenceJointVenture
v. FederalSav. & LoansIns. Corp.,
489 U.S. 561 (1989)
N
N
N/A
None
1988 TERM
Well-reasoned
1036
[Vol. 101: 969
The Yale Law Journal
33. Bowen v. Georgetown Univ. Hosp.,
488 U.S. 204 (1989)
N
N
N/A
Longstanding
34. Pittston Coal Group v. Sebben,
488 U.S. 105 (1989)
N
Y
S1
None
35. Communications Workers v.
N
N
N/A
Longstanding
36. Mississippi Power & Light Co. v. Mississippi
ex rel. Moore, 487 U.S. 354 (1988)
Y
N
N/A
None
37. Huffman v. Western Nuclear, Inc.,
486 U.S. 663 (1988)
Y
N
N/A
38. K Mart Corp. v. Cartier, Inc.,
486 U.S. 281 (1988)241
Y/N
Y
S2/S1
None
39. FERC v. Martin Exploration
Management Co.,
486 U.S. 204 (1988)
Y
N
N/A
None
40. EEOC v. Commercial Office Prods. Co.,
486 U.S. 107 (1988)
Y
N
N/A
None
41. Edward J. DeBartolo Corp. v. Florida
Gulf Coast Bldg. & Constr. Trades Council,
N
N
N/A
None
Y
N
N/A
None
43. Traynor v. Turnage,
485 U.S. 535 (1988)
Y
N
N/A
Longstanding
Ratification
44. Federal Labor Relations Auth. v.
N
N
N/A
None
45. Bethesda Hosp. Ass'n v. Bowen,
485 U.S. 399 (1988)
N
Y
S1
None
46. Etsi Pipeline Project v. Missouri,
484 U.S. 495 (1988)
N
Y
S1
None
47. Honig v. Doe,
484 U.S. 305 (1988)
Y
Y
S2
None
48. NLRB v. United Food & Commercial
Y
Y
S2
Longstanding
N
N/A
None
1987 TERM
Beck, 487 U.S. 735 (1988)
Post hoc
rationalization
485 U.S. 568 (1988)
42. Regents of the Univ. of Cal. v.
Public Employment Relations Bd.,
485 U.S. 589 (1988)
Aberdeen Proving Ground, 485 U.S. 409 (1988)
Workers Union, Local 23,
484 U.S. 112 (1987)
1986 TERM
49. Utah Div. of State Lands v.
N
United States, 482 U.S. 193 (1987)
241. For purposes of tabulation,this case has been counted as "Y" and "S2."
1992]
1037
Executive Precedent
50. Bowen v. Yuckert,
482 U.S. 137 (1987)
Y
N
N/A
Ratification
51 Fall RiverDyeing & FinishingCorp.
v. NLRB,482 U.S. 27 (1987)
Y
N
N/A
None
Bhd. of Elec.
52. NLRBv. International
Workers,Local 340, 481 U.S. 573 (1987)
N
N
N/A
None
53. Lukhardv. Reed,
481 U.S. 368 (1987)
Y
N
N/A
None
54. INS v. Cardoza-Fonseca,
480 U.S. 421 (1987)
N
Y
S1
Longstanding
55. Clarkev. SecuritiesIndus.Ass'n,
479 U.S. 388 (1987)
Y
Y
S2
None
56. Wrightv. City Of Roanoke
Redev.& Hous.Auth.,
479 U.S. 418 (1987)
Y
N
N/A
None
57. AnsoniaBd. of Educ.v. Philbrook,
479 U.S. 60 (1986)
N
N
N/A
None
58. CommodityFuturesTradingComm'n
v. Schor,478 U.S. 833 (1986)
Y
N
N/A
Expertise
Longstanding
Ratification
59. Local 28 of Sheet MetalWorkers'
Int'l Ass'n v. EEOC,
478 U.S. 421 (1986)
Y
N
N/A
Longstanding
Contemporaneous
60. JapanWhalingAss'n v. American
CetaceanSoc'y, 478 U.S. 221 (1986)
Y
Y
S2
None
61. UnitedStatesDep't Of Transp.
v. ParalyzedVeteransof Am.,
477 U.S. 597 (1986)
Y
Y
S1
Longstanding
62. Atkinsv. Rivera,
477 U.S. 154 (1986)
Y
N
N/A
None
63. MeritorSav. Bankv. Vinson,
477 U.S. 57 (1986)
Y
N
N/A
None
64. Youngv. CommunityNutritionInst.,
476 U.S. 974 (1986)
Y
Y
S2
Longstanding
Ratification
65. Bowen v. AmericanHosp. Ass'n,
476 U.S. 610 (1986)
N
N
N/A
None
66. FTCv. IndianaFed'n of
Dentists,476 U.S. 447 (1986)
Y
N
N/A
None
67. FDICv. PhiladelphiaGearCorp.,
476 U.S. 426 (1986)
Y
N
N/A
Longstanding
Ratification
1985TERM
1038
[Vol. 101: 969
The Yale Law Journal
68. United States v. City of Fulton,
475 U.S. 657 (1986)
Y
Y
S2
None
69. NLRB v. Financial Inst. Employees,
Local 1182,
N
N
N/A
None
N
Y
S1
None
Y
Y
S2
Ratification
N
N/A
Expertise
475 U.S. 192 (1986)
70. Board of Governors of the Fed.
Reserve Sys. v. Dimension Fin. Corp.,
474 U.S. 361 (1986)
71. United States v. Riverside
Bayview Homes, Inc.,
474 U.S. 121 (1985)
1984 TERM
72. Pattern Makers' League of N. Am.
v. NLRB, 473 U.S. 95 (1985)
Y
73. Jean v. Nelson, 472 U.S. 846 (1985)
Y
N
N/A
None
74. Cornelius v. Nutt,
472 U.S. 648 (1985)
Y
N
N/A
None
75. Central States S.E. & S.W. Areas
Y
N
N/A
None
76. Western Air Lines v. Criswell,
472 U.S. 400 (1985)
Y
N
N/A
Longstanding
77. Mountain States Tel. & Tel. Co.
Y
N
N/A
Longstanding
Longstanding
Pension Fund v. Central Transp.,
Inc., 472 U.S. 559 (1985)
v. Pueblo of Santo Ana,
472 U.S. 237 (1985)
Contemporaneous
78. Lowe v. SEC, 472 U.S. 181 (1985)
N
N
N/A
None
79. Connecticut Dep't of Income
Maintenance v. Heckler,
471 U.S. 524 (1985)
Y
N
N/A
Longstanding
80. INS v. Rios-Pineda,
471 U.S. 444 (1985)
Y
N
N/A
None
81. Tony & Susan Alamo Found. v.
Secretary of Labor,
471 U.S. 290 (1985)
Y
N
N/A
Longstanding
82. United States v. Locke,
471 U.S. 84 (1985)
Y
N
N/A
None
83. Bennett v. Kentucky Dep't of Educ.,
470 U.S. 656 (1985)
Y
N
N/A
None
84. Heckler v. Turner,
470 U.S. 184 (1985)
Y
N
N/A
Longstanding
Ratification
85. NAACP v. Hampton County Election
Comm'n, 470 U.S. 166 (1985)
Y
N
N/A
None
1992]
Executive Precedent
1039
86. ChemicalMfrs. Ass'n v.
NationalResourcesDefense Council
470 U.S. 116 (1985)
Y
Y
S2
Longstanding
87. NLRBv. Action Automotive,Inc.,
469 U.S. 490 (1985)
Y
N
N/A
None
88. Alexanderv. Choate,
469 U.S. 287 (1985)
y
N
N/A
None
89. LawrenceCountyv. Lead-Deadwood
Sch. Dist., 469 U.S. 256 (1985)
Y
N
N/A
None
90. UnitedStatesv. Boyle,
469 U.S. 241 (1985)
Y
N
N/A
None
91. IrvingIndep,Sch. Dist. v. Tatro,
468 U.S. 883 (1984)
Y
N/A
N/A
None
92. SecuritiesIndus.Ass'n v. Board
of Governorsof Fed. ReserveSys.,
468 U.S. 207 (1984)
Y
N/A
N/A
Longstanding
Legis./Interp.
93. SecuritiesIndus.Ass'n v. Board
of Governorsof Fed. ReserveSys.,
468 U.S. 137 (1984)
N
N
N/A
Longstanding
Post hoc
rationalization
94. Sure-tan,Inc. v. NLRB,
467 U.S. 883 (1984)
Y
N/A
N/A
None
95. ChevronU.S.A., Inc. v.
NationalResourcesDefenseCouncil
467 U.S. 837 (1984)
Y
Y
S2
Expertise
96. UnitedStatesv. Morton,
467 U.S. 822 (1984)
Y
N/A
N/A
Legis./nterp.
97. SEC v. JerryT. O'Brien,Inc.,
467 U.S. 735 (1984)
Y
N/A
N/A
Longstanding
98. CapitalCities Cable,Inc. v.
Crisp,467 U.S. 691 (1984)
Y
N/A
N/A
None
99. AluminumCo. of Am. v. Central
LincolnPeoples'Util. Dist.,
467 U.S. 380 (1984)
Y
N/A
N/A
Expertise
Contemporaneous
100. ICCv. AmericanTruckingAss'ns,
Inc., 467 U.S. 354 (1984)
Y
N/A
N/A
None
101. EscondidoMut.WaterCo. v. La Jolla
Band of MissionIndians,
466 U.S. 765 (1984)
N
N/A
N/A
Longstanding
102. NLRBv. City DisposalSys., Inc.,
465 U.S. 822 (1984)
Y
N/A
N/A
Longstanding
Expertise
1983Term
Ratification
1040
[Vol. 101: 969
The Yale Law Journal
103. Consolidated Rail Corp. v.
Darrone, 465 U.S. 624 (1984)
Y
N/A
N/A
Contemporaneous
Ratification
104. NLRB v. Bildisco & Bildisco,
465 U.S. 513 (1984)
N
N/A
N/A
Expertise
105. Grove City College v. Bell,
465 U.S. 555 (1984)
Y
N/A
N/A
Longstanding
Ratification
106. Dickman v. Commissioner,
465 U.S. 330 (1984)
Y
N/A
N/A
None
107. Commissioner v. Engle,
464 U.S. 206 (1984)
N
N/A
N/A
None
108. INS v. Phinpathya,
464 U.S. 183 (1984)
N
N/A
N/A
None
109. Bureau of Alcohol, Tobacco &
Firearms v. FLRA,
N
N/A
N/A
None
464 U.S. 89 (1983)
1982 TERM
110. Rice v. Rehner,
463 U.S. 713 (1983)
Y
N/A
N/A
Contemporaneous
111. Guardians Ass'n v. Civil Serv.
Y
N/A
N/A
Longstanding
Comm'n of N.Y.,
463 U.S. 582 (1983)
Contemporaneous
112. Belknap, Inc. v. Hale,
463 U.S. 491 (1983)
N
N/A
N/A
None
113. Public Serv. Comm'n v.
N
N/A
N/A
None
Y
N/A
N/A
Longstanding
115. NLRB v. Transportation Management
Corp., 462 U.S. 393 (1983)
Y
N/A
N/A
Longstanding
116. Bell v. New Jersey,
461 U.S. 773 (1983)
Y
N/A
N/A
None
117. Bill Johnson's Restaurants, Inc.
N
N/A
N/A
Expertise
Mid-Louisiana Gas,
463 U.S. 319 (1983)
114. National Ass'n Of Greeting
Card Publishers v. United States
Postal Serv., 462 U.S. 810 (1983)
v. NLRB, 461 U.S. 731 (1983)
118. Morrison-Knudsen Constr. Co. v.
Longstanding
Y
N/A
N/A
Longstanding
Y
N/A
N/A
Ratification
Director, Office of Workers'
Compensation Programs,
461 U.S. 624 (1983)
119. Bob Jones Univ. v. United States,
461 U.S. 574 (1983)
1992]
Executive Precedent
1041
120. Heckler v. Campbell,
461 U.S. 458 (1983)
Y
N/A
N/A
Legis./Interp.
121. American Paper Inst, Inc. v. American
Elec. Power Serv. Corp.,
Y
N/A
N/A
Contemporaneous
122. Commissioner v. Tufts,
461 U.S. 300 (1983)
Y
N/A
N/A
Contemporaneous
123. Metropolitan Edison Co. v. NLRB,
460 U.S. 693 (1983)
Y
N/A
N/A
None
124. Bowsher v. Merck & Co.,
460 U.S. 824 (1983)
N
N/A
N/A
Longstanding
125. Fidelity Fed. Sav. & Loan Ass'n v.
De La Cuesta, 458 U.S. 141 (1982)
Y
N/A
N/A
Longstanding
126. Schweiker v. Hogan,
457 U.S. 569 (1982)
Y
N/A
N/A
None
127. Blum v. Bacon, 457 U.S. 132 (1982)
Y
N/A
N/A
None
128. North Haven Bd. of Educ. v. Bell,
456 U.S. 512 (1982)
Y
N/A
N/A
Longstanding
Ratification
461 U.S. 402 (1983)
1981 TERM
Interagency
disagreement
129. Herwig v. Ray,
455 U.S. 265 (1982)
Y
N/A
N/A
Legis./Interp.
130. United States v. Vogel Fertilizer
Co., 455 U.S. 16 (1982)
N
N/A
N/A
Longstanding
Ratification
Legis./Interp.
131. United States v. Clark,
454 U.S. 555 (1982)
Y
N/A
N/A
Longstanding
132. Charles D. Bonnano Linen Serv.
Y
N/A
N/A
Longstanding
Well-reasoned
N/A
N/A
Longstanding
v. NLRB, 454 U.S. 404 (1982)
133. NLRB v. Hendricks County Rural
Elec. Membership Corp.,
V
454 U.S. 170 (1981)
Ratification
134. Ridgway v. Ridgway,
454 U.S. 46 (1981)
V
N/A
N/A
None
135. Federal Election Comm'n v.
V
N/A
N/A
Longstanding
Well-reasoned
Democratic Senatorial Campaign
Comm., 454 U.S. 27 (1981)
Expertise