The Myth of the Federal Private Nondelegation Doctrine, Part 2

Serial-blogging my recent article in the Notre Dame Law Review

Yesterday, I started serial-blogging my article, The Myth of the Federal Private Nondelegation Doctrine, which has just come out in the Notre Dame Law Review. I'll continue serial-blogging it here over the next couple of days. This is a timely issue, because of the horseracing case currently pending in the Fifth Circuit (in which I filed an amicus brief on behalf of the Reason Foundation and others). Here's Part I, explaining the variety of different delegation-related doctrines and providing a handy categorization. (Please be sure to refer to the real version if you want all the footnotes!)

*     *     *

I. Our Many Nondelegation Doctrines

Nondelegation is easy to get wrong because there's more than one nondelegation doctrine.  Everyone knows about the classic doctrine—the one that's usually called the Nondelegation Doctrine, which derives from Article I's Vesting Clause.  But other doctrines also have implications for delegations, and the Supreme Court and others sometimes talk about them using the word "delegation."  This isn't wrong: these doctrines really are relevant to delegations.  But we shouldn't confuse these similar-sounding doctrines.

Here's a way to illustrate the different doctrines:

• Consider Gary Lawson's "Goodness and Niceness Commission" hypo. Congress passes a Goodness and Niceness Act, where section 1 outlaws transactions not promoting goodness and niceness, and section 2 lets the Commission promulgate regulations defining the content of the statute.  The problem is from the giver's side: Congress has given up too much power.
• Now suppose Congress creates specialized Article III courts and gives them jurisdiction over cases where plaintiffs lack standing, or that fall outside the Article III jurisdictional categories. Congress hasn't given up too much power; perhaps the set of cases is very narrowly defined (and maybe even there would be no problem with those cases being heard by non–Article III administrative tribunals).  The problem is that Article III courts are constitutionally prohibited from hearing such cases; i.e., the recipient lacks the power to act.
• Now imagine Congress tells an agency how to conduct adjudications. (The guidance is detailed, so Congress hasn't given up too much power, and the adjudications are within the agency's powers.)  Under that guidance, the parties are deprived of liberty or property interests with no notice or procedural rights.  The problem is that, for rights/‌justice/‌fairness reasons, individuals' rights under the Due Process Clause are being violated, i.e., the use of the delegated power is unconstitutional.  The same would be true with any other due process issue, like if the adjudicators received bonuses when they ruled against a claimant.  More generally, something about the application of the delegated powers would be improper.

This taxonomy divides nondelegation rules into three categories: (1) giver-based, i.e., don't delegate power that you can't grant; (2) recipient-based, i.e., don't delegate power to entities that can't exercise it; and (3) application-based, i.e., don't delegate power where the circumstances of its application will be unjust.  (Perhaps this third category is a residual category for everything not fitting into the other two.)  This Part looks at each in turn.

A. Giver-Based

1. The Article I Nondelegation Doctrine

The main giver-based nondelegation doctrine is the classic Nondelegation Doctrine stemming from Article I's Vesting Clause—"[a]ll legislative Powers herein granted shall be vested in a Congress of the United States"—which holds that Congress can't give up legislative power.  The caselaw elaborates on this principle in several ways:

• Because the doctrine stems from Article I, it applies only to congressional delegations, not state delegations—though many states have constitutional nondelegation doctrines of their own.
• Delegations of power aren't wrong as such—only delegations of legislative power. And these are defined not formalistically (i.e., as the power to vote on laws in legislatures) but functionally, as the power to make law in an unconstrained way.  Accordingly, delegations don't cross the line into legislative delegations unless Congress has failed to provide an "intelligible principle" to guide the delegate's discretion.
• The presence of (congressionally mandated) procedures and judicial review can prevent a violation of this doctrine, because if an agency can't act without following particular procedures, and can get reversed for exceeding its authority, that constrains its scope of action.
• This doctrine applies to any delegate. While most Nondelegation Doctrine cases concern executive agencies, others concern the judiciary, Indian tribes, or even (as discussed below) private parties; and (except as discussed below) the doctrine doesn't differ depending on who is involved.
• When the delegate already has inherent authority over the subject matter, the intelligible principle doctrine is weakened or dropped entirely. (I've dubbed this the "Inherent-Powers Corollary.")  This applies, for instance, when Congress delegates to the executive branch in areas of foreign affairs or military governance, to Indian tribes in matters of regulation of Indian country, or to courts in certain procedural areas, or when Congress dynamically incorporates state law in various areas.
• Some commentators have argued that delegations should be scrutinized more carefully when the delegate fares poorly on a menu of "functional considerations," such as "relative expertise, accountability, flexibility, accessibility, and . . . ability to achieve uniformity."

The doctrine isn't explicitly used often—the Supreme Court has applied it only twice, both times in 1935—though it operates behind the scenes, in constitutional avoidance decisions or delegation-constraining administrative doctrines like Chevron minimalism or the major questions doctrine.  And Gundy v. United States suggests that the doctrine might be revived in the near future (and perhaps that the specific "intelligible principle" formulation might be replaced with something else).

Nonetheless, for the moment, the test is whether Congress has provided an "intelligible principle" to accompany the delegation; and so, whenever a delegation is problematic under this theory, a solution would be "Just narrow the delegation."

2. Some Other Giver-Based Doctrines

Other, less famous, giver-based doctrines have also shown up in cases or scholarship:

• The void-for-vagueness doctrine is rooted in due process, not Article I—thus applying equally to state laws. (A version of this doctrine is also rooted in the First Amendment.)  This doctrine is also aimed at avoiding delegating standardless authority to the police, judges, and juries.
• The Constitution lets Congress "define and punish . . . Offences against the Law of Nations." Can Congress delegate that power, or must any definition of new offenses come from Congress?  When the Alien Tort Statute grants federal courts jurisdiction over "any civil action by an alien for a tort only, committed in violation of the law of nations," does this let federal courts develop an evolving federal common law of law-of-nations torts?  Can Congress authorize military tribunals to develop new war crimes?  Perhaps the Offenses Clause makes this an exclusive congressional power, not available to military tribunals or federal courts.
• The Constitution lets Congress create lower federal courts. Does this prevent Congress from delegating to courts the power to create other courts, as when judicial councils create bankruptcy appellate panels, or as with proposals to similarly allow "judicial councils to create district court appellate panels"?
• The Constitution provides that "[e]ach State shall appoint, in such Manner as the Legislature thereof may direct, a Number of [presidential] Electors" and that "[t]he Times, Places and Manner of holding [congressional] Elections . . . shall be prescribed in each State by the Legislature thereof." Does this prevent a state legislature from delegating the elector-appointing power or the district-drawing power to state courts?  The theory of the Bush v. Gore concurrence, or of the petitioners in Moore v. Harper, suggests that these state legislative powers would be nondelegable, though the Supreme Court has rejected a strong version of this theory.

B. Recipient-Based: The Appointments Clause and More

The second class of nondelegation doctrines is recipient-based: even if Congress can delegate a power and is suitably specific, the recipient can't exercise that power.  Whenever a delegation is problematic under this theory, a solution would be to "just delegate to someone else."

Many separation-of-powers doctrines can be thought of as recipient-based nondelegation doctrines (though they're often not presented with the "delegation" label):

• Congress can't delegate significant federal authority to someone, unless that someone is appointed according to the rules governing officers of the United States (i.e., the Appointments Clause) and is properly removable. (But Congress could delegate that same power to someone else who was properly appointed and removable.)
• One could similarly argue that a private attorney general, like a qui tam relator, can't validly get the power to enforce federal law (though Congress could delegate that same power to an actual prosecutor).
• Congress can't delegate certain types of adjudicative power to non–Article III tribunals (but could delegate it to Article III courts).
• Congress can't delegate power to federal courts to hear cases where the plaintiffs lack standing—and, relatedly, can't delegate prosecutorial power to would-be plaintiffs who lack standing. (But it could delegate such power to non–Article III tribunals.)
• Congress might not be able to delegate to Article III courts powers that aren't "appropriate to the central mission of the Judiciary," and might not be able to delegate to any branch the power to appoint inferior officers where there's some "incongruity" between the functions of the appointing branch and the functions of the appointed officer. (But it could delegate to a different party where the incongruity would not be present.)
• Congress can't delegate power to one (or both) of its own houses that would allow that house to alter the rights or responsibilities of actors outside of Congress (though it could delegate that power to an agency).

More generally, the "separation of powers principle," which prevents aggrandizement of one branch at the expense of another, could be thought of as recipient-based.

Mistretta v. United States, which considered the constitutionality of the U.S. Sentencing Commission, illustrates how the Supreme Court has distinguished between giver-based and recipient-based theories.  The Court considered whether the statute contained an intelligible principle that would satisfy the Article I Nondelegation Doctrine—but then it also had a separate section discussing whether the delegated power could be wielded by the Sentencing Commission, an entity within Article III.  The Court answered yes to both questions; on the second question, Justice Scalia, in dissent, wrote that delegation to "all manner of 'expert' bodies" would be bad "not because of the scope of the delegated power, but because its recipient is not one of the three Branches of Government."  Everyone on the Mistretta Court understood that two different sorts of delegation problems were at play.

Not that Article I nondelegation and recipient-based theories can't talk to each other: one can readily imagine a recipient-based theory nested within (or closely related to) a giver-based theory.  For instance, what if the Article I Nondelegation Doctrine limits Congress in its ability to delegate generally, but absolutely bars its ability to delegate to particular disfavored recipients?  An Article I private nondelegation theory (which would prohibit, among other things, dynamic delegations of standard-setting power to private expert organizations) would be an example of such a theory.  So would be a view that the federal government can't delegate to "foreign" governments (including not only other countries but also municipal governments or other state governments, or the federal government if the delegator is a state government), for instance by dynamically incorporating another government's law.  Some delegates might be acceptable (the theory would say), but these aren't, because they can't exercise the power they're asked to exercise.  Such theories could even rule out delegations back to the people, in the form of plebiscites or referenda.

Recipient-based theories might be based on "political accountability" rationales—a government can't delegate to anyone it can't control or that can't be removed by the (relevant jurisdiction's) people at the next election, or that are beyond the reach of the (relevant jurisdiction's) checks and balances.

Or a recipient-based theory could be based on the idea that only proper elements of the federal government can be proper recipients of federal delegations, because nobody else is allowed to exercise federal power.  Justice Scalia's Mistretta dissent has that flavor: there, he argued that "the power to make law cannot be exercised by anyone other than Congress, except in conjunction with the lawful exercise of executive or judicial power."  In Justice Scalia's view, delegations to the executive or judiciary could be valid because "a certain degree of discretion, and thus of lawmaking, inheres in most executive or judicial action"—but the U.S. Sentencing Commission had no functions other than rulemaking ("a sort of junior-varsity Congress"), so there was no excuse for delegation in that case.

Recipient-based theories also show up outside of administrative law:

• A legislature can't write sentencing rules that delegate to a judge the power to find facts that increase a sentence beyond the maximum that would be justified based on facts found only by a jury—this would violate the criminal jury trial right, though delegating the same factfinding power to the jury itself would be fine.
• A legislature can't delegate a zoning-type power to churches to veto the licensing of bars—this would violate the Establishment Clause if churches were singled out as preferred recipients of the power, though it could delegate the same zoning power to a broader group of landowners that happened to include churches.

C. Application-Based: The Due Process Clause

Finally, there's the third category of nondelegation challenges.  I call these "application-based," because the power is one that the giver can give and that the recipient can exercise, but something about the application or use of the power is unconstitutional.

Not all rights-related issues fall into this category: we've seen that some jury-trial-right and Establishment Clause issues can be considered recipient-based.  This is a residual category, encompassing whatever doesn't fall within the other two.  Often the problem is due process—for instance, the doctrine preventing people from wielding coercive power when they have bias, especially financial bias.

1. Due Process and the Structure of Delegations

Due process usually constrains specific actions that any governmental actor might take.  For instance: Has the government deprived me of something without notice or a hearing?

This is typically unrelated to the structure of delegations: if a statute permitted an agency to deprive me of liberty without due process, the agency could conceivably cure the problem by voluntarily adopting adequate procedures.  This is not the case with giver- or recipient-based doctrines: A delegate's own voluntary actions can't enlarge Congress's power to delegate or give itself a power it doesn't have.

But sometimes, due process is relevant to the structure of delegations.

First, suppose a legislature passes a statute requiring an agency to use specific (inadequate) procedures when depriving people of certain interests.  The offending procedures are statutory, so the agency can't cure the problem by adopting different ones.  The due process problem inheres in the delegation itself.

Second, consider delegations that are entirely standardless.  Even when the delegates are disinterested public servants, such delegations can be unconstitutional merely by their structure (whether under due process or under some other doctrine, like equal protection or the First Amendment), because there is no constraint against the public officials' giving free rein to racist, speech-discriminatory, or other impulses—or yielding to similar impulses of others.

Third, consider Tumey v. Ohio, where a mayor also acted as a judge.  His compensation was suspect: he was paid, in part, out of fines assessed on defendants he convicted, so he had an interest in convictions.

If you were convicted under this system, you could complain of a due process violation based on the judge's pecuniary bias—even if you got every imaginable procedure, and even if you couldn't point to any way the judge's bias manifested itself or any decision tainted by his bad incentives.  Bias is subtle and hard to detect; the test isn't whether the decisionmaker is actually biased, but whether "the probability of actual bias on the part of the judge or decisionmaker is too high to be constitutionally tolerable."  Justice must also "satisfy the appearance of justice."  Acting badly is bad, but the bias caselaw doesn't demand proof of actual bad acts; it condemns biased structures that make bad acts more probable, and the appearance of such bias.

2. Indirect and Nonfinancial Bias

Because bias cases rely on questions of degree like the probability of bias and the appearance of impartiality, the Supreme Court hasn't insisted on any specific mechanism of bias, like payments to identifiable decisionmakers.  If an entity as a whole—particularly in the judicial or quasi-judicial context—has stood to benefit financially from its decisions, the Court has condemned the arrangement without requiring a showing that decisionmakers were under pressure from elsewhere in the organization to achieve a particular result.  Usually, it's been enough to observe that the entity's self-interest was implicated, in the sense that the entity as a whole could be made better off by its decisions.

In fact, in conducting the necessary "realistic appraisal of psychological tendencies and human weakness," the Court has relied on behavioral factors to find a probability of bias excessive—unapologetically calling this a process of "informed" "speculation."

In Tumey, the mayor-judge's direct incentive was enough to find a due process violation, but the Court also gave another reason: as chief municipal executive, he was responsible for municipal finances.

In Caperton v. A.T. Massey Coal Co., the Court held that it violated due process for an elected state supreme court justice to hear a case where a party had spent millions to get him elected.  The justice's compensation didn't depend on his ruling, and the donor had no power to get him removed, so what was the source of bias?  Just the "debt of gratitude" he would feel toward his benefactor.

Likewise, in Williams v. Pennsylvania, the problem was that a state supreme court justice hearing a case had been involved in the same case as a prosecutor years earlier.  One could reasonably fear that the justice "'would be so psychologically wedded' to his . . . previous position as a prosecutor that [he] 'would consciously or unconsciously avoid the appearance of having erred or changed position.'"  In In re Murchison—where a judge acted as a "one-man grand jury" under Michigan law, charged a witness with contempt, and tried that contempt proceeding—one of the concerns was that the judge "cannot be, in the very nature of things, wholly disinterested in the conviction or acquittal of those accused" and might have "the zeal of a prosecutor."  And in Morrissey v. Brewer, the Court said that a "determination that reasonable ground exists for revocation of parole should be made by someone not directly involved in the case," i.e., not the supervising parole officer, because "[t]he officer directly involved in making recommendations cannot always have complete objectivity in evaluating them."

The emphasized phrase from In re Murchison—"in the very nature of things"—echoes the same phrase from Carter Coal that we saw in the Introduction.  The Supreme Court has assumed, realistically, that people can be tempted in subtle, complex, and hard-to-detect ways, which don't always involve financial gain.  Therefore, especially since we also care about the appearance of justice, we should avoid tempting arrangements.

3. Curing Bias Through Disinterested Review

How can bias be cured?  Perhaps some bias could be cured by the official's voluntary acts (could the Tumey judge announce, before trying a case, that he would forgo any fine-based compensation?).  But not all cases can be cured this way (could the Caperton justice announce that he would feel no gratitude toward his benefactor?).

While financial bias can be addressed by changing the compensation structure, another general cure is to have a disinterested party ratify the decision; there's generally no due process violation when the self-interested party's power is limited by the involvement of a neutral decisionmaker by the time of the first adjudication.  So, as we'll see, due process generally isn't implicated when private parties can merely set (disinterested) legal machinery in motion—even when they're frankly self-interested, as when they're motivated by the prospect of rewards.

If this were always true, perhaps prosecutorial bias would be harmless as long as the judge is neutral.  But in United States v. James Daniel Good Real Property, the Supreme Court insisted on a predeprivation hearing before a seizure of real property, in light of the government's "direct pecuniary interest in the outcome of the proceeding": the government's "financial stake in drug forfeiture."  Was any official directly compensated from drug forfeiture proceeds?  Presumably not—but there was a memo from the Attorney General urging U.S. Attorneys to meet the DOJ's budget targets.  The Court was concerned with structural bias, not just individual compensation arrangements—even though the seizure had been approved by a (neutral) magistrate judge.  Thus, while due process imposes lesser constraints on prosecutors and plaintiffs than on judges and quasi-adjudicative officials, some constraints do exist.

D. Why Should We Care?

1. Good Functional Reasons

Does this three-part categorization matter?  In the D.C. Circuit's first Amtrak decision, Judge Brown suggested that "the distinction evokes scholarly interest" but wouldn't "effect a change in the inquiry."

But we should care.  Each of these doctrines corresponds to a particular constitutional concern, hinted at in the rubrics of "giver-based," "recipient-based," and "application-based."  Congress can't give up too much power; officers can't exercise executive power without proper appointment; nobody can deprive someone of life, liberty, or property through biased decisionmaking.

And each category implies a fix for unconstitutional delegations.  Has Congress given up too much?  Narrow the delegation.  Has Congress given power to someone incapable of wielding it?  Give the power to someone else.  Has a legislature given power to someone with financial bias?  Change the compensation structure.

If we miscategorize a delegation, we'll not only misunderstand the constitutional problem, we'll also misunderstand how to salvage the delegation.

Thus, consider Judge Brown's original holding in the Amtrak case—that, because Amtrak was private, the delegation violated the Article I Nondelegation Doctrine.  Though that doctrine is generally giver-based (Congress can't give up too much power), a supposed subdoctrine specifically against private delegations would be recipient-based.  If she were right, then one solution would be to instead delegate to a public entity.

The Supreme Court held that Amtrak was in fact public, and sent it back to Judge Brown for reevaluation under this new understanding.  So was everything now constitutional?  No: on remand, Judge Brown correctly identified the real problem, which was Amtrak's pecuniary bias.  Public or not, the statute required Amtrak to maximize profits, which required it to act adversely to the freight railroads that were its competitors for scarce track.  Judge Brown reached the same result as before, but on the better ground of due process.

The same scenario shows up repeatedly.  A popular account of the Article I Nondelegation Doctrine is that it's about preventing "arbitrariness" and "uncontrolled discretionary power."  This theory folds the doctrine into something like due process, and suggests that Article I nondelegation problems can be cured by the agency's unilaterally adopting adequate procedures.

But beyond the difference in doctrinal philosophy (Article I nondelegation is about separation of powers, while due process is about fundamental fairness), this recharacterization has real effects.  For instance, due process theories (unlike Article I theories) would apply against the states and would open up the possibility of Bivens damages.  And, more fundamentally, one theory is giver-based while the other is application-based, which means the fixes are different.  When we ask how we can cure the problem, it's important to know whether the solution is make Congress delegate a narrower power or provide better procedures.

And indeed, the Supreme Court has squarely rejected the Article-I-nondelegation-as-due-process theory, noting that, if Congress has delegated too broadly, separation of powers has already been breached.  If an agency gets excessive power and then clearly announces that it will voluntarily limit itself, that self-limitation is itself a forbidden exercise of power—even though there's no unfairness (and no due process violation), since everyone's on notice as to the conduct required or prohibited.

A violation of the Article I Nondelegation Doctrine thus needn't violate due process (or even implicate arbitrariness).  The same is true in reverse: If Congress passes a hyperspecific statute allowing welfare benefits to be withdrawn without process, due process will be violated but there will be no impermissible delegation under Article I.

2. Overlap Is Not a Problem

This doesn't mean the theories can't overlap.  Consider the Inherent-Powers Corollary to the Article I Nondelegation Doctrine: no intelligible principle is necessary when the delegate already has some inherent power over the subject matter.  Thus, for instance, no intelligible principle is necessary when delegating to Indian tribes a power to regulate commerce in Indian country, because Indian tribes have "attributes of sovereignty."  But an identical (standardless) delegation to the Bureau of Indian Affairs (BIA) might violate the Nondelegation Doctrine because the BIA lacks such sovereignty.  If Congress made such a standardless delegation to the BIA, would this be a giver-based problem (because Congress gave up too much power—just narrow the delegation by specifying a standard next time!) or a recipient-based problem (because Congress gave this power to a nonsovereign entity—just delegate to the Tribe next time!)?

Why not both?  Congress can have multiple ways of salvaging a delegation.

Similarly, recall that the presence of procedures can save a delegation under the Article I Nondelegation Doctrine.  But some of those procedures might also be mandated by due process.  If those procedures are absent, is this a giver-based problem (because the absence of procedures made the delegation unbounded and thus a delegation of legislative power) or an application-based problem (because the absence of procedures violated rights)?

On the appointments side: insisting on power being wielded by properly appointed officers can be thought of as a structural protection for the executive branch (a recipient-based problem).  But it can also be thought of as a fairness protection for the would-be targets of non–politically accountable enforcement (an application-based problem).  Indeed, that's the standard defense of structural constitutional doctrines like separation of powers or federalism: they're good not in themselves but because they preserve liberty.

All these options can be valid.  But we should still keep the doctrines analytically distinct.  The procedures that save a delegation from being "legislative" under the Article I Nondelegation Doctrine are ones that constrain discretion, for instance by enforcing rationality under "hard look" review.  But the procedures that save a delegation under due process are ones that ensure fairness, for instance by minimizing bias or providing notice.  Maybe there's some overlap, maybe not.

And couldn't we characterize any giver-based doctrine as a recipient-based doctrine?  If Congress must retain its legislative power and can't give it away (that's giver-based), isn't it also true that no one else may exercise legislative power (that's recipient-based)?  Sure.  But this categorization isn't about abstract elegance; it's about useful functional questions like Should Congress narrow the power? or Should Congress find a different delegate?  If there turns out to be more than one way to salvage an unconstitutional delegation, so much the better.