Retaining Control After Delegating

Retaining Control After Delegating

When Congress hands an agency broad authority, how much control may it keep over the way the agency uses it? In the 1970s, many members of Congress concluded that ordinary legislation was too cumbersome a way to supervise administrative action, and they turned to an old device, the legislative veto.¹ Statutes of this kind told agencies to send final rules to Congress before the rules took effect, and a vote of two chambers, of one chamber, or even of a committee or its chair could kill a rule.² Nearly 200 such provisions were on the books when INS v. Chadha (1983) held that a one-house veto of an executive decision violates Article I, § 7.³

The veto answered a worry that runs through this chapter. Only Congress holds the federal legislative power, because only Congress is politically accountable for its policy choices.⁴ Agencies that write binding rules under broad grants are less accountable. Congress can always narrow or revoke an agency's authority by passing a later statute,⁵ but a statute needs both houses and the President, and the veto promised a cheaper path.

The veto also had a respectable theory behind it, an argument from symmetry. If Congress may delegate lawmaking power to agencies that issue rules with the force of law without a vote of both houses or a presidential signature, why may it not reserve a check on that power for itself? Justice White pressed that question in dissent, and the majority answered it directly; both positions appear below. The rest of the unit follows Chadha's reasoning into two related problems, agents that Congress controls and a line-item veto that the President would hold, and then lists the checks that Congress still has.

INS v. Chadha

Article I, § 7 requires every bill that passes the House and the Senate to be presented to the President before it becomes law, and it extends the same requirement to any order, resolution, or vote that needs the concurrence of both Houses, except on a question of adjournment.⁶ In INS v. Chadha the Court held that a one-House legislative veto of an executive branch decision is unconstitutional, because the veto bypasses presentment to the President and approval by both Houses.⁷ The holding does not depend on immigration. Any exercise of legislative power by Congress that has the purpose and effect of altering the rights, duties, or legal relations of persons outside the legislative branch must conform to bicameralism, meaning approval by both Houses, and to presentment, meaning delivery to the President for signature or veto.⁸ Chief Justice Burger wrote for a 7–2 Court.⁹

Jagdish Rai Chadha was born in Kenya, held a British passport, and entered the United States in 1966 on a nonimmigrant student visa that expired on June 30, 1972.¹⁰ After immigration officials told him he had overstayed, he conceded that he was deportable and applied for suspension of deportation under § 244(a)(1) of the Immigration and Nationality Act.¹¹ On June 25, 1974, an immigration judge acting for the Attorney General suspended the deportation after finding that Chadha had been continuously present for seven years, was of good moral character, and would suffer extreme hardship if removed.¹²

The statute required a report of every such suspension to go to Congress, and § 244(c)(2) let either House invalidate the Attorney General's decision by resolution.¹³ On December 16, 1975, the House passed a resolution disapproving the suspension for Chadha and five other aliens on the ground that they did not meet the statutory requirements, particularly as to hardship.¹⁴ The House acted without a recorded vote, and the resolution was never submitted to the Senate or presented to the President.¹⁵ The immigration judge then reopened the proceedings and ordered Chadha deported. The Ninth Circuit held the veto unconstitutional and set the order aside, and the Supreme Court affirmed.¹⁶

The majority began from a presumption that the House's veto was an exercise of legislative power.¹⁷ The presumption does not end the inquiry, because not every action by either House must satisfy bicameralism and presentment. Whether a House has exercised legislative power depends not on the action's form but on “whether they contain matter which is properly to be regarded as legislative in its character and effect.”¹⁸ The Framers had seen the relabeling problem coming. In the final debate on § 7, Madison worried that Congress could evade presentment by calling a proposed law a “resolution” or “vote” instead of a “bill,” and the third clause of § 7 was added in response.¹⁹ If a title could decide the question, Congress could make binding law by changing a word.

Measured by that test, the House's resolution was legislative. It had the purpose and effect of altering the legal rights, duties, and relations of persons outside the legislative branch, among them the Attorney General, other executive officials, and Chadha himself.²⁰ Without § 244(c)(2), Congress could have forced Chadha's deportation, if at all, only by legislation requiring it.²¹ Nor could the House recast its veto as an amendment or partial repeal of § 244, because amendment and repeal of statutes, no less than enactment, must conform to Article I.²² Both of Congress's moves in the case follow the same rule. Its original choice to delegate and its later disagreement with the Attorney General's decision each involve determinations of policy, and Congress can implement either only by bicameral passage followed by presentment to the President.²³ Until Congress alters the delegation by that route, the statute's terms govern the delegate's authority.²⁴

Each step in the Article I procedure does separate work, which is why skipping any of them matters. The bicameral requirement, the Presentment Clauses, the President's veto, and Congress's power to override a veto were meant to erect enduring checks on each branch and to protect the people from the improvident exercise of power by mandating certain prescribed steps.²⁵ The power of two-thirds of both Houses to override a veto keeps the President's own veto from becoming final arbitrary action by one person.²⁶ The Court called the result “a single, finely wrought and exhaustively considered, procedure.”²⁷ A one-House veto removes the second chamber and the President in one stroke, so a decision with the force of law about a person's right to remain in the country was made by one body, with no second look and no executive judgment to overcome.

The Constitution does allow one House to act alone, with binding and unreviewable force, but only in four situations, and the Court used the short list to show how narrow the category is:²⁸

• The House alone may initiate impeachments.
• The Senate alone may try impeachments and convict.
• The Senate alone may approve or disapprove presidential appointments.
• The Senate alone may ratify treaties negotiated by the President.

Each power is stated in explicit terms and separately justified, and none covers a veto over an executive decision about deportation. The Court read that explicitness as a lesson about the rest of the Constitution. When the Framers wanted one House to act outside its bicameral role, they defined the procedure narrowly and precisely, so a power to act alone is not to be implied.²⁹ Because the House's action was legislative and fit none of the exceptions, it was subject to Article I's standards.³⁰

Congress's best reply drew on delegation itself. Agencies issue rules with the force of law without a vote in both Houses or the President's signature, so why should the House's veto face a stricter test than the delegate's decision? The majority answered that Congress's authority to delegate provides no support for controlling the administration of the laws through a veto.³¹ The Attorney General acts under a statute that Congress enacted through Article I, and “his administrative activity cannot reach beyond the limits of the statute that created it.”³² That executive action is always subject to check by the terms of the authorizing legislation, it is open to judicial review, and Congress can modify or revoke the authority entirely.³³ A one-House veto is legislative in both character and effect and has none of those checks, so the check that Article I supplies is needed.³⁴ Whether the delegation was too broad is a separate question under the intelligible principle test taught earlier in this chapter, and the Court has applied that test generously, as when it called it hard to deem lawless a broad delegation of rulemaking discretion to the President where the delegated duty is interlinked with duties the Constitution already assigns him.³⁵

The Court did not deny that the veto is efficient and useful. It held that convenience, standing alone, will not save a procedure that is contrary to the Constitution.³⁶ Nothing in the Constitution or the Court's decisions supports the proposition that the cumbersomeness and delays of explicit constitutional standards may be avoided by Congress or by the President. With all its flaws of delay, untidiness, and potential for abuse, the Court said, making the exercise of power subject to the carefully crafted restraints in the Constitution is still the best known way to preserve freedom.³⁷ A rule that excused the steps whenever skipping them saved time would let the safeguards give way exactly when Congress finds them most inconvenient.

Exam questions rarely repeat Chadha's facts. They change the House, the subject, and the label, so ask each time what the action does to persons outside Congress.

In both scenes the answer follows the effect on persons outside Congress, and the label “resolution” does not change it.

Justice Powell concurred only in the judgment, on narrower grounds.³⁸ He warned that the Court's reasoning would invalidate every use of the legislative veto, which Congress had placed in literally hundreds of statutes dating to the 1930s, and he thought the holding should be no broader than necessary out of respect for Congress's judgment about the device's usefulness.³⁹ In his view, the House exceeded its constitutionally prescribed authority when it applied its own rules to Chadha. The House enacted no general rule. It decided that six specific persons did not meet statutory criteria, a determination ordinarily left to courts or agencies, and so it assumed a judicial function.⁴⁰ He would not have reached the broader question whether legislative vetoes violate the Presentment Clauses.⁴¹

Justice White dissented and warned that the decision “sounds the death knell for nearly 200 other statutory provisions” in which Congress had reserved a legislative veto.⁴² His central argument was functional and took the form of the symmetry question. Agencies with delegated power issue regulations with the force of law without bicameral approval or the President's signature, so it is hard to see why Article I forbids Congress from reserving a check on that power for itself. In both cases, he said, it is enough that the initial statutory authorization complied with Article I.⁴³ Because the Constitution neither authorizes nor prohibits the veto directly, he would ask whether it is consistent with the purposes of Article I and the principles of separation of powers. He read those purposes as limiting the methods for enacting new legislation, not as restraining the scope of authority exercised under a duly enacted law.⁴⁴ He also attacked the majority's reliance on judicial review. The Court itself had noted that it knew of no decision reviewing a suspension of deportation, because no party has the motivation or the right to appeal one, so judicial review was a thinner check than the majority suggested.⁴⁵

Steps two and three carry most of the weight. Students instinctively defend a veto as a mere condition on the delegation, on the theory that Congress could have withheld the power altogether. Congress can indeed narrow or revoke an agency's authority, but it does so by later legislation, which means passage by both Houses and presentment to the President.⁴⁶ A resolution of one House changes a person's legal position without that process, so it fails whatever the authorizing statute said. The holding also reaches well beyond individual adjudications. Chadha appeared to invalidate all of the nearly 200 legislative veto devices then on the books, and two weeks later the Court summarily upheld challenges to two other vetoes in rulemaking proceedings.⁴⁷ Justice Powell's narrower ground therefore did not become the rule.

Congress-Controlled Agents In Bowsher v. Synarr (1986) the Court held that Congress cannot reserve for itself the power to remove an officer charged with executing the laws, except by impeachment.H0 The Gramm-Rudman-Hollings Act of 1985 set maximum allowable deficits and, when spending exceeded the limit, directed the Comptroller General to impose spending cuts under the statute's terms. The Comptroller General headed the General Accounting Office and could be removed by Congress. Applying a statutory formula to order cuts is a way of executing a law, so the question was whether Congress could assign that job to an officer it controlled, and the answer was no. The structure of the Constitution does not permit Congress to execute the laws, and it follows that Congress cannot grant to an officer under its control what it does not possess.H1 The removal-power cases belong to another reading, so Bowsher matters here for one proposition, and it is a short bridge from Chadha.

The bridge is the logic of the veto. Congress's disagreement with a delegate's decision, like its original choice to delegate, is a determination of policy that Congress can carry out in only one way, bicameral passage followed by presentment.⁵⁰ An officer answerable only to Congress would open a way around that requirement, because Congress could simply remove, or threaten to remove, the officer for executing the laws in any fashion it disliked. To permit an officer controlled by Congress to execute the laws would be, in essence, to permit a congressional veto exercised through a person instead of a resolution.⁵¹ Madison's warning, borrowed from Montesquieu, names the danger: when legislative and executive powers are united in the same person or body, apprehension arises that the same monarch or senate will enact tyrannical laws in order to execute them tyrannically.⁵²

Justice Stevens, concurring in the judgment, reached the same result without relying on the removal power. He saw no need to consider the Decision of 1789, the President's removal power, or the abstract nature of “executive powers.”⁵³ Once it was clear that the Comptroller General's statutory duties made him an agent of Congress and gave him the task of making policy determinations that bind the Nation, the question was simply one of congressional process, and those duties did not follow the prescribed procedures for congressional lawmaking.⁵⁴ The majority asks who controls an officer who executes the law, and the concurrence asks whether Congress is making binding policy through its own agent. Both treat Congress's agent as Congress.

The Court met the same problem in Metropolitan Washington Airports Authority v. Citizens for the Abatement of Aircraft Noise (1991). Congress had made administrative decisions about managing the Washington-area airports subject to veto by an oversight board staffed by members of the House and Senate transportation committees.⁵⁵ The Court treated the board as an agent of Congress, and that classification produced a dilemma with no exit. If the board's functions were legislative, bicameralism and presentment were required, and neither occurred.⁵⁶ If its functions were executive, Congress could not hold them at all, because Congress may not give itself, its members, or its agents a role in executing the laws.⁵⁷ Whatever the label, a Congress that wants to take legally effective action must use the full enactment process of Article I, § 7.⁵⁸

The same logic reaches arrangements that look unrelated to a veto. Consider a federal statute that transfers operating control of the national parks to the states, on condition that each state administer them through a five-person commission with at least two members of Congress from that state. The statute is invalid because it lets members of Congress help implement a federal law.⁵⁹ That a state governor appoints the commissioners changes nothing, because the decisive fact is that Congress is vesting its own members with a role in executing a federal law. If federal oversight of the parks matters, the Constitution requires assigning that function to the federal executive branch, and not to Congress.⁶⁰

What, then, may Congress keep after it delegates? It may keep the power to change the delegation through legislation, and a program run on those terms looks like this.

The parks statute and the loan statute both leave Congress with an interest in how a program is run, so the difference is where that interest is exercised. In the parks statute it is exercised from inside the machinery of execution, by legislators who sit on the commission. In the loan statute it is exercised from outside, through legislation that must satisfy Article I, which is the only way Chadha allows Congress to revisit what it delegated. The next section turns to a related device that gave the President a role in what Congress had enacted.

The Line-Item Veto

Article I, § 7, clause 2 requires that every bill passed by the House and the Senate be presented to the President before it becomes law, and it gives him two options, to sign the bill or to return it.⁶³ The choice attaches to the whole bill, and Presidents are not permitted to sign only parts of federal bills.⁶⁴ The Line Item Veto Act of 1996 tried to supply a partial veto by ordinary statute. In Clinton v. City of New York (1998) the Court held the attempt unconstitutional because “[i]n both legal and practical effect, the President has amended two Acts of Congress by repealing a portion of each,” and repeal of statutes, like enactment, must comply with Article I procedures.⁶⁵ Justice Stevens's majority opinion put the consequence in a phrase: a President lacks “the unilateral power to change the text of duly enacted statutes.”⁶⁶

The Act operated only after Congress and the President had finished the ordinary process. The President would sign an appropriations bill and then, immediately afterward, notify Congress that specific expenditures were cancelled.⁶⁷ The power reached items of new spending, discretionary budget authority, and limited tax benefits.⁶⁸ A cancellation took effect at once, and it became “null and void” only if Congress passed a disapproval bill and the President signed it. If the President vetoed the disapproval bill, the cancellation stayed in place unless two-thirds of both Houses overrode the veto.⁶⁹ That sequence explains why the Court saw an amendment. When the President signed the appropriations bill its spending items were law, and after the cancellation they had no legal force, a change produced by one officer's decision. Article I's ordinary burden ran the other way, because changing a law requires both Houses and the President, and under the Act the President's change stood unless a supermajority restored the item.

The Court's objection was to the cancellation of parts of duly enacted statutes, which happened only after the Presentment Clause had been satisfied for the bill as a whole.⁷⁰ Chadha had already established that legislative action altering legal rights, duties, or relations must satisfy bicameralism and presentment, and Clinton drew on its statement that “repeal of statutes, no less than enactment, must conform with Art. I.”⁷¹ The reason follows from what Article I protects. The Constitution limits the President's unilateral veto by letting two-thirds of both Houses override it, which precludes final arbitrary action by one person.⁷² A cancellation power reverses that design, because it turns one officer's judgment into a change in the law that only an override-sized majority can undo.

The contrast with Field v. Clark shows why a long history of suspension statutes did not control. The Tariff Act of 1890 exempted sugar, molasses, coffee, tea, and hides from duty and authorized the President to suspend the exemption whenever he was satisfied that a producing country imposed duties on American agricultural products that he deemed “reciprocally unequal and unreasonable.” The Court upheld the Act in Field, reciting statutes since 1798 that gave the President similar powers.⁷³ The Clinton Court distinguished those authorizations on two grounds: they were contingent upon an executive finding of fact, and they related to foreign affairs, where the President has a special degree of discretion and freedom.⁷⁴ The first ground has a structural meaning. The tariff law as Congress wrote it already contained both outcomes, exemption and duty, and a fact in the world decided which applied, so a suspension carried out the statute's own terms. Under the Line Item Veto Act the appropriation was complete law directing money to be spent, and a separate statute let the President decide, after signing, that the item should not take effect because cancelling it would reduce the deficit without impairing essential government functions or harming the national interest.⁷⁵ Those are judgments about wise policy, and the item's legal force disappeared on the strength of that judgment.

The dissenting Justices read the same facts as a delegation problem. Justice Scalia, concurring in part and dissenting in part, argued that Article I, § 7 no more categorically forbids executive reduction of congressional dispositions under a statute authorizing it than it forbids executive augmentation through substantive rulemaking.⁷⁶ On his view the limit comes from the doctrine of unconstitutional delegation, because authorized executive reduction or augmentation that goes too far usurps a function of Congress that cannot be delegated. The question he posed was whether authority to cancel an item of spending gives the President a power that history and tradition show must reside exclusively in the Legislative Branch.⁷⁷

Justice Breyer, dissenting, divided the problem the same way. He did not think a cancellation was literally a repeal or an amendment, and because the Act itself had been enacted through Article I's procedures, he found its departure from those procedures beside the point.⁷⁸ That left the nondelegation question, where he found an intelligible principle in three places: a procedural direction to consider the legislative history, construction, and purposes of the law containing each item; a purpose of greater fiscal accountability and the elimination of wasteful spending and special tax breaks; and a requirement that the President determine that cancellation would reduce the deficit, impair no essential government function, and not harm the national interest.⁷⁹ The standards were broad, but the Court had upheld equally broad ones, and the Act involved no roving commission, no private recipients, and no power over all American industry, the features this chapter saw in Schechter and Panama Refining.⁸⁰ To the objection that the Act reversed the usual burden, he answered that where the burden of overcoming legislative inertia lies is within the power of Congress to determine by rule.⁸¹

The two framings ask different questions, and the difference decides outcomes. The majority asked what the President did to a statute already in force. If the President removed a provision's legal force on his own decision, a precise standard cannot help, because the objection concerns the procedure for altering a law, and no standard replaces the vote of both Houses and the President's signature. The dissenters asked how much policy discretion Congress handed over, a question a statute can pass by stating a purpose and limits. A student who meets this Act on an exam and writes only about the intelligible principle has answered the dissenters' question, while the Court's answer lies in Article I, § 7. The case also pairs with Chadha. There Congress kept a one-House shortcut around Article I for itself; here Congress handed the President a shortcut; and the same procedure defeated both.

Permissible Congressional Checks

Chadha tells Congress what it may not do, and the same rule tells it what remains available. Any exercise of legislative power that alters the rights, duties, or legal relations of persons outside the legislative branch must pass both Houses and be presented to the President.⁸⁴ A check on an agency that takes that form is ordinary lawmaking, so it raises none of the problems that sank the one-House veto. Congress can always revoke or narrow the authority it has granted through subsequent legislation, and every tool below is a variation on that theme.⁸⁵ Chadha's own facts left Congress that path, because undoing the suspension of one person's deportation would have required legislation requiring deportation.⁸⁶ The tools differ in when they act and in how much they ask of Congress, but each runs through Article I, § 7.

The most direct tool is a statute that narrows or reverses what an agency has done. When large numbers of consumers complained about the National Highway Traffic Safety Administration's seat belt “interlock” rule, Congress promptly amended the agency's organic act to provide that its safety standards could not require belt systems that prevented the car from starting or sounded a continuous buzzer when belts were unfastened.⁸⁷ The Food and Drug Administration's attempts to ban saccharin, the only approved artificial sweetener, aroused such strong popular opposition that Congress imposed a moratorium on regulatory action and substituted a warning label for products containing the chemical.⁸⁸ In both episodes Congress acted by statute, so the change in the agency's authority came only after a bill cleared both Houses and the President. That sequence is why these checks survive Chadha: the new statute is itself an exercise of legislative power that took the route Article I prescribes.

Congress can also change how agencies decide, which steers many future choices instead of reversing one past decision. The National Environmental Policy Act of 1969 forced all federal agencies, for the first time, to consider the environmental impacts of major decisions, and the Regulatory Flexibility Act requires agencies to study the economic effect of regulations on small businesses and to consider ways of minimizing those effects.⁸⁹ The Unfunded Mandates Reform Act of 1995 goes further for rules that impose high compliance costs on other governmental units or the private sector. Such a rule must be accompanied by an extensive economic analysis, and the agency must adopt the least burdensome alternative that meets the rule's objectives or explain why it did not.⁹⁰ These statutes rest on the premise of the earlier discussion of procedures and judicial review, that procedural controls make a broad delegation easier to tolerate.

A third tool slows an agency without overruling it. Report-and-wait provisions defer the effective date of an agency action so that Congress has an opportunity to pass a statute blocking the action and to secure the President's signature on it.⁹¹ The waiting period gives no one a veto, because the rule takes effect on schedule unless Congress completes the enactment process in the interval. That feature separates the device from the legislative veto, which let a vote of one House, a committee, or even a chair end an executive decision with no statute at all.⁹² Ordinary legislation takes time, and the waiting period gives Congress that time before the rule binds anyone.

The most structured tool is the Congressional Review Act of 1996, which allows Congress to adopt, using expedited procedures, a joint resolution of disapproval of any regulation adopted within the previous 60 days of legislative session.⁹³ If the President signs the resolution, the regulation is nullified, and under 5 U.S.C. § 801(b)(2) the agency may not reissue the rejected rule “in substantially the same form,” or issue a new rule that is substantially the same, unless a law enacted after the resolution specifically authorizes it.⁹⁴ The bar on reissuance means a disapproval does more than pause the rule, because the agency needs fresh statutory authority before it can return to the same policy. Congress invoked the Act only once between 1996 and 2016. Then, at the beginning of the Trump Administration, it rejected a host of Obama Administration regulations in a flurry of activity, among them an Interior Department rule restricting disposal in streams of debris from mountaintop removal mining.⁹⁵

An exam will ask why a joint resolution that kills an agency rule survives when the House resolution that overrode the suspension of Chadha's deportation did not, given that both end an executive decision. The answer lies in the route the action takes, since the effect on the regulated party can be the same. In Citizens for Constitutional Integrity v. United States, 57 F.4th 750 (10th Cir. 2023), a court of appeals distinguished the Review Act from a legislative veto because its resolutions must comply with bicameralism and presentment.⁹⁶

The committee device in that statute belongs to the family Chadha condemned. Legislative vetoes took many forms, including a vote of one chamber, of a committee, or of a committee chair, and Chadha appeared to invalidate all of the nearly 200 devices then on the books.⁹⁷ The Review Act resolution produces a similar result for the rule but only after it has satisfied Article I, and that difference in route decides the constitutional question.

The price of staying within Article I is strength. A one-House veto promised that a single vote could end an executive decision, while every tool above must clear both Houses and the President, and if the President vetoes, the measure becomes law only when two-thirds of each House repass it.⁹⁸ Congress's control over delegated power is therefore real but weaker than the legislative veto promised, and each check is only as effective as the majorities Congress can assemble.