The Statute That Stopped Chasing the Deficit

The operative question behind the Budget Enforcement Act of 1990 was narrower and harder than the question Congress had been asking during the preceding five years. The earlier question concerned an outcome: how large would the federal deficit be in a given fiscal year, and how quickly would that figure fall toward balance under a schedule written into law. The newer question concerned conduct: what limits would bind Congress at the moment Congress acted, when it voted an appropriation, when it expanded a benefit program, or when it cut a tax. That shift from outcome to conduct supplied the statute with its logic, its enforcement machinery, and its durability during the decade that followed enactment. A law that punished Congress for a deficit figure produced in large part by an economy no legislator controlled invited evasion, optimistic forecasting, and emergency spending that escaped the accounting altogether. A law that measured Congress against its own recorded votes promised a different kind of discipline, because the conduct being judged and the institution being judged were the same. The distance between those two designs explains why the 1985 experiment failed on its own terms and why the 1990 replacement, although less dramatic in its stated ambition, proved harder to circumvent in ordinary legislative practice.

What were the Budget Enforcement Act’s controls designed to protect?

It replaced fixed deficit targets with enforceable limits on discretionary appropriations and a pay-as-you-go rule for new mandatory spending and revenue laws, preserved the deficit reduction enacted in its parent reconciliation act, and backed both controls with automatic sequestration when Congress breached a limit or recorded a scorecard debit.

The formal identity of the statute deserves precision, because the short title concealed the size of the vehicle that carried it into law. The Budget Enforcement Act of 1990 formed Title XIII of the Omnibus Budget Reconciliation Act of 1990, cited as Public Law 101-508 and printed at 104 Stat. 1388. President George H. W. Bush signed the reconciliation act on November 5, 1990, and the measure had been enacted by the 101st Congress after a negotiation and a legislative collapse that tested the agreement before final passage. In legal terms, the enforcement title amended two earlier frameworks rather than creating a free standing budget code. It rewrote central parts of the Balanced Budget and Emergency Deficit Control Act of 1985, the deficit control statute associated with Senators Phil Gramm, Warren Rudman, and Ernest Hollings, and it amended the Congressional Budget Act of 1974, the statute that had created the House and Senate Budget Committees, the Congressional Budget Office, and the annual budget resolution process. The placement mattered. Because the new controls lived inside the 1985 act as amended, sequestration, the automatic cancellation of budgetary resources used to enforce a breach, retained a statutory home with definitions, exemptions, and reporting duties that the Office of Management and Budget and the Congressional Budget Office could administer without building an enforcement bureaucracy from nothing. Because the reconciliation act itself carried the tax increases and spending reductions negotiated during 1990, the enforcement title could perform the task assigned to it in the same law that created the savings worth protecting.

Understanding the replacement requires an account of the regime replaced. The Balanced Budget and Emergency Deficit Control Act of 1985 had set fixed deficit targets for successive fiscal years, declining on a path intended to reach balance, and had provided that a projected breach of a target would trigger sequestration across a broad range of programs. The appeal of that design in 1985 was plain. Deficits had grown during the early 1980s through a combination of tax reduction, defense buildup, recession, and high interest costs, and repeated annual negotiations had failed to produce a credible downward path. A target written into statute appeared to remove discretion from a process that discretion had repeatedly defeated. Yet the design contained a structural flaw that experience exposed within a few budget cycles. The deficit is a residual. It records the difference between revenues collected under law and outlays spent under law, but the size of that difference in any particular year depends on forces that no appropriations bill and no tax bill determines by itself. Economic growth governs receipts from individual and corporate income taxes and from payroll taxes. Unemployment governs spending on income support and other automatic stabilizers, which rise during a contraction by design and without a new statute. Interest rates govern the cost of servicing debt accumulated over decades. Inflation, asset prices, and the timing of receipts and payments all move the final figure. Forecasting error compounds each of those influences, because a target must be judged against a projection made months before the fiscal year ends, and a projection rests on assumptions about growth, rates, and caseloads that honest analysts dispute in good faith.

A sequester tied to such a residual therefore threatened punishment for events that Congress had neither caused nor approved. A recession beginning after enactment of a responsible budget could widen the deficit through falling receipts and rising benefit payments and could bring automatic cuts into force at the precise moment when private demand had contracted and when deeper public retrenchment carried the greatest economic risk. Conversely, an unexpected surge in receipts could carry the government inside a target even after Congress had enacted expensive legislation, rewarding conduct that the regime existed to restrain. The target measured luck and judgment in a single number and treated both alike. Political behavior adapted to that defect in predictable ways. Because the penalty followed a projection rather than a vote, energy flowed into the projection. Assumptions grew more optimistic near the margin of compliance. Costs were timed to fall outside a measured year, savings were claimed early, and asset sales and other one time transactions acquired attraction because a target cares about a figure in a year rather than about the underlying obligations a transaction creates. Congress also revised the targets when the targets became inconvenient, most notably through the 1987 amendments that pushed the path to balance outward, a concession that an outcome rule can be rewritten by the same majority that finds obedience painful. The 1985 statute had intended to create a commitment device. In practice the device committed Congress to a number, and numbers can be restated, reestimated, and renegotiated more easily than conduct can be denied once the Congressional Record and the Statutes at Large have preserved a vote.

The 1990 act abandoned that architecture deliberately rather than by drift. Its organizing idea can be stated as a namable claim: Constrain the action, not the outcome. The sentence compresses an institutional theory into a rule of legislative craftsmanship. Congress cannot legislate the growth rate of the economy for the next fiscal year, cannot set the interest rate on Treasury securities by statute, and cannot repeal forecasting error by declaring an estimate authoritative. Congress can, however, decide how much discretionary budget authority to provide in annual appropriations acts, subject to the constitutional role of the President in signing or vetoing those acts. Congress can also decide, bill by bill, whether legislation affecting mandatory spending or revenues will increase the deficit over a measured period, because the text of each such bill, scored under agreed conventions, produces an estimate attributable to that act of lawmaking alone. A regime built on those two controllable choices assigns responsibility where responsibility rests. If appropriations exceed a limit that Congress wrote into law, the excess traces to identifiable bills passed through identifiable committees. If a tax cut or a benefit expansion records a debit, the debit traces to a statute with sponsors, a committee report, and roll call votes. Enforcement that follows an action carries a legitimacy that enforcement following a macroeconomic surprise cannot carry, because the sanctioned institution can point neither to the weather of the economy nor to the errors of forecasters as the true author of the breach.

Constrain the action, not the outcome also described a more modest and more honest relationship between law and the deficit itself. The 1990 act did not promise balance by a date certain, and its drafters gave up the theatrical certainty that a target schedule performs for public audiences. Deficit reduction would result, if reduction resulted at all, from the accumulation of restrained choices over successive sessions: appropriations held beneath ceilings year after year, and new commitments paid for at enactment rather than charged against an unspecified future. The approach traded a guarantee that could not be kept for a discipline that could be observed. Observability was central to the trade. The Office of Management and Budget would report whether appropriations had breached a cap, and the same office would maintain a scorecard recording the cumulative effect of enacted mandatory spending and revenue legislation. Both judgments concerned documents and statutes rather than prophecy, and both could be audited against figures that committees possessed when they acted. The claim did not deny that outcomes mattered; the whole enterprise existed because deficits mattered to interest costs, to national saving, and to the room available for response during a future emergency. The claim denied only that an outcome made a serviceable trigger for automatic punishment, and it insisted that a legislature serious about an outcome begins by governing the levers within reach of its own hands.

Two instruments gave the theory practical force, and each addressed a distinct channel through which deficits grow. The first instrument placed statutory limits, generally called discretionary caps, on the budget authority and outlays provided through annual appropriations. Discretionary spending covers the operations of the federal government that Congress funds each year, from defense activities to national parks to the administration of courts, and before 1990 that spending had been guided by the allocations of the annual budget resolution, enforced through congressional points of order that a determined majority could waive. Statutory caps converted guidance into law and attached sequestration to a breach, so that appropriations exceeding a ceiling would be reduced automatically within the affected category rather than excused by a procedural vote. During the opening years of the regime, separate ceilings for defense, international, and domestic spending, with statutory walls between the categories, prevented a familiar maneuver in which restraint in one area quietly financed expansion in another without an explicit decision to transfer resources across national priorities.

The second instrument, pay-as-you-go, addressed the parts of the budget that appropriations do not reach. Mandatory spending flows from eligibility rules and benefit formulas in permanent law, and revenues flow from the tax code, so a Congress determined to expand a program or reduce a tax could worsen the deficit for years without touching an appropriations bill. Pay-as-you-go required that the cumulative effect of newly enacted legislation in those areas remain deficit neutral over a rolling multi year period recorded on a scorecard maintained by the Office of Management and Budget. A tax cut had to be matched by revenue increases or mandatory savings elsewhere in the enacted record, and a benefit expansion had to carry its own financing, unless Congress and the President designated the measure as an emergency and kept it off the scorecard under the statutory procedure. A net debit standing at the close of a session exposed nonexempt mandatory programs to sequestration. The two instruments together closed the main exits from restraint. Caps without pay-as-you-go would have left the tax code and the entitlement statutes available as substitutes for appropriated spending. Pay-as-you-go without caps would have left annual appropriations free to absorb pressure that the scorecard had displaced.

The purpose assigned to that machinery differed from the purpose assigned to the machinery of 1985, and the Congressional Research Service drew the distinction with care in its accounts of the statutory history. The earlier deficit control act had sought to use budget controls to force future deficit reduction legislation, holding sequestration over Congress as a prod toward bargains that had not yet been negotiated. The 1990 act used controls for a narrower and more defensible end: controls preserved the deficit reduction already achieved in the accompanying reconciliation act. The distinction reverses the relationship between enforcement and agreement. Under the 1985 theory, enforcement preceded agreement and attempted to manufacture agreement through fear of automatic cuts. Under the 1990 theory, agreement came first, embodied in a reconciliation act that the Congressional Research Service projected would reduce the deficit by about $482 billion over five years, through about $158 billion in revenue increases and about $324 billion in spending cuts and debt service savings, and enforcement followed to guard the settlement against erosion. A future Congress remained free to legislate, but legislation that spent the savings or surrendered the revenue would have to reckon with caps that left no room in appropriations and with a scorecard that demanded offsets for mandatory and tax changes. The barrier was procedural and statutory rather than constitutional, yet it raised the price of reversal from a quiet line in a later bill to an overt act of undoing, visible to markets, to journalists, and to voters who had been told that sacrifice in 1990 purchased a smaller deficit over the following years.

That preservative purpose also explains why the statute attracted support from legislators who had resisted the target regime or had watched the target regime lose credibility. Guardianship of a completed bargain asks less of human foresight than compulsion of a bargain not yet struck. Negotiators could calculate the reconciliation package because the package consisted of specified tax provisions and specified spending changes, each capable of estimation by the Joint Committee on Taxation and the Congressional Budget Office under conventions known to the parties. No negotiator could calculate the economy of 1993 during the autumn of 1990. Enforcement tied to the specified package therefore rested on knowledge that existed, while enforcement tied to a deficit figure for a distant year rested on knowledge that nobody possessed. The 1990 settlement also distributed sacrifice across revenue and spending in a single enactment, and preservation protected both sides of that distribution. Participants who had accepted tax increases had reason to insist that spending restraint survive, and participants who had accepted spending restraint had reason to insist that the revenue not be returned through later tax legislation recorded as a debit. Each faction became a monitor of the part of the settlement that the opposing faction might prefer to unwind, and the statute gave those monitors an automatic remedy rather than a complaint.

None of that discipline arrived without a political crisis, and the remainder of this profile begins with the budget summit of 1990, where executive and legislative negotiators met first in Washington and then at Andrews Air Force Base in Maryland, where President Bush conceded in a written statement of June 26, 1990 that deficit reduction would require tax revenue increases, and where an initial agreement collapsed on the floor of the House before a revised reconciliation act passed in late October. The profile then examines the two instruments in operational detail, tracing how separate discretionary categories governed appropriations during the opening years of the regime, how walls between categories gave way to consolidated limits during the middle of the decade, how the pay-as-you-go scorecard accumulated debits and credits across sessions, and how emergency designations and exemptions, including the treatment of Social Security and the limits placed on reductions to Medicare during a mandatory sequester, defined the boundaries of both controls. A separate section follows the Federal Credit Reform Act of 1990, enacted within the same title and effective beginning in fiscal year 1992, which replaced cash flow accounting for federal loans and loan guarantees with an accrual measure of subsidy cost and thereby closed a channel through which credit assistance had appeared cheaper than equivalent grants. Later sections follow the extensions of the caps and pay-as-you-go through the Omnibus Budget Reconciliation Act of 1993 and the Balanced Budget Act of 1997, weigh the operation of the regime during the years that produced federal surpluses in fiscal years 1998 to 2001 without assigning those surpluses to enforcement alone, and record the expiry of the statutory controls on September 30, 2002, when a Congress facing different fiscal conditions allowed the framework to lapse rather than renew it. The profile closes with the restoration of statutory pay-as-you-go in the Statutory Pay-As-You-Go Act of 2010, a successor that revived the scorecard and the mandatory sequester while omitting discretionary caps altogether, and with an assessment of what the 1990 design demonstrated about the limits of law as a restraint on the institution that writes the law. Each of those subjects returns to the question with which this section opened, because the history of the statute is the history of a legislature learning, at considerable political cost, that control over its own recorded actions was the only form of fiscal discipline that a statute could honestly supply.

A Summit, a Reversed Pledge, and a Defeated Package

The Budget Enforcement Act of 1990 emerged from failure. By the opening months of 1990, the deficit control regime enacted in 1985 had lost the capacity to command belief or fear in the form its authors had intended. That regime, the design the Gramm-Rudman-Hollings statute had tried to impose on the federal budget, rested on a simple proposition. Congress and the President would commit in advance to a declining series of maximum deficit amounts. If the projected deficit exceeded a target by more than a permitted margin, automatic spending reductions, divided between defense and nondefense programs, would close much of the gap. The threat was meant to discipline bargaining before bargaining began, by giving legislators and Presidents a consequence each disliked more than compromise. The statute treated the deficit as an outcome that law could order downward, year by year, until balance arrived.

Experience between 1985 and 1990 weakened every part of that proposition. Targets were missed, targets were revised, and deadlines were moved. The Gramm-Rudman-Hollings Act of 1985 had promised balance by fiscal year 1991 under its first schedule. The 1987 revision pushed balance to fiscal year 1993 and reset the intermediate targets. Each revision preserved the form of discipline while conceding that the previous form had demanded more deficit reduction, more quickly, than the political system had delivered. A target that Congress could amend when the target became inconvenient taught a lesson opposite to the lesson intended.

The record of actual deficits reinforced the doubt. The federal deficit for fiscal year 1990 reached about 221 billion dollars, a figure recorded in historical budget data compiled from Congressional Budget Office and Office of Management and Budget sources. Economic weakness during 1990 added pressure, because slower growth reduced revenue, and the savings and loan crisis added another burden through federal deposit insurance obligations. Those pressures arrived unevenly, in defense, in deposit insurance, and in revenue, and uneven pressure strained a mechanism designed for symmetrical pain.

Fiscal year 1991 posed the sharpest version of the problem. The revised Gramm-Rudman-Hollings target for that year set a maximum deficit of 64 billion dollars. Contemporary reporting in Education Week, published at the start of October 1990, described negotiators working against that ceiling while confronting a deficit estimate far above the ceiling. If the regular process failed to bring the projected deficit within the permitted tolerance of the target, automatic reductions would follow under the statutory formula. Contemporary reporting put the scheduled sequestration at about 106 billion dollars in spending reductions to take effect at the start of fiscal year 1991. A reduction of that magnitude, concentrated in the portion of the budget exposed to sequestration, threatened cuts deep enough to disrupt defense operations and domestic services alike. The Congressional Record preserved a later Senate recollection that the required sequester would have cut defense by between 25 percent and 35 percent.

Events in the Persian Gulf turned a difficult calculation into an urgent one. Iraq invaded Kuwait on August 2, 1990. The United States deployed forces to Saudi Arabia and the surrounding region in the weeks that followed, building a military presence whose costs could not be reconciled easily with a sequester that threatened to cut defense spending by a large fraction. A government preparing for a possible war in the Gulf faced a budget law poised to impose indiscriminate defense reductions on October 1, 1990, the first day of fiscal year 1991. The choice was framed less as a choice between a negotiated package and continued drift than as a choice between a negotiated package and an automatic fiscal shock during a military crisis. That framing did not make agreement easy, because agreement required tax increases that the President had promised to resist and Medicare savings that exposed both parties to attack. The framing did make inaction harder to defend.

The summit that produced the statute began before the invasion. President George H. W. Bush and congressional leaders convened a budget summit in May 1990, bringing senior administration officials and senior legislators into sustained negotiation over a multi year deficit reduction package. The negotiations extended through the summer and moved during September 1990 to Andrews Air Force Base in Maryland, where the relative isolation of a military installation reduced the audience for posturing and concentrated attention on the tradeoffs that a package required. The format also signaled seriousness, because failure at a summit conducted by principals would be attributed to named leaders rather than to anonymous process.

The arithmetic governing the summit was severe. Negotiators sought about 50 billion dollars of deficit reduction for the first year and about 500 billion dollars over five years. A package reaching that total required discretionary spending restraint, changes in mandatory programs, savings in Medicare, reductions in defense spending from earlier plans, and additional revenue. Each element carried an organized constituency. Defense restraint collided with the Gulf deployment. Domestic restraint collided with Democratic commitments to education, housing, and social services. Revenue increases collided with the central political promise of the Bush presidency. The summit therefore turned on a preliminary question. Revenue had to be admitted to the table, or the remaining categories had to bear reductions larger than their defenders would accept.

President Bush answered that question on June 26, 1990. In a written statement issued during the negotiations, Bush accepted that a deficit reduction package would require, among other elements, “tax revenue increases.” The wording was bureaucratic in form and explosive in effect, because a President who had excluded tax increases from acceptable outcomes had placed revenue increases inside the definition of an enactable package. The statement also paired revenue with entitlement and mandatory program reform, discretionary spending reductions, growth incentives, and orderly reductions in defense expenditures. That pairing allowed the administration to present the concession as one part of a balanced settlement rather than as a surrender on revenue alone. Opponents inside the Republican Party treated the concession differently, as a reversal that voters would remember in its simplest form.

The reversal collided with the most quoted promise of the 1988 presidential campaign. At the Republican National Convention in 1988, Bush had accepted the presidential nomination with the pledge rendered in shorthand ever afterward as “read my lips: no new taxes.” The pledge had been framed as a prediction about congressional pressure and presidential resistance. Congress would push for higher taxes, the candidate had said, and the response would be refusal, repeated for as long as the pressure continued. The force of the pledge came from its absoluteness. The pledge left no room for a distinction between new taxes and increases in existing taxes, between rate increases and base changes, or between revenue raised by legislation and revenue raised by economic growth. The June 1990 statement abandoned that campaign commitment under negotiating pressure from a Democratic Congress, from deficit projections that left the Gramm-Rudman-Hollings target out of reach, and from advisers who concluded that no package large enough to matter could pass without revenue.

The political cost became visible through named actors rather than through abstraction. Newt Gingrich, the House Minority Whip, treated the concession as a betrayal of the Republican electoral position, and accounts of the episode described Gingrich ending a telephone call from White House Chief of Staff John Sununu in anger after learning of the decision. Vice President Dan Quayle and members of the Senate Republican leadership also objected to the reversal. Senator Trent Lott questioned the change, and Sununu responded in public by diminishing Lott, a response that illustrated the strain inside the party. Ed Rollins, a co-chair of the Republican National Committee, advised Republican candidates in a memorandum to distance themselves from the President if electoral survival required distance, and Rollins lost his position after the memorandum became known. The New York Post captured the attack from outside the party in a headline published the day after the statement, a headline that accused Bush of having lied during the campaign. Together, the reactions established that the concession would cost Bush support among Republicans whose help the administration needed to pass the very package the concession had made possible.

Democratic negotiators carried exposure of a different kind. Democrats held majorities in both chambers of the 101st Congress, and Senate Majority Leader George Mitchell and Speaker of the House Thomas Foley led a negotiation whose product would require Democratic votes in large numbers for Medicare savings and domestic restraint that Democratic campaigns had promised to resist. The summit structure spread responsibility across both leaderships, and that spreading of responsibility was part of its value.

Negotiators announced agreement on September 30, 1990, one day before fiscal year 1991 began. The announced package promised about 500 billion dollars of deficit reduction over five years and about 40 billion dollars in the first year. About 134 billion dollars of the five year total came from taxes. Medicare savings formed a major part of the mandatory spending component, alongside savings in agriculture and other entitlement programs, and discretionary spending restraint supplied another large component, divided among defense, international, and domestic programs. Administration talking points dated October 1, 1990 divided the total among discretionary spending, entitlement and mandatory programs, and taxes, with debt service effects completing the arithmetic. The agreement also contained the enforcement architecture that became the Budget Enforcement Act. Discretionary caps would hold appropriations within negotiated limits, and a pay-as-you-go requirement would hold new mandatory spending and revenue legislation to deficit neutrality. Enforcement allowed negotiators to argue that the promised savings would survive later sessions of Congress, because caps, scorekeeping, and sequestration gave future restraint a legal form.

Speed then became the enemy of consent. Leaders could announce totals, categories, and enforcement rules before committees had converted the totals into legislative text that every member could examine. A continuing resolution funded the government through October 5, 1990 and suspended the sequester through the same date, giving the House a narrow window in which to approve the budget resolution embodying the agreement. Urgency compressed the time available to build a majority. Members who might have accepted a painful package after consultation or a visible concession to a constituency had little time to obtain either, while opposition required only a coalition large enough to say no.

What happened in the days after the House defeated the package?

The House rejected the package because its majority was assembled from opposite objections. Conservative Republicans led by Minority Whip Newt Gingrich opposed its tax increases, liberal Democrats opposed its Medicare savings and the distribution of its burdens, and Minority Leader Robert Michel’s support could not hold enough Republicans to offset defections from both wings.

The vote arrived in the early morning of October 5, 1990. The House rejected the agreement by 179 to 254. Contemporary reporting in Education Week described the rejection as overwhelming and emphasized that the coalition against the package cut across party lines. Conservative Republicans objected to the 134 billion dollars in tax increases over five years and to the abandonment of the no new taxes pledge that the tax component embodied. Liberal Democrats objected that the tax increases, Medicare reductions, and Medicare premium increases placed unfair burdens on lower income households and older Americans. Between those poles stood the leadership of both parties and the President, holding a package that each flank could portray as a concession to the other side. The agreement carried the burden of every compromise it contained, while the opposition carried none of the burden of an alternative.

Robert Michel, the House Minority Leader, supported the deal and worked for its passage. Gingrich, serving as Michel’s Whip, broke with Michel in public and led the conservative rebellion against a package negotiated by a Republican President. The split carried institutional meaning beyond a single vote, because a Whip who organizes against the Leader’s position announces that the party contains two competing theories of governance. Michel’s theory treated the agreement as the best settlement available under divided government and treated presidential credibility as an asset worth protecting. Gingrich’s theory treated the tax reversal as a threat to the Republican brand and treated refusal as the means of forcing a different settlement. The October 5 vote settled the immediate contest. Gingrich assembled enough Republican defections, in combination with Democratic defections from the left, to defeat a package that Michel and Bush had endorsed.

Defeat produced an immediate governing crisis because the continuing resolution expired on October 5 and the sequester suspension expired with the funding. Congress passed another continuing resolution intended to extend funding, and President Bush vetoed the measure. In the veto message for House Joint Resolution 660, which would have extended funding through October 12, 1990, Bush objected that the resolution would suspend the Gramm-Rudman-Hollings sequester even though Congress had failed to act on deficit reduction. The message restated the position Bush had announced on September 30, that a further continuing resolution would not be signed until a satisfactory budget resolution had passed. The veto converted a legislative defeat into a funding lapse. Federal appropriations for fiscal year 1991 had not been enacted for most agencies, the temporary funding law had expired, and the President had rejected the replacement that Congress sent. The lapse ran over the Columbus Day weekend, from Saturday, October 6 through Monday, October 8, 1990. National parks closed during the weekend. Smithsonian museums in Washington closed. Federal employees faced furlough uncertainty, and visitors encountered locked doors at institutions that symbolized the federal presence in ordinary life. The lapse ended after a further continuing resolution was signed on October 9, 1990, restoring funding while negotiators returned to the task that the October 5 vote had left unfinished.

The shutdown weekend clarified the bargaining position of each side. Bush had demonstrated a willingness to accept a visible funding lapse rather than sign temporary funding detached from a budget settlement. Congressional leaders had demonstrated that the September 30 totals could not pass the House in the form first presented. Renegotiation therefore had to move the package at both margins at once, giving conservatives a reason to reduce their opposition while giving liberals a reason to accept a package that retained tax increases, Medicare savings, and discretionary restraint in substantial form. The overall size of the package survived the renegotiation in broad terms, because abandoning the five year total would have surrendered the rationale for the summit. The composition changed around that total.

The Congressional Research Service and the Congressional Budget Office later described the enacted reconciliation act as reducing the deficit by about 482 billion dollars over five years, a figure close to the announced summit ambition and inclusive of revenue increases, spending reductions, and debt service savings. The fact sheet totals preserved for the summit and for enactment tell a story of adjustment rather than abandonment. Negotiators softened the tax and Medicare pieces that had drawn the sharpest objections during the first House vote, and leadership shifted allocations between defense and domestic discretionary spending as the appropriations consequences of the agreement were worked through. Specific claims about individual rates and premium shares require care, because summit proposals, committee products, and enacted provisions differed at several stages during October, and a figure attached to one stage cannot be transferred safely to another. The enacted tax provisions are clearer in the statutory record. The Omnibus Budget Reconciliation Act of 1990 raised the top statutory individual income tax rate from 28 percent to 31 percent, according to Congressional Research Service summaries of the individual income tax, and the act limited the value of itemized deductions for higher income taxpayers by reducing allowable deductions in relation to adjusted gross income above a threshold of 100,000 dollars. The enacted law also raised the individual alternative minimum tax rate and capped the maximum rate on capital gains at 28 percent. Those enacted provisions, rather than any unattributed comparison between drafts, establish the direction of the final settlement. Revenue remained in the package. The form of the revenue changed as negotiators searched for a combination that a House majority could accept.

Medicare followed a comparable path. Savings from Medicare remained necessary to the arithmetic of a package measured in hundreds of billions of dollars, and renegotiation reduced the exposure created by the first package without removing Medicare from the total. A member who had rejected the first package over Medicare could support a revised package by pointing to moderated provisions. The final vote therefore depended in part on whether enough members concluded that continued refusal, after a funding lapse and a presidential veto, carried greater risk than acceptance.

The legislative vehicle for the settlement was reconciliation. The Omnibus Budget Reconciliation Act of 1990, introduced in the House as House Bill 5835 by Representative Leon Panetta of California on October 15, 1990, gathered the revenue, mandatory spending, credit reform, and enforcement provisions into a single measure enacted under the reconciliation instructions for fiscal year 1991. The House passed the bill on October 16, 1990 by 227 to 203, according to the action record maintained by Congress. The Senate passed the measure on October 19, 1990 by 54 to 46, in place of the Senate companion measure. Conference negotiations then resolved the differences between the chambers during the following week. The conference report, House Report 101-964, reached the floors of both chambers on October 27, 1990. The House agreed to the conference report by 228 to 200. The Senate agreed to the conference report the same day by 54 to 45, according to the action record for House Bill 5835 maintained through Congress.gov. The Senate conference tally differed by one vote from the earlier Senate passage tally, a distinction that matters for accuracy because the two votes decided different questions at different stages. Passage of the bill and agreement to the conference report were separate actions, separated by eight days and by the work of the conference committee.

The margins revealed the character of the final coalition. A House margin of 28 votes and a Senate margin of 9 votes represented passage without consensus and without the bipartisan breadth that summit negotiators had sought on September 30. The final coalition combined the congressional leadership of both parties, the President, and enough members from the middle of each party to overcome the combined opposition of the conservative Republican right and a portion of the Democratic left. The difference from October 5 lay in the adjustments made during renegotiation, in a funding lapse that raised the cost of a second failure, and in the pressure of an approaching end to the session.

President Bush signed the Omnibus Budget Reconciliation Act of 1990 on November 5, 1990. The act became Public Law 101-508, published at 104 Stat. 1388. Title XIII of the act contained the Budget Enforcement Act of 1990, the enforcement regime built around discretionary caps and pay-as-you-go discipline. Signature completed a process that had begun with the collapse of confidence in fixed deficit targets and had passed through a summit, a reversed tax pledge, a defeated agreement, a veto, and a funding lapse. The sequence explains why the enforcement design took the shape it took. Negotiators who had watched fixed targets miss and bend built a system intended to preserve a negotiated reduction rather than to command an outcome that economic events could falsify. The summit supplied the reduction, and the statute supplied the preservation mechanism.

The Caps: Walls Around the Appropriations Process

The first instrument of the Budget Enforcement Act of 1990 was a set of statutory limits on discretionary spending, the spending that Congress provides year by year through appropriations acts rather than through permanent law. The Congressional Research Service, in Report R41901, Statutory Budget Controls in Effect Between 1985 and 2002, described the caps as the half of the 1990 design that governed the appropriations process directly, while pay-as-you-go governed new legislation affecting direct spending and revenues. The distinction mattered because discretionary programs, from military operations to foreign aid to parks, housing, and scientific research, received funding through annual decisions that appropriators could measure, total, and compare against a number written into statute. A limit of that kind converted the 1990 budget agreement from a political understanding into a test that each enacted appropriations bill either passed or failed.

Each category carried two limits rather than one, a ceiling on budget authority and a separate ceiling on outlays, and the pairing shaped behavior in ways that a single ceiling could not have achieved. Budget authority is permission to incur obligations, the legal capacity to sign contracts, hire personnel, and make grants, while outlays are the payments that flow out of the Treasury as those obligations are liquidated. An appropriations act could provide budget authority that produced outlays slowly over several years, as with shipbuilding or major construction, or authority that spent out quickly, as with salaries and operating expenses. A ceiling on budget authority alone would have left Congress free to accelerate spending from authority provided in earlier years, and a ceiling on outlays alone would have permitted enormous commitments whose costs landed beyond the enforcement horizon. By limiting both measures, the statute closed the two main channels through which the total could be evaded, and appropriators had to fit each bill inside a pair of constraints that reflected both the commitments made in a year and the cash consequence of commitments made in that year and in prior years. The Congressional Budget Office tracked both measures in its scorekeeping through the 1990s, and its study Supplemental Appropriations in the 1990s, published in March 2001, recorded that the statutory limits applied to budget authority and to outlays alike.

For fiscal years 1991 to 1993, the statute divided discretionary spending into three categories, defense, international, and domestic, and placed a separate pair of limits on each category. Defense covered the military functions of the Department of Defense and related atomic energy defense activities. International covered foreign assistance, State Department operations, and other international affairs programs. Domestic covered the broad remainder of annually funded government, from transportation and housing to education, health research, law enforcement, and natural resources. The division mirrored the structure of the 1990 summit agreement itself, which had allocated the pain and the protection of deficit reduction across those same three areas. Separate caps meant that the agreement could not be satisfied in the aggregate while being quietly rewritten in its parts. A year in which defense spending fell below its ceiling produced no usable room for domestic appropriations, because the domestic ceiling stood on its own figure in the statute. The categories functioned as compartments with sealed bulkheads, and each appropriations subcommittee worked inside a compartment whose boundaries the enacted law, rather than committee preference, had fixed.

The walls between the categories gave the early caps their distinctive force, and the walls rested on a deliberate distrust of cross category trade offs. Negotiators at the 1990 summit had constructed a bargain in which defense spending would decline from Cold War levels while domestic programs received defined, limited room, and each side feared that the other side would raid the settlement after the headlines faded. Administration officials and defense oriented lawmakers worried that savings from the post Cold War drawdown would be harvested to finance domestic expansions, converting a deficit reduction agreement into a transfer agreement. Domestic oriented lawmakers held the mirror image fear, that a defense establishment resisting the drawdown would reclaim resources by squeezing the domestic category whenever an international crisis or a weapons program overrun created pressure. Walls answered both fears at once. Because transfers between categories were barred, the defense drawdown bargain held as a matter of law rather than goodwill, and a raid across categories required an affirmative amendment of the statute, visible to every participant and subject to presidential signature or veto, instead of a quiet reallocation inside an omnibus bill. The Congressional Research Service account in Report R41901 treated the separate categories of fiscal years 1991 to 1993 as the defining structure of the original caps, before consolidation into a single discretionary total for fiscal year 1994 under the framework extended by the Omnibus Budget Reconciliation Act of 1993.

Enforcement matched the compartmental design, because a breach in one category triggered sequestration confined to that category alone. Sequestration under the deficit control act meant the cancellation of budgetary resources by a uniform percentage across nonexempt accounts, ordered by the President after the Office of Management and Budget reported a breach. If enacted domestic appropriations exceeded the domestic ceiling, the resulting order cut domestic accounts, and defense and international accounts stood outside the calculation entirely. A domestic breach therefore could not be cured by cutting defense, and a defense breach could not be shifted onto foreign aid. The confinement rule served two purposes. First, confinement preserved the bargain, since the category that overspent bore the correction, and no coalition could overspend in one compartment while expecting the penalty to fall in another compartment whose supporters lacked the votes to resist. Second, confinement made the penalty predictable, since appropriators could calculate which accounts faced exposure if a bill crossed a ceiling. Exemptions written into the deficit control act removed specified accounts from sequestration, and the uniform percentage applied to the nonexempt base that remained within the breached category.

What happens when a spending cap is breached?

A breach in one category triggers a sequestration order confined to that category. The Office of Management and Budget calculates the excess, and the President orders uniform percentage reductions across nonexempt accounts in the breached category sufficient to eliminate the excess in both budget authority and outlays.

The calculating authority under the caps was the Office of Management and Budget, and the statute organized its work into a rhythm of three sequestration reports across each budget cycle. The preview report accompanied the submission of the President budget for the coming fiscal year and set out the cap levels for that year, together with the adjustments the statute authorized, so that Congress began the appropriations season with the administration official statement of the limits that would govern enforcement. The update report followed in August, in the middle of the appropriations process, restating the limits with adjustments recorded through that point in the year and giving appropriators a midcourse measurement while bills remained amendable. The final report, issued within fifteen days after Congress adjourned to end a session, compared enacted appropriations against the adjusted limits for the fiscal year and determined whether a breach had occurred. The Congressional Research Service summary of the process in its budget process materials described adjustments as occurring three times per year, in the preview report in the President budget, in the August update report, and in the final report after adjournment. The sequence converted enforcement from a single year end confrontation into a running disclosure system, and a committee drifting toward a breach received two formal warnings, each tied to the same statutory figures, before any order issued.

The President sequestration order followed the final report as a ministerial act rather than a discretionary policy choice. When the Office of Management and Budget reported a breach, the statute directed the President to issue an order eliminating the excess, and the order applied the uniform percentage that the report had calculated to the nonexempt accounts in the breached category. The design deliberately separated measurement from mercy. The Office of Management and Budget controlled the estimates, consulting Congressional Budget Office estimates of appropriations legislation under the procedures of the deficit control act, while the President signature executed a result that the numbers had produced. Neither branch could waive a breach by silence, because the report deadlines ran from events, adjournment and enactment, rather than from administrative convenience. Congress retained the sovereign remedy of amending the caps by law, and later Congresses used that remedy on several occasions, but an amendment required passage through both chambers and presentation to the President, which kept relief inside the same constitutional process that had created the limits.

The first order under the act demonstrated both the speed of the machinery and the dependence of the machinery on accurate estimates. On November 9, 1990, four days after President George H. W. Bush signed the Omnibus Budget Reconciliation Act of 1990, the President issued a sequestration order intended to close a breach of 395 million dollars in the international category for fiscal year 1991. The breach had emerged as Congress completed action on the appropriations bills for the year that had begun on October 1, 1990, under the compressed calendar that the summit and the funding lapse had produced. The Office of Management and Budget then discovered an error in the estimates underlying the determination. The office corrected the error, the November 9 order was rescinded, and the sequestered funds were restored. The Congressional Budget Office history in Supplemental Appropriations in the 1990s recorded the episode as the first sequestration under the new legislation, followed by correction of the error, rescission of the order, and restoration of the funds. The episode carried a double lesson. Enforcement was real enough to operate within days of enactment, and enforcement was estimate driven enough that a single scoring mistake could move hundreds of millions of dollars and then move those dollars back again.

Statutory adjustments kept the caps from becoming a brittle instrument that genuine emergencies would shatter, and the adjustment provisions formed the safety valve of the entire structure. Section 251 of the deficit control act, as amended in 1990, authorized the Office of Management and Budget to raise the caps by specified amounts for specified events, recorded in each of the three reports. The largest adjustment covered emergency requirements. Appropriations designated as emergency requirements in law, with the concurrence of the President, raised the caps by the amount of the designated spending, so that the spending stood outside the limits for breach purposes. Either branch could initiate a designation, Congress through statutory language in an appropriations act and the President through the budget or another formal statement, but a valid designation required agreement by both branches, a requirement that prevented either Congress or the President from unilaterally relabeling ordinary spending as an emergency. Disaster relief reached the caps principally through that channel, since appropriations for the Disaster Relief Fund and related accounts after hurricanes, earthquakes, and floods were routinely designated as emergency requirements during the 1990s, and the designation, once made by both branches, produced a matching cap adjustment.

Other adjustments were narrower and more technical, and each one reflected a judgment that a particular category of spending should not consume room intended for ordinary program decisions. Congressional Research Service summaries of the 1990 framework listed adjustments for changes in budgetary concepts and definitions, for specified inflation differences in fiscal years 1993 and 1994, for allowances tied to credit subsidy reestimates under the new credit reform accounting, for Internal Revenue Service tax compliance initiatives funded to raise revenue, and for specified debt forgiveness for Egypt and Poland connected to the diplomacy that followed the Persian Gulf War and the end of the Cold War. Each adjustment shared a common logic. The caps were meant to discipline choices, and an adjustment removed from the breach calculation those amounts that did not represent a choice to expand the ordinary operations of government beyond the summit settlement. An emergency designation differed from the technical adjustments in scale and in politics, since the emergency channel had no fixed ceiling and depended on the willingness of both branches to apply the label, a dependence that later debates over supplemental appropriations in the 1990s examined at length. The formal structure, however, remained constant through the decade. Designated emergency appropriations were added outside the caps by adjustment, undesignated appropriations counted in full, and the Office of Management and Budget reported the adjustments as part of the official cap figures rather than as informal exceptions.

The caps also interlocked with the older machinery of the Congressional Budget Act of 1974, and the interlock explains how a statutory limit administered by the executive branch shaped drafting decisions inside congressional committees months before any sequestration report appeared. Under section 302 of the Congressional Budget Act, the annual budget resolution allocated spending totals among the committees of each chamber, and the allocation to the Appropriations Committees, known as the section 302(a) allocation, set the total that the appropriations bills of that chamber could provide. The Appropriations Committees then subdivided the allocation among their subcommittees through section 302(b) suballocations, and each suballocation operated as an enforceable ceiling for the bill within that subcommittee jurisdiction, backed by points of order on the floor of each chamber. For the system to function, the section 302(a) allocation had to fit inside the statutory caps. A budget resolution that allocated more than the caps allowed would have set the Appropriations Committees on a collision course with sequestration, since bills written to the allocation would breach the statute. Budget Committees therefore constructed the resolution aggregates and the section 302(a) allocation with the adjusted cap figures from the preview report in view, and appropriators subdivided a total that already embodied the statutory constraint. Points of order under sections 302 and 311 of the Congressional Budget Act gave members an in session remedy against bills or amendments that exceeded allocations, so enforcement operated at three levels, committee drafting against suballocations, floor objections against allocation breaches, and sequestration against enacted breaches of the statute. The statutory caps sat at the top of that hierarchy as the one level that no chamber rule, waiver practice, or committee agreement could alter, because only a law could amend a law.

A further design choice separated the 1990 caps from the deficit target regime that the statute replaced, and the choice concerned the treatment of estimates that changed for reasons outside congressional action. Under the Gramm-Rudman-Hollings framework of the Balanced Budget and Emergency Deficit Control Act of 1985, enforcement ran against fixed maximum deficit amounts, and the deficit was an outcome that economic performance, revenue collections, interest costs, and technical estimating changes all influenced. A recession that lowered revenues or a technical reestimate that raised projected entitlement outlays could push the projected deficit through the target and toward sequestration even in a year when Congress had enacted no new legislation at all, a feature that summit negotiators in 1990 regarded as a central defect of the earlier design. The caps inverted the relationship. The measured quantity was enacted discretionary appropriations, a product of congressional action, compared against limits that the Office of Management and Budget adjusted only through the channels the statute specified. Economic deterioration that raised mandatory spending or reduced receipts changed the deficit, sometimes by large amounts, without changing the appropriations total that the caps measured, and technical reestimates of mandatory programs belonged to the pay-as-you-go and baseline side of the budget rather than to a cap breach determination. The Congressional Research Service characterized the 1990 shift as a move from controlling budgetary outcomes to controlling congressional action on legislation, and the caps embodied that characterization in mechanical form. Appropriators answered for the numbers that appropriators produced.

Outlay estimates required care inside that design, because outlays from enacted budget authority depended on spend out rates that estimators projected and revised. A breach determination therefore rested on estimates, as the November 1990 episode proved, but the estimates concerned the budgetary effects of appropriations legislation that Congress had enacted. Congress could respond to an adverse estimate by rescinding authority, by reducing a later bill, or by legislating a cap change, and each response was again an act of legislation rather than an argument about the weather of the economy.

Timing rules for supplemental appropriations completed the enforcement scheme, because supplementals were the appropriations most likely to arrive after the regular bills had consumed the available room under a ceiling. The statute distinguished between the end of session determination and a within session sequestration. The ordinary course ran through the final report after adjournment, when the Office of Management and Budget measured the full set of enacted appropriations for the fiscal year against the adjusted limits. A supplemental appropriation enacted during the session for the fiscal year in progress, after adjournment of the prior session and before July 1 of that fiscal year, received faster treatment. If such a supplemental caused a breach in a category, sequestration followed fifteen days after enactment, confined to the breached category and calculated after taking account of any prior sequestration in that category. The within session rule prevented a spring supplemental from spending a full season beyond the ceiling while enforcement waited for adjournment, and the July 1 boundary reflected the practical judgment that supplementals enacted late in a fiscal year, with only weeks left before the year closed on September 30, belonged in a different accounting treatment. For appropriations enacted after June 30 that caused a breach, the statute applied a look back rule under which the limits for the affected category in the following fiscal year were reduced by the amount of the breach, so that late year excess spending borrowed against the room available in the next year rather than escaping measurement altogether.

The within session mechanism operated at least once in the form the statute contemplated. The Congressional Budget Office recorded that Public Law 102-27, enacted on April 10, 1991, provided supplemental appropriations that exceeded the domestic discretionary cap then in place by 2.4 million dollars. The determination triggered a sequestration on April 25, 1991, fifteen days after enactment, reducing budget authority in each nonexempt domestic account by 0.0013 percent. The amounts were minute beside the totals of the federal budget, and the episode demonstrated a property of automatic enforcement that discretionary enforcement rarely possesses. A threshold that yields to small breaches invites larger breaches, because each exception teaches appropriators that the ceiling bends under pressure, while a mechanism that sequesters for 2.4 million dollars establishes that the ceiling is a property of the statute rather than a negotiating position of the administration. The same CBO study recorded that seven supplemental laws of the 1990s were enacted after July 1 and that the Office of Management and Budget determined that none of those laws breached the caps then in place, which meant that the look back reduction, like the end of session sequestration for regular bills, functioned principally as a deterrent whose existence shaped drafting before enactment.

Taken together, the cap mechanics formed a closed system with a defined input, a defined measurement process, and a defined consequence. The input was enacted discretionary appropriations, sorted into the categories the statute named for fiscal years 1991 to 1993 and measured in both budget authority and outlays. The measurement process belonged to the Office of Management and Budget, exercised through preview, update, and final reports, disciplined by statutory adjustments for emergencies and other designated events, and connected to congressional drafting through the section 302 allocations of the Congressional Budget Act. The consequence was sequestration inside the breached category, delivered by presidential order at the end of a session for regular appropriations and within fifteen days for a breaching supplemental enacted before July 1, with a look back reduction carrying late year breaches into the limits of the following year. Every element pointed back to the central decision of 1990, the decision to constrain the action rather than the outcome. The caps promised no particular deficit figure in any particular year. The caps promised only that the annual appropriations process, the most regular and controllable spending decision the federal government made, would occur inside walls that Congress and the President had built in statute and could dismantle only in statute.

Pay-As-You-Go: A Scorecard for Legislation

The second instrument created by the Budget Enforcement Act of 1990 approached the deficit from the opposite direction to the discretionary spending limits. The limits governed money provided through annual appropriations acts, a flow that Congress revisited every year. Pay-as-you-go governed the other channel through which Congress changed the fiscal position by statute, namely legislation affecting direct spending and revenues. Direct spending, also called mandatory spending, flowed from eligibility rules and benefit formulas written into permanent law rather than from annual appropriations, and revenue law set tax rates, deductions, credits, and collections under the Internal Revenue Code. A change in either body of law altered federal deficits for years after enactment without any further vote. The pay-as-you-go requirement, recorded in the enforcement titles of the 1985 deficit control act as rewritten in 1990, answered that persistence with a ledger. Every enacted law that changed direct spending or revenues received an entry, the entries accumulated, and the accumulated balance determined whether automatic cuts followed at the end of the session. The Congressional Research Service described the governing idea as a shift away from enforcing deficit outcomes and toward controlling congressional action on legislation, a design summarized in the claim that the statute would constrain the action, not the outcome.

The attachment point of the rule defined its reach. Pay-as-you-go attached to enacted legislation, and only to enacted legislation. A proposal, a committee draft, a passed bill awaiting a presidential signature, and a vetoed measure all stood outside the scorecard, because none of them had changed the law against which deficits were measured. Enactment converted an estimate into an obligation of record, and the Office of Management and Budget entered the estimated effect of the law on the scorecard. The rule also attached only to two categories of legislative effect. A law that increased direct spending, for example by expanding eligibility for a benefit program or creating a new entitlement, produced a cost entry. A law that reduced revenues, for example by cutting a tax rate or enlarging a credit, likewise produced a cost entry, because a revenue loss and a spending increase widened the deficit by the same arithmetic. A law that reduced direct spending or raised revenues produced a savings entry. Laws that did none of those things, including ordinary appropriations acts governed by the discretionary limits, generated no entry at all.

Equally important was the class of fiscal change that the rule left untouched. The scorecard recorded the estimated effects of legislation measured against the law in force before enactment, and the scorecard recorded nothing else. Growth in spending under a program whose authorizing statute remained unchanged generated no entry, even when caseloads rose sharply during a recession. A shortfall in revenues caused by an economic contraction generated no entry, even when collections fell far below forecast. Interest costs, which responded to debt levels, interest rates, and Treasury financing decisions rather than to any single direct spending or revenue statute, stood outside the mechanism, as did baseline drift of every kind. The Congressional Research Service emphasized that pay-as-you-go did not seek to control the effects of any direct spending or revenue law already in effect, and that limitation was structural rather than incidental. Congress in 1990 had paired the enforcement statute with a reconciliation act that the Congressional Research Service projected would reduce the deficit by about 482 billion dollars over five years, and the enforcement problem was the risk that later statutes would surrender portions of that negotiated reduction piece by piece. A ledger of legislative effects addressed precisely that risk, because each surrender would arrive in the form of an enacted law with an estimable cost.

The ledger itself was the rolling five-year pay-as-you-go scorecard maintained by the Office of Management and Budget. Five years matched the horizon over which the 1990 reconciliation act had been scored and over which the discretionary limits initially ran, and the scorecard preserved that horizon as a moving window. When a law was enacted, the Office of Management and Budget recorded the estimated deficit effect of the law for the fiscal year of enactment and for the succeeding years within the window, so a single statute could leave entries spread across several fiscal years rather than a single lump sum. A tax provision phased in over three years entered according to the estimated revenue loss in each year, and a benefit expansion with a delayed effective date entered according to the estimated outlays in each year after the effective date arrived. As one fiscal year closed and another opened, the window rolled forward, and balances recorded in earlier sessions carried into the calculation for later sessions within the window. Legislation enacted during a session after the final sequestration report for that session had been issued escaped nothing by that timing accident, because the statute directed that the budgetary effects of such legislation be recorded on the scorecard in the following session.

Against that ledger the statute set a single test. The cumulative balance of all recorded effects for a fiscal year, combining the legislation of the session just ended with any balance carried forward from legislation of earlier sessions inside the window, had to be zero or negative. A balance of zero meant that the costs of new direct spending and revenue legislation had been matched exactly by savings from other new legislation, whether the savings came from direct spending reductions or from revenue increases. A negative balance meant that enacted legislation had, on net, reduced the deficit relative to prior law, and a credit balance of that kind provided headroom against which later costs inside the window could be recorded. A positive balance, called a debit, meant that Congress had enacted net deficit-increasing legislation without offsets, and a debit was the sole condition that activated the enforcement sanction. The test applied on an aggregate and annual basis rather than bill by bill, so one statute could increase the deficit in isolation provided that other statutes enacted within the relevant period supplied offsetting savings of equal or greater size. Enforcement waited until the session ended, a design that gave Congress the whole session to assemble offsets.

Does PAYGO require a balanced budget?

No. The requirement governed only the estimated deficit effect of newly enacted legislation affecting direct spending and revenues. A debit balance on the scorecard triggered cuts, while a preexisting deficit left untouched by new law triggered nothing. Deficit neutrality for new law and a balanced budget were separate conditions under the statute.

The distinction between deficit neutrality for new legislation and balance in the overall budget deserves emphasis, because the two conditions were persistently confused in public discussion of the 1990 statute. A federal budget in balance would have required total revenues to equal total outlays in a fiscal year, a condition that depended on the entire inherited structure of spending and taxation, on economic performance, on demographics, and on interest rates. Pay-as-you-go required none of that. Historical tables published by the Office of Management and Budget recorded a federal deficit of about 221 billion dollars in fiscal year 1990, before the enforcement regime took effect, and a deficit of about 164 billion dollars in fiscal year 1995. Neither deficit, standing alone, violated the pay-as-you-go requirement, because the requirement asked only whether legislation enacted during the covered period had added to the deficit relative to the law that preceded enactment. Conversely, a fully compliant scorecard could coexist with a large and growing overall deficit when spending under preexisting law expanded or when revenues sagged for economic reasons.

Two institutions divided the work of estimation and enforcement, and the division mattered to the credibility of the ledger. The Congressional Budget Office supplied cost estimates for legislation as Congress considered and enacted measures, drawing on the estimating responsibilities that the Congressional Budget Act of 1974 had assigned to that office. Committees received Congressional Budget Office estimates during drafting, and the estimates gave legislators advance knowledge of the entries that enactment would produce. The Office of Management and Budget, an executive branch agency inside the Executive Office of the President, kept the official scorecard, made the official determinations of budgetary effect for enforcement purposes, and issued the sequestration reports on which any automatic cuts depended. That allocation placed the estimate that guided legislators and the determination that bound the government in different hands. Administrations and Congresses disputed scorekeeping questions repeatedly during the years the regime operated, but the statute assigned the final enforcement reading to the Office of Management and Budget.

The enforcement calendar converted the ledger into a sanction through a fixed sequence of reports and orders at the end of each session. The Office of Management and Budget reported on sequestration three times in the annual cycle under the deficit control act as rewritten in 1990. A preview report accompanied the submission of the budget of the President early in the calendar year, an update report followed in August, and a final sequestration report closed the sequence. The final sequestration report was due within 15 calendar days after Congress adjourned at the end of a session. In that report the director of the Office of Management and Budget determined whether the scorecard carried a debit for the fiscal year just ended or for the preceding fiscal year that legislation of the session had produced. When the final report showed such a debit, the President was required to issue a sequestration order immediately, and the statute placed any pay-as-you-go sequester on the same day as any sequester required for enforcement of the discretionary spending limits, so that the two sanctions, when both applied, took effect together. The order permanently canceled budgetary resources in nonexempt direct spending programs in an amount calculated to eliminate the debit. No further vote, and no appropriation, stood between the report and the cancellation.

The sequester that a debit triggered fell on nonexempt mandatory programs according to a uniform percentage calculation rather than according to any judgment about the merits of individual programs. The Office of Management and Budget computed the percentage reduction required to produce savings equal to the debit from the base of nonexempt direct spending available for reduction, and the same percentage applied across the covered accounts. A program whose authorizing legislation had supplied offsets that kept other bills compliant received the same percentage cut as a program whose growth had contributed to the debit, because the sequester operated on accounts rather than on responsibility. The cancellation of budgetary resources was permanent rather than a deferral, so amounts sequestered did not return to the accounts in a later year, and that combination of uniformity, automaticity, and permanence explains why the threat of a sequester influenced behavior even in sessions when no sequester occurred.

Congress shielded large parts of the mandatory budget from that sanction, and the pattern of exemptions expressed the priorities of the statute. The exemption list in section 255 of the Balanced Budget and Emergency Deficit Control Act placed Social Security benefits outside sequestration entirely, apart from administrative expenses, which remained subject to reduction. Medicaid, the joint federal and state health program for people with low incomes, likewise stood exempt from a pay-as-you-go sequester, as the Congressional Research Service recorded in summaries of both the 1990 procedures and the later statutory pay-as-you-go framework that retained the same protections. Railroad retirement benefits stood exempt, as did veterans compensation and veterans pensions. Payments of net interest on the federal debt could not be reduced, and payments under the earned income tax credit were exempt as well. Beyond the named programs, sections 255 and 256 of the deficit control act together carried a wider structure of protection under which federal retirement benefits and a broad range of programs directed at people with low incomes were exempt from reduction or governed by special rules that limited exposure, so the sequester base consisted disproportionately of the remaining nonexempt accounts. Administrative expenses remained subject to reduction even for programs whose benefits were exempt.

Medicare occupied an intermediate position between full exposure and exemption, governed by a special rule rather than by the uniform percentage alone. Reductions in Medicare spending under a pay-as-you-go sequester were limited by statute to 4 percent. The Congressional Research Service stated that limit in its report on statutory budget controls in effect between 1985 and 2002, in its report on the termination of the pay-as-you-go requirement for fiscal year 2003 and later years, and in its analyses of Medicare and budget sequestration, which described the 4 percent ceiling as the limit applicable to Medicare benefit payments under a statutory pay-as-you-go sequestration order. The ceiling operated as a cap on the percentage reduction applied to Medicare, not as a guarantee of any particular dollar amount of savings. When the uniform percentage required to eliminate a debit exceeded 4 percent, Medicare absorbed reductions up to the ceiling and the remaining accounts absorbed a recomputed and higher percentage so that total savings equaled the debit. When the required uniform percentage fell below 4 percent, Medicare took the same percentage as every other nonexempt account.

Between the two instruments of the 1990 act, the statute built a firewall, and the logic of the firewall prevented each instrument from being gamed against the other. The discretionary limits and pay-as-you-go measured different actions, kept different accounts, and triggered different sequesters. Savings achieved under the discretionary limits, for example when appropriations came in below a ceiling, generated no credit on the pay-as-you-go scorecard and could not cancel a pay-as-you-go debit. Savings recorded on the pay-as-you-go scorecard, for example when a revenue increase exceeded the cost of a new benefit, raised no discretionary ceiling and created no room for additional appropriations. A breach of a discretionary category before fiscal year 1994 produced a sequester confined to that category, while a pay-as-you-go debit produced a sequester confined to nonexempt direct spending, and neither calculation borrowed savings from the other enforcement account. The firewall foreclosed that maneuver by assigning each form of legislative action to a single ledger with a single sanction.

One category of legislation escaped the scorecard by design rather than by accounting, through the emergency designation. Legislation designated as an emergency requirement stayed off the pay-as-you-go scorecard, and the estimated effects of such legislation entered neither the cost column nor the savings column of the ledger. The escape hatch required agreement across the separation of powers. Congress had to designate the provision as an emergency requirement in the statute itself, and the President had to designate the provision as an emergency requirement as well, so neither branch acting alone could exempt legislation from the neutrality rule. A congressional designation that the President declined to match left the legislation on the scorecard, and a presidential preference for emergency treatment that Congress declined to write into law likewise left the legislation on the scorecard. The double designation ensured that exempting a cost from the ledger was itself a recorded act of both branches.

The scorecard changed the drafting of legislation long before any session ended, because the end-of-session test was knowable in advance and committees wrote bills with the ledger in view. An authorizing committee that reported a bill expanding a mandatory program faced a concrete question that the process before 1990 had posed only in general terms, namely which provisions would supply the offset. Offsets could be found inside the same bill, through reductions in another mandatory program within the jurisdiction of the committee, through revenue provisions, or through negotiated savings in companion legislation expected to move during the same session. Drafting practice adapted accordingly. Bills arrived on the floor with offset titles already assembled, phased effective dates arranged to fit estimated costs inside the five-year window, and sunsets or delayed expansions calibrated to the Congressional Budget Office estimate that the text would receive. Provisions whose estimated costs resisted offsetting were narrowed, postponed to later sessions, or abandoned in committee, and those casualties of the scorecard formed a large part of the practical effect of the regime.

Aggregation across a session amplified the behavioral effect. Because the test weighed the cumulative record rather than any single enactment, a savings bill enacted early in a session created headroom that a later bill could consume, and a costly bill enacted early imposed an obligation on the remainder of the session that later bills had to retire. Committee chairs and party leaders therefore tracked the running balance as a session asset, and the balance entered negotiations over the order in which bills would move. A popular expansion could be scheduled after the savings measure that would pay for the expansion had cleared both chambers, reducing the risk that the session would close with the cost enacted and the offset stranded. Reconciliation legislation, which combined many direct spending and revenue provisions in a single measure under expedited procedures, interacted with the scorecard as a large net entry rather than as a collection of separate entries, so a reconciliation act that reduced the deficit on net added credit to the ledger even though individual titles inside the act increased spending. The 1990 reconciliation act itself stood apart from that treatment, because the enforcement regime existed to preserve the reduction that the act achieved rather than to spend the reduction as credit against later legislation.

Taken together, the mechanics of the original pay-as-you-go requirement formed a coherent enforcement theory. The rule attached to enacted legislation rather than to economic outcomes, so responsibility for a violation could always be traced to identifiable votes on identifiable statutes. The scorecard accumulated those statutes over a rolling five-year window, so timing maneuvers and end-of-session enactments could not dissolve responsibility. The zero-or-negative test converted the accumulation into a binary judgment that required no forecast of the overall deficit. The sequester supplied a sanction that operated without further legislation, confined the sanction to nonexempt direct spending, protected Social Security and Medicaid outright, capped the exposure of Medicare at 4 percent, and left administrative expenses exposed even where benefits were shielded. The firewall kept the two forms of congressional action in separate accounts, each policed on its own terms, and the emergency designation admitted a single deliberate exception conditioned on the agreement of Congress and the President. Preservation of a negotiated reduction, rather than compulsion toward an outcome that events could defeat, was the whole of the undertaking, measured session by session and entry by entry until the statutory procedures expired at the end of fiscal year 2002.

The Two Instruments Side by Side

The two-instrument table records the comparison in compact form, stating for the discretionary caps and for pay-as-you-go what each instrument constrains, what triggers enforcement, which accounts face reduction when enforcement fires, and what stands exempt from that reduction.

Dimension	Discretionary caps	Pay-as-you-go
What each instrument constrains	Annual appropriations, measured in budget authority and outlays	Enacted legislation affecting direct spending and revenues
What triggers enforcement	Enacted appropriations above an adjusted category or total limit	A net debit on the five year scorecard at the end of a session
Which accounts are cut when enforcement fires	Nonexempt discretionary accounts inside the breached category	Nonexempt direct spending programs
What is exempt	Accounts exempted under the deficit control act and designated emergency appropriations moved outside the limits by adjustment	Social Security and Medicaid, with Medicare reductions limited to 4 percent and designated emergency legislation kept off the scorecard

Credit Reform: Making the Cost of a Loan Visible

The Budget Enforcement Act of 1990 dealt mainly with limits on appropriations and with a pay-as-you-go rule for new mandatory spending and revenue legislation, but the same statute carried a third reform that changed the accounting for a different kind of federal activity. Title XIII, Subtitle B of the Omnibus Budget Reconciliation Act of 1990, in section 13201, enacted the Federal Credit Reform Act of 1990. The new act did not stand alone in the United States Code. It was placed as Title V of the Congressional Budget Act of 1974, alongside provisions governing budget resolutions, the Congressional Budget Office, and congressional budget procedure. Credit reform was a budget concept before it was a lending concept, concerned with how the budget recorded a promise stretching across decades.

Placement inside a budget enforcement statute followed from the defect the reform addressed. A grant appeared once, at its full amount, in the year Congress provided it. A loan appeared through yearly cash movements, large at the start and then gradually reversed by repayments. A guarantee appeared at almost nothing at the start and then, years afterwards, through scattered payments when borrowers defaulted. Because the discretionary limits created in 1990 rationed budgetary room year by year, any activity whose recorded cost could be pushed outside the measured year, or made to look smaller than the commitment behind it, held an advantage inside that system. Without it, the caps could have encouraged substitution of guarantees for grants and loans for appropriations, preserving the appearance of restraint while expanding federal obligations beyond the year of decision.

Under the cash method that preceded reform, a direct loan was scored by the money that left and entered the Treasury in each fiscal year. The full disbursement counted as an outlay in the year of the loan, and repayments of principal and interest counted as offsetting receipts in the years of collection. The first distortion followed directly. A sound loan expected to be recovered in full, with interest, looked in its first year as costly as an outright grant of the same amount, because the disbursement dominated the record and the recoveries lay in the future. Committees therefore compared a grant at its true lifetime cost against a loan at a figure overstated by roughly the amount expected to be repaid.

The treatment of a loan guarantee inverted the error. When the government guaranteed a loan made by a private lender, no federal cash moved at the moment of commitment in the ordinary case. The budget recorded nothing, or recorded only fees collected from borrowers or lenders, which could make a guarantee program appear to raise money rather than spend it. Federal cash moved only if the borrower defaulted, and defaults on a new body of guaranteed loans tended to arrive years after the guarantees were issued, when the loans had seasoned and economic conditions had turned. The second distortion ran in the opposite direction. A guarantee whose expected losses ran into hundreds of millions of dollars could enter the budget at zero cost in the year the exposure was created, and the eventual default claims fell into later budgets, detached from the volume decisions that had produced them.

Together the two distortions biased policy toward guarantees and away from direct loans, and toward credit in general and away from grants, for reasons unrelated to which instrument served a borrower best. A committee seeking to assist home buyers, farmers, or exporters within a tight annual allocation found the guarantee nearly free on paper and the direct loan nearly prohibitive, even when analysis of expected repayments and defaults might have shown the direct loan to be the cheaper instrument. The bias also hid subsidy. Every loan made at a rate below the market rate for that borrower, and every guarantee issued for a fee below the value of the risk assumed, transferred value from the government to the borrower. Cash accounting left that transfer invisible at the moment it occurred, although the support embedded in a credit decision varied with interest rates, maturities, default probabilities, and recovery prospects.

Why is the discount rate behind the subsidy cost disputed?

The subsidy cost is the estimated long-term cost to the government of a direct loan or a loan guarantee, calculated on a net present value basis at the time of commitment. The calculation excludes administrative costs and discounts expected cash flows using rates on comparable Treasury securities.

The definition in section 502 of the Federal Credit Reform Act turned the budget record from a diary of cash movements into an estimate of cost made at the moment of decision. For a direct loan, estimators projected the expected stream of cash connected with the loan, disbursements, repayments of principal, interest, fees, recoveries on defaulted balances, and proceeds from collateral, and discounted that stream back to the year of disbursement. For a guarantee, estimators projected the payments expected on default claims, net of fees and recoveries, and discounted that stream in the same way. The result was a single figure recorded up front, at commitment, representing what the commitment was expected to cost across its life. A loan expected to be repaid in full at a rate equal to the discount rate carried a subsidy cost near zero.

Discounting performed the central intellectual work of the reform, because money expected in a distant year does not carry the same value as money in hand. The act directed that expected cash flows be discounted using rates on Treasury securities of comparable maturity, tying the recorded cost of credit to the cost at which the government itself borrowed. The spread between the rate charged to the borrower and the rate paid by the Treasury, adjusted for expected defaults and fees, measured the drain or contribution the loan made to the public finances. Administrative costs stood outside the definition by statute. Salaries, servicing expenses, and program overhead were recorded separately on a cash basis, so the subsidy figure isolated the financing concession and the credit risk from the operating expense of delivery.

Reform applied from a stated starting point rather than retroactively. The act governed direct loans obligated and loan guarantees committed beginning in fiscal year 1992, while earlier loans and guarantees continued under liquidating accounts that recorded their cash flows under the older approach as those commitments wound down. From fiscal year 1992 forward, a new commitment stood in the budget at its estimated lifetime cost in the year the government made it, and later cash movements on that commitment no longer masqueraded as new spending or new savings attributable to later decisions.

Recording a cost up front changed little unless the cost had to be paid for under the budget rules, and the act supplied that consequence. For a discretionary credit program, the subsidy cost required an appropriation in advance, before an agency could obligate a direct loan or commit a guarantee carrying that cost. Congress had to provide budget authority sufficient to cover the estimated subsidy, together with authority for administrative expenses, and the loan volume an agency could support followed from the appropriation divided by the estimated subsidy rate. A program whose loans carried an average subsidy of ten cents per dollar lent could support ten dollars of volume for each dollar appropriated, while a riskier or more concessional program supported less volume on the same appropriation.

That requirement placed credit in direct competition with every other claim on discretionary resources. Because the Budget Enforcement Act capped total discretionary budget authority and outlays, subsidy dollars appropriated for a housing or export credit program occupied room under the caps that could otherwise have funded grants, salaries, or procurement. Loans and grants stood, for the first time under a general rule, on an equivalent budgetary basis at the moment of choice. A committee could prefer a loan to a grant for sound reasons, because a loan recovered part of its cost, reached borrowers a grant could not reach, or leveraged private capital, but the preference had to survive an honest comparison of lifetime costs charged against the same limited allocation. Mandatory credit programs occupied a different position, since their subsidy costs were funded through permanent authority rather than annual appropriations, but the measurement rule applied across the portfolio.

Each program operated through a pair of accounts with distinct budgetary characters. The program account was the budgetary account, where Congress appropriated the subsidy cost and administrative expenses and where the recorded cost entered the budget totals and the enforcement calculations. The financing account was nonbudgetary, deliberately excluded from the budget totals, and handled the cash mechanics of the loans. It received the subsidy payment from the program account, borrowed from the Treasury as needed, disbursed direct loans or paid default claims on guarantees, and collected repayments, interest, fees, and recoveries. Because the financing account stood outside the totals, the large gross flows of lending and repayment that had distorted the pre-reform record no longer moved the deficit figures. Only the subsidy moved them, and it moved them once, in the year of commitment.

No estimate made at commitment could survive contact with decades of economic change, borrower behavior, and movements in interest rates, and the act therefore required the estimate to be revisited. Agencies, working under methodology coordinated by the Office of Management and Budget and reviewed by the Congressional Budget Office, reestimated the subsidy cost of each cohort as experience accumulated. A cohort grouped the loans obligated or guarantees committed by a program in a single fiscal year, and each cohort carried its own record through its life. When a cohort performed better than projected, because defaults ran below expectations or recoveries ran above them, a downward reestimate returned value to the Treasury and reduced the recorded cost. When performance disappointed, an upward reestimate recorded additional cost, and the additional amounts were covered through permanent indefinite authority provided by the act, so a worsening cohort did not force an agency to halt new lending while seeking a supplemental appropriation for losses on old promises, with interest on the reestimates reflecting the time passed since the original estimate.

Congress stated the ambitions of the reform in section 501 of the act. The first purpose was to gauge the costs of federal credit programs with greater accuracy than cash accounting allowed, so that budget figures corresponded to the commitments described. The second purpose was to put credit programs on a budgetary footing equivalent to that of other federal spending, ending the privileged or penalized position credit had occupied by accident of accounting. The third purpose was to encourage the provision of benefits in the form best suited to the purpose at hand, so that the choice among a grant, a direct loan, and a guarantee would turn on which instrument delivered the intended assistance most appropriately rather than on which instrument the accounting flattered. The fourth purpose was to improve the allocation of resources, both among credit programs competing with each other and between credit programs and other spending competing for the same public funds.

The reach of the reformed accounting extended across the main fields of federal lending. In housing, the Federal Housing Administration insured mortgages for buyers whose down payments or credit histories placed them outside conventional market terms, the Department of Veterans Affairs guaranteed mortgages for veterans, and rural housing programs lent and guaranteed in communities private lenders served thinly. In agriculture, federal direct loans and guarantees supported farm ownership and operating expenses through seasons and commodity cycles that made ordinary commercial terms hard for many producers to obtain. Small business lending through the Small Business Administration relied heavily on guarantees of loans originated by private lenders, a structure that multiplied the importance of honest guarantee pricing, since federal exposure could grow to many times the annual appropriation a cash record would have suggested. Export finance through the Export-Import Bank combined direct lending, guarantees, and insurance in support of American sales abroad, in competition with the credit agencies of other exporting nations.

Student lending illustrated both the scale of federal credit and the sensitivity of its measurement. Loans to students and parents under the programs authorized by the Higher Education Act framework governing federal student lending combined features that made subsidy estimation unusually dependent on assumptions, including long repayment periods, repayment options tied to borrower income in the programs as they developed during the 1990s and 2000s, deferment and forbearance, and default patterns tied to the fortunes of borrowers early in working life. Before 2010 the federal government both made direct loans and guaranteed loans originated by private lenders under the bank-based program Congress ended in 2010, after which new federal student loans were made directly. A small change in an assumed default or discount rate, applied to annual volume measured in tens of billions of dollars, altered the estimated subsidy by sums large enough to matter in an appropriations bill, which made student lending a recurring subject in arguments over the statutory method.

That argument became the most substantial challenge to the settlement of 1990, and it concerned the discount rate at the center of the method. The statute directed discounting at Treasury rates, which reflected the borrowing cost of the government and excluded any charge for the risk that expected cash flows might not arrive. A private buyer pricing the same stream of expected student loan or mortgage payments would have paid less than the Treasury-rate present value, because private markets demanded compensation for bearing risk. Analysts at the Congressional Budget Office developed this objection into a sustained program of work during the 2010s, publishing fair-value estimates of major federal credit programs, including student loans and housing guarantees, alongside the official estimates produced under the statute. In those presentations the office reported that several programs recorded as saving money under the statutory method, most prominently student lending in a number of years, appeared instead to cost money once a market price for risk was included, and that housing guarantee programs carried larger costs on a fair-value basis than the budget recorded. Its analysts argued that a measure ignoring the cost of risk understated what taxpayers stood to lose in bad economic states, precisely when losses on many programs would arrive together.

Defenders of the statutory method answered from within the logic of a government budget. Officials at the Office of Management and Budget, and analysts who supported the Treasury-rate approach in the debates of the 2010s, argued that the federal government was not a private investor and did not need to earn a market return, that it borrowed at Treasury rates and could spread risk across generations of taxpayers, and that the purpose of the budget figure was to record the expected effect on federal borrowing and debt rather than the price a private firm would pay for the loans. On that view, adding a market risk premium recorded a cost the government would never pay in cash terms, overstated the burden on the Treasury, and could discourage programs whose expected cash cost was genuinely negative. The same defenders observed that fair-value estimates rested on choices about the size of the risk premium, drawn from markets for assets that resembled federal loans imperfectly, so the alternative introduced a new layer of estimation judgment into figures governing appropriations. The disagreement, as it stood in the published exchanges through 2019, concerned what a budget number is for: a record of expected consequences for the public debt, or a statement of economic value on the terms a market would set.

The debate left the structure of 1990 intact while altering the terms in which credit programs were discussed. No statute enacted before February 1, 2020 displaced Treasury-rate discounting as the official basis of the budget, and subsidy appropriations, program accounts, financing accounts, and cohort reestimates operated on the statutory method in the years through 2019. Credit reform had insisted that a promise be priced before it was made, and the argument of the 2010s was, at bottom, an argument over the right price. That the argument proceeded through published estimates of identifiable programs marked the distance traveled from the era when a guarantee entered the budget at zero and a loan entered at its face value. The reform of 1990 had not settled the value of federal credit for all time, but it had made that value a question the budget was obliged to ask, in numbers, at the moment of commitment, under caps that forced the answer to compete with every other use of public money.

Extensions, Surpluses, and the Expiry

The Budget Enforcement Act of 1990 was built to operate quietly. Earlier deficit law had promised a result and then demanded that Congress find the savings later, under threat of automatic cuts large enough to frighten both parties. The 1990 statute reversed that sequence. The savings came first, inside the Omnibus Budget Reconciliation Act of 1990 itself, and the enforcement machinery followed behind the savings as a guard. Discretionary caps limited the annual appropriations bills to negotiated totals. Pay-as-you-go placed enacted changes in direct spending and revenue law on a scorecard maintained by the Office of Management and Budget, so that a tax cut or a benefit expansion had to be paid for rather than merely announced. The design had a nameable logic, constrain the action, not the outcome, and the following twelve years tested that logic against recession and recovery, a stock market boom, and a political argument over money that the federal government had not expected to have.

In operation, the caps bit in the least dramatic and most consequential way available to a budget rule. Appropriations bills were written to the caps. The Congressional Budget Act allocations that divided a budget resolution total among committees, and the suballocations that divided the Appropriations Committee total among its subcommittees, had to be fitted inside statutory ceilings that carried a sequestration penalty for breach. A subcommittee chairman who wanted more for one program had to take the money from another program inside the same total, or find an adjustment the statute allowed, or persuade Congress and the President to change the law. Argument did not disappear from the appropriations process. Argument changed address. Disputes that might once have been fought over the size of the whole discretionary budget were fought instead over allocations within it, over the boundary between defense and domestic spending during the years when separate categories carried statutory walls, and over the scoring of particular items by the Congressional Budget Office and the Office of Management and Budget.

The emergency designation was the safety valve built into that arithmetic, and the record of the 1990s shows the valve opening wider as the decade advanced. Spending that Congress and the President together designated as an emergency requirement raised the caps by the amount involved and stayed off the pay-as-you-go scorecard. In the early years the designation served the purposes on the label: disaster relief, military operations, and other needs that arrived too quickly for the regular appropriations cycle. Congressional Budget Office histories of supplemental appropriations in the 1990s recorded emergency amounts through most of the decade as comparatively modest beside the totals appropriated under the caps. For several years the amounts passing through the valve did not change the character of a system in which the caps governed the regular bills, and the regular bills carried the bulk of discretionary spending.

The first extension arrived before any serious test of expiry could occur. The Omnibus Budget Reconciliation Act of 1993, Public Law 103-66, carried the discretionary caps and pay-as-you-go through fiscal year 1998. The extension did not travel alone. The 1993 act was a deficit reduction measure in its own right, combining revenue increases concentrated at higher incomes, a higher tax on motor fuels, restraint in Medicare and other spending, and discretionary limits that the extended caps then locked into the appropriations process for the middle of the decade. The pairing repeated the 1990 settlement in a new partisan setting. In 1990 a Republican President had signed a package negotiated with a Democratic Congress. In 1993 a Democratic President signed a package passed with Democratic votes alone. Enforcement was written to survive the argument over the package itself: once the 1993 law was enacted, the caps and the scorecard guarded its savings against erosion by later bills, whichever party held the gavels.

The architecture of the caps simplified as the first extension took hold. For fiscal years 1991 to 1993, three categories, defense, international, and domestic, had stood apart with statutory walls between them, so that an appropriator could not solve a domestic shortfall by reaching into the defense total. Congressional Research Service Report R41901, Statutory Budget Controls in Effect Between 1985 and 2002, recorded the consolidation that followed: from fiscal year 1994, the three categories gave way to a single discretionary total. The consolidation matched the strategic facts of the period. The Cold War division that had justified a protected defense category and a protected international category carried less force after the Soviet Union dissolved, and a single total gave appropriators flexibility to move money across functions while holding the aggregate to the negotiated path. Flexibility inside the total and rigidity at the total were the twin features of the design, and consolidation strengthened the first without surrendering the second.

One separate category interrupted that simplicity. The Violent Crime Control and Law Enforcement Act of 1994, Public Law 103-322, created a separate category for the Violent Crime Reduction Trust Fund, which operated from fiscal year 1995 to fiscal year 2000 alongside the single general total. Money deposited in the trust fund, drawn from projected savings in federal personnel, financed crime programs outside the competition inside the general cap, under a dedicated limit of its own. The category guaranteed that a signature program would not be crowded out by other domestic priorities, and in the other direction that crime spending would not silently expand the general total. Additional narrow categories for highway, mass transit, and conservation spending appeared toward the end of the decade, each carving a protected channel through the general discipline while accepting a limit on the channel itself.

The second extension repeated the pattern of the first and then widened the strain on it. The Balanced Budget Act of 1997, Public Law 105-33, extended the discretionary caps and pay-as-you-go through fiscal year 2002. Congressional Research Service accounts of the 1997 extension recorded that separate categories returned for fiscal years 1998 and 1999, divided among defense, nondefense, and crime reduction. For fiscal year 2000, two limits applied, one for crime reduction and one for all other discretionary spending. For fiscal years 2001 and 2002, a single overall limit applied again. The 1997 act was paired, as the earlier extensions had been paired, with substantive budget legislation: restraint in Medicare and Medicaid growth, welfare changes carried over from the legislation of 1996, and a companion tax act that cut capital gains rates and created new education and child credits. The political setting had also changed. A Democratic President negotiated with a Republican Congress, and both sides claimed the emerging improvement in the fiscal outlook as vindication. Enforcement survived the negotiation because enforcement served both claims. Neither side wanted the other to spend or give away the improvement first.

The strain arrived with success. Through the middle 1990s, the caps had restrained a government that wanted to spend more than a deficit politics allowed. By the end of the decade, actual surpluses changed the moral of every appropriations argument. A cap written in 1997 for a world of deficits governed fiscal years in which the Treasury collected more than it spent, and members of both parties assembled lists of uses for money that the caps treated as unavailable. Defense readiness, education, health research, farm support, and tax reduction all competed against a ceiling whose justification appeared to have been overtaken by events. The caps did not collapse. The caps held formally while the political system routed around them, through the emergency designation, through adjustments agreed after the fact, and through appropriations written at levels the original ceilings would not have permitted.

The emergency designations of the closing years illustrate the routing in its clearest form. The Congressional Budget Office history of supplemental appropriations in the 1990s, published in March 2001, recorded that emergency designations inside regular appropriations acts, rather than supplemental bills alone, rose to more than 21 billion dollars in the fiscal 1999 omnibus measure and to about 31 billion dollars in fiscal 2000. Among the items carried under the designation were funding for the fiscal 2000 census, a constitutionally required count whose date had been set a decade in advance, and repeated packages of farm assistance responding to low commodity prices and crop losses at the end of the 1990s. Neither item resembled a sudden disaster in the ordinary sense. Each had a constituency, each had a plausible claim on federal money during a surplus, and each could be financed without breaching the caps only by calling the spending an emergency. The designation required agreement between Congress and the President, so the practice was not evasion by one branch against the will of the other. The practice was a joint decision to keep the caps on the books and to set them aside for the spending both branches wanted.

Cap adjustments in the final cycles completed the picture. The ceilings for fiscal years 2001 and 2002, set in 1997 at levels intended to force discretionary spending down as a share of the economy, proved tighter than the government that Congress and two Presidents intended to fund. Rather than accept sequestration or compress appropriations to the statutory path, the political branches raised the effective levels through the appropriations process itself, adjusting the caps upward to accommodate enacted totals, with budget resolution assumptions, statutory adjustments, and advance appropriations that shifted budget authority between fiscal years. In the last two years of the regime, the caps described a negotiating position more than a binding destination, and appropriations landed materially higher than the 1997 statute had contemplated. A rule that can be amended by the same majority that it restrains always carries that weakness, and the record of 2001 and 2002 showed the weakness operating in the open, under surplus conditions that had removed the fear which had made the rule binding.

Did the caps cause the surpluses?

No. The caps and pay-as-you-go guarded deficit reduction after legislation had enacted it, and the improvement had several causes at once. Tax and spending packages in 1990 and 1993 did the direct reduction, a capital gains revenue surge lifted receipts beyond forecasts, and falling defense outlays freed resources throughout the decade.

The surplus record itself is not in dispute. The federal budget recorded a surplus of about 69 billion dollars in fiscal year 1998, the first surplus since 1969, after a generation in which annual deficits had become the ordinary condition of federal finance. The surplus grew through fiscal 1999, reached about 236 billion dollars in fiscal year 2000, and continued through fiscal year 2001, giving the Treasury four consecutive years in the black. Debt held by the public fell, and the central argument in Washington during the presidential campaign of 2000 concerned the division of projected surpluses among tax reduction, debt retirement, and the protection of Social Security and Medicare. That argument, unthinkable in the deficit politics of 1990, is the strongest surface evidence for crediting the enforcement regime with a transformation. Surface evidence is where the crediting should stop, because the composition of the improvement points elsewhere.

The direct deficit reduction was legislated, not enforced into existence. The reconciliation act of 1990 combined revenue increases with spending restraint and debt service savings that the Congressional Research Service projected at about 482 billion dollars over five years. The reconciliation act of 1993 added a second large package of tax increases and spending restraint. Those two statutes changed the law that determined how much the government collected and how much it committed. The caps and pay-as-you-go, by contrast, changed the cost of changing that law again. Enforcement preserved a trajectory. Preservation is a real contribution, and its value grows with every year in which a Congress might otherwise have spent the improvement, but a guard is not the author of the thing guarded. Any account that credits the caps with the surpluses has confused the lock on the door with the income of the household.

The revenue surge supplied the second cause, and the forecasters of the period were candid about its mystery. Through the late 1990s, federal receipts ran ahead of projections year after year, and Congressional Budget Office analyses of those years attributed a large share of the surprise to capital gains realizations during the equity market boom. Realized gains nearly doubled as a share of individual income tax receipts across the middle of the decade, as asset sales generated taxable income that standard forecasting relationships, built on historical patterns of gains relative to the size of the economy, had not predicted. The Congressional Budget Office treated part of the surge as temporary and then revised its treatment as the receipts kept arriving, a sequence that amounted to an admission that the models could describe the windfall better than they could explain it. Legislation had set the rates at which gains were taxed, and the 1997 act had lowered those rates even as collections rose. No provision of the Budget Enforcement Act taxed a single capital gain.

The defense drawdown supplied the third cause, measured in the largest single movement of federal spending during the decade. Defense outlays fell from roughly five percent of gross domestic product in 1990 to roughly three percent in 2000, as the force structure built for the Cold War contracted through base closures, procurement reductions, and personnel cuts under Presidents of both parties. A decline of about two percentage points of gross domestic product in one spending category exceeds the entire fiscal 1998 surplus measured against the size of the economy. The drawdown was a strategic decision following the collapse of the Soviet Union, taken through the defense bills of the period and ratified in a series of reviews and budgets. The caps accommodated the drawdown and, through their category structure in the early years and their consolidated total afterward, helped determine how much of the peace dividend was banked against the deficit and how much was redirected to domestic programs. Helping to allocate a dividend is different from creating one.

The strongest defensible claim for the regime is therefore a precise and narrower one. Between 1991 and the end of the decade, no major tax cut was enacted without offsets under pay-as-you-go discipline, discretionary appropriations followed the negotiated path through the middle 1990s, and proposals to spend the emerging surplus met a procedural obstacle that forced advocates to argue for changing the rules rather than ignoring them. The question of whether the reforms controlled spending in those years turns on that distinction between originating restraint and enforcing it. The 1990 and 1993 packages originated the restraint. The enforcement regime kept the packages from being quietly repealed by increments, held discretionary growth to the totals the negotiators had chosen, and converted deficit reduction from a single vote in a crisis year into a standing condition of legislating. Regimes that prevent backsliding rarely receive credit proportional to their work, because the backsliding that did not happen leaves no event to commemorate.

Expiry, when the end came, was a lapse rather than a repeal. The caps and statutory pay-as-you-go expired on September 30, 2002, the last day of fiscal year 2002, and Congress did not renew them. The pay-as-you-go scorecard balances for later years were set to zero by statute, closing the account without a sequester. President George W. Bush had proposed extending the enforcement procedures in the budget for fiscal year 2002, but the budget for fiscal year 2003 retreated to a general willingness to work with Congress on enforcement mechanisms, and no extension legislation gathered a coalition. The constituency that had sustained renewal in 1993 and 1997 had dissolved, and the dissolution had identifiable causes in the events of 2001 and 2002 rather than in any judgment that the machinery had malfunctioned.

The return of deficits removed the premise on which renewal politics rested. The tax act of 2001, the Economic Growth and Tax Relief Reconciliation Act, committed the projected surplus to rate reductions and rebates on a scale that assumed the surplus would persist. The recession that ran from March to November 2001 cut receipts and raised automatic spending at the same time. The attacks of September 11, 2001, brought military operations in Afghanistan, a new homeland security apparatus, and reconstruction assistance for New York, financed substantially through emergency supplemental appropriations of the kind that the caps had strained to contain even during surplus years. By the close of fiscal year 2002 the federal budget had returned to deficit, and the argument for renewing a discipline designed to protect a surplus had lost its subject. Members who had voted for enforcement as guardians of deficit reduction faced a different choice in 2002: accept caps that would bind wartime and recession spending, or let the statute lapse and legislate without it. The lapse required no vote against enforcement in the abstract, which suited a Congress that preferred the outcome to the declaration.

The procedural afterlife of the regime ran through the chambers rather than the statute book. In the Senate, a pay-as-you-go point of order survived the expiry of the statutory requirement, because the Senate had maintained a rule of that kind since 1993 alongside the statute. The surviving rule operated during consideration of legislation, exposing a deficit increasing bill to a point of order that took a three fifths vote to waive, but it lacked the sequestration machinery that had given the statutory version its automatic character. The Senate modified the rule in 2003 in a direction that weakened it materially, exempting direct spending and revenue changes that the adopted budget resolution had assumed. A rule that defers to the majority plan of the year restrains improvisation more than direction. In the House, no pay-as-you-go rule existed during the first years after expiry. The House adopted one on January 5, 2007, at the opening of the 110th Congress, requiring that legislation affecting direct spending or revenues avoid increasing the deficit across six year and eleven year measurement periods. The Senate strengthened its own rule during 2007 as well, removing the budget resolution exemption and applying tests across five year and ten year horizons. Both chambers had rebuilt a fence at the point of consideration. Neither chamber restored the scorecard, the sequester, or the statutory caps that had made the 1990 regime a system rather than a custom.

Measured across its twelve years, the regime did the work its design assigned and no other work. The caps governed appropriations while deficits frightened the political branches, bent through emergency designations when surpluses tempted them, and were adjusted upward when the governing majority preferred higher spending to the discipline of its own earlier statute. Pay-as-you-go deterred unpaid-for tax cuts and benefit expansions during the years when the scorecard carried a sequestration threat. The surpluses of fiscal years 1998 to 2001 belonged to the tax and spending packages of 1990 and 1993, to an equity boom that filled the Treasury beyond every forecast, and to a defense establishment shrinking after a won Cold War. The enforcement regime belonged in the same history at a different rank: the mechanism that kept those gains from being given back in the ordinary traffic of legislating, for as long as the political branches wanted them kept.

The Restoration That Arrived Without the Caps

The Statutory Pay-As-You-Go Act of 2010 entered law through a vehicle built for a different emergency. President Barack Obama signed House Joint Resolution 45 on February 12, 2010, and Title I of that joint resolution carried the restored pay-as-you-go statute into the United States Code at sections 931 to 939 of Title 2. The measure received the designation Public Law 111-139, and the same enactment raised the statutory debt limit to about $14.3 trillion. The pairing mattered for interpretation from the first day. A Congress willing to authorize a large increase in federal borrowing also wrote a rule against adding to the deficit through new revenue and direct spending legislation, and the two actions sat side by side in a single signed text. The Congressional Research Service, in the report titled The Statutory Pay-As-You-Go Act of 2010: Summary and Legislative History, treated the statute as a revival of the enforcement idea that had operated from 1990 to 2002 rather than as a fresh invention. The lineage was deliberate. Sponsors in 2009 and 2010 described the bill as a return to the discipline associated with the surplus years at the close of the twentieth century, and the statutory language borrowed the central mechanism of the earlier regime, a scorecard policed by the threat of automatic cuts.

The resemblance to the Budget Enforcement Act of 1990 ended at that central mechanism, and the differences defined the restored rule. The 1990 act had deployed two instruments together. Statutory limits on discretionary spending constrained the annual appropriations process category by category and then in consolidated form, while pay-as-you-go neutrality constrained legislation affecting direct spending and revenues. The 2010 act brought back the second instrument alone. No discretionary caps accompanied the restoration, so appropriations for defense, education, infrastructure, and the rest of the annually funded government faced no statutory ceiling from this statute, however large those bills grew. A later Congress addressed that half of the enforcement problem through separate legislation, and the Budget Control Act of 2011 guide traces the statute enacted on August 2, 2011, that revived discretionary limits alongside a different sequestration design. The 2010 act also set no expiration date, a sharp departure from the practice of 1990, 1993, and 1997, when each extension carried a defined end and forced a fresh legislative decision. Absent a sunset, the restored rule continued by inertia, and repeal or amendment required affirmative action rather than passive lapse. Permanence sounded stronger than a fixed term. In practice, permanence also removed the scheduled moments at which Congress had once been obliged to debate whether the rule deserved renewal on its record.

The scorekeeping machinery differed from the 1990 design in a second respect. The earlier regime had relied on a single rolling scorecard covering five years. The 2010 act directed the Office of Management and Budget to keep two scorecards, one covering five years and one covering ten years, and to record on each the average annual budgetary effect of enacted revenue and direct spending legislation. Averaging spread the measured effect of a law across the years of the relevant window, so a statute with costs concentrated early and savings promised later entered the record as a smoothed annual figure rather than as a year by year profile. The budgetary effects themselves came, in the first instance, from statements submitted for printing in the Congressional Record by the chairs of the House and Senate Budget Committees, and legislation enacted after 2010 routinely carried language directing scorekeepers to those statements. Where no such statement existed for a measure, the Office of Management and Budget determined the effects on its own authority. A debit on either scorecard for the budget year, measured at the end of a session of Congress, obliged the President to issue a sequestration order cutting nonexempt direct spending by an amount sufficient to eliminate the debit.

Has the restored PAYGO rule ever forced a spending cut?

No. Through the end of the session that closed in December 2019, the Office of Management and Budget reported no debit requiring a sequestration order under the 2010 act, and the Congressional Budget Office recorded no instance in which the backstop cut any program, including Medicare, during the first decade after enactment.

That unbroken record invites two readings, and the honest assessment in this profile holds both open. Deterrence offers one reading. A backstop that never fires may have done its work in committee rooms and conference negotiations, where drafters shaped bills against the knowledge that an unoffset debit carried a formal consequence, and the absence of a sequestration order would then measure success rather than failure. Evasion offers the second reading, and the legislative record through 2019 supplied substantial material for it, because Congress repeatedly reached the end of a session with debits removed, excluded, or moved rather than offset. The design of the backstop itself limited the damage a firing would have inflicted, and that limitation weakened deterrence in a measurable way. Social Security stood wholly exempt from a pay-as-you-go sequester, and Medicaid stood exempt as well, so the two largest direct spending commitments for retirees and for low income health coverage sat outside the penalty. Medicare faced exposure, but a statutory ceiling protected the program, and verification against the Congressional Research Service report titled Medicare and Budget Sequestration and against Congressional Budget Office explanations of the 2010 act confirms the figure stated in this profile: a sequestration order under the 2010 act could reduce Medicare payments by no more than 4 percent. A full account of enforcement also requires a comparison of sequestration and pay-as-you-go as distinct devices, because the 2010 act fused a rule about legislative behavior to a penalty aimed at programs that had caused no offense. Pay-as-you-go governed the drafting decision, requiring offsets at the moment of enactment. Sequestration governed the aftermath, canceling budgetary resources across nonexempt accounts when the drafting decision produced a debit that survived to the end of a session. The mismatch between the actor and the target ran through every debate about the restored statute. Defenders of the design answered that indiscriminate pain was the point, since a penalty that spared favored programs would invite drafters to load costs onto exempt accounts, and the 1990 experience had shown that a broadly feared sequester concentrated legislative minds during drafting. The record through 2019 never tested that answer, because no order issued, and an untested threat derives force only from the credibility of the institution that issued it.

Congress set the restored rule aside through channels the statute itself supplied, and the first channel opened on the day of enactment. Section 4 of the 2010 act excluded specified policies from scorecard treatment, so that extensions of designated elements of the tax and health payment landscape could proceed without offsets and without creating a debit. The Committee for a Responsible Federal Budget, in an analysis dated July 22, 2009, listed the exempted territory in the proposal that became the statute: renewal of the income tax reductions enacted in 2001 and 2003, extension of estate tax rules then in force, continued patches against the alternative minimum tax, and continued updates to Medicare physician payments. The same analysis cited a Congressional Budget Office estimate that the exceptions in an earlier version of the legislation would allow enactment of policies increasing deficits by an amount in the vicinity of $3 trillion over the period from 2010 to 2019 without triggering a sequestration. The enacted statute moderated some of that territory without eliminating the principle, and the principle deserves plain statement: the restoration Congress wrote its largest anticipated exceptions into the rule at birth, deciding in advance that certain extensions counted as continuation rather than as new action, whatever their measured cost. That decision embodied the founding maxim of the 1990 regime in inverted form. The earlier act had promised to constrain the action, not the outcome. The 2010 exclusions constrained only the actions Congress had not placed beyond reach in advance.

The second channel ran through emergency designations under section 4(g) of the act. A provision designated as an emergency requirement in pay-as-you-go legislation kept its budgetary effects off both scorecards entirely, by instruction to the Congressional Budget Office and the Office of Management and Budget alike. The designation carried procedural friction, including a separate vote on consideration in the House and exposure to a point of order in the Senate that required sixty votes to waive, yet the friction was procedural rather than substantive, since the statute imposed no test of suddenness, unpredictability, or temporary need comparable in rigor to a definition. A majority determined to label a provision an emergency, with the required Senate supermajority for a waiver where a point of order was raised, could remove the provision from enforcement by declaration. Genuine emergencies supplied the justification for the channel, and the financial crisis conditions surrounding the 2010 enactment made the case vividly, because a rule that forced offsets during a collapse in private demand could deepen a contraction. The same channel, however, asked no questions about duration or scale once the label attached, and each use through the 2010s normalized the next. Emergency treatment and specified policy exclusions together meant that the scorecards never recorded the full deficit effect of legislation enacted under the restored regime, and any assessment that treats the absence of sequestration as proof of neutrality mistakes a filtered record for a complete one.

The third channel appeared when a debit reached the scorecard anyway, and the tax legislation of 2017 supplied the defining case before the February 2020 boundary of this profile. The Tax Cuts and Jobs Act, Public Law 115-97, signed on December 22, 2017, reduced federal revenues on a scale that the Congressional Research Service later summarized as a projected deficit increase of about $1.46 trillion over the period from 2018 to 2027. The act carried no internal provision erasing that effect from the pay-as-you-go scorecards. Instead, the remedy arrived in separate legislation, as the design of the reconciliation process effectively required, and the Bipartisan Budget Act of 2018, Public Law 115-123, signed on February 9, 2018, shifted the scorecard debits attributable to the tax act out of the budget year rather than requiring offsets or permitting a sequestration order. The Office of Management and Budget scorecard reports that followed recorded balance shifts affecting budget years 2019 and 2020, a pattern in which a debit survived on paper while its enforcement moment receded into a later year. Deferral differed from forgiveness in form and converged with forgiveness in effect across repeated use, since a balance moved once could be moved again by the same legislative technique. The episode demonstrated, in a single sequence dated across three months, the structural truth of statutory enforcement in a system of legislative supremacy: a statute binds a future Congress only until that Congress passes another statute, and the majority that enacted the tax reduction commanded the votes to neutralize the consequence in the next vehicle that moved.

The enforcement argument for the restored rule, stated at full strength, began from that structural truth and answered it with political rather than legal force. The Committee for a Responsible Federal Budget declared in July 2009 that the organization strongly supported statutory pay-as-you-go, together with tough but realistic discretionary spending caps, as a means of keeping the budget situation from deteriorating further. The logic assigned the statute a gatekeeping function. Every unoffset proposal had to survive an additional argument, in public, about why an exception was warranted, and the scorecard converted quiet drift into a recorded number attached to named legislation. Alice Rivlin of the Brookings Institution, testifying before the House Budget Committee in June 2009, called statutory pay-as-you-go a small but important step toward restoring fiscal discipline and argued that writing the pay-as-you-go rules adopted in 2007 into law would highlight their importance and make the rules easier to enforce. Deficit hawks in both parties added a comparative claim across the decades covered by this series: the years from 1990 to 2002, when statutory enforcement operated, coincided with the last sustained movement from deficit to surplus in federal finances, while the years after the 2002 expiration coincided with large tax reductions and new mandatory commitments enacted without offsets. Correlation carried the usual hazards, and the surpluses of 1998 to 2001 rested on revenue growth, defense restraint, and economic expansion far beyond any rule, yet the enforcement case insisted that a rule preventing new legislation from giving improvement back had done real work inside that larger story.

The flexibility argument, stated at equal strength, denied that deficit neutrality deserved the status of a universal test for legislation. The Center on Budget and Policy Priorities, in testimony presented during the 2009 consideration of the statute, argued against pretending that designated extensions of expiring policy would either lapse or be paid for, warning that a rule built on a promise nobody intended to keep would produce an inevitable series of waivers, and that those waivers would undermine support for the rule itself and open the door to waivers for other costly policies. The point generalized beyond the specified exclusions. A tax reduction, a health coverage expansion, or a countercyclical measure in a recession fails the neutrality test on enactment even when long run returns exceed the recorded debit, because the scorecard counts legislated effects inside a fixed window and assigns no value to benefits outside the federal ledger. Lawmakers who pressed versions of this case during the debates of 2009 and 2010 accepted deficits as a legitimate instrument of policy in defined circumstances, and the emergency designation in section 4(g) amounted to a statutory concession that the flexibility case was partly right. A rule with no escape valve would have commanded less support at enactment and might have broken, rather than bent, under the first genuine crisis.

Measured against the ambitions announced in February 2010, the restored statute therefore occupies a narrow middle ground in the record closed at the start of 2020. The statute raised the cost of unoffset legislation at the margin, forced repeated public votes and recorded findings whenever Congress chose an exception, and preserved a vocabulary of neutrality that budget committees invoked in ordinary drafting throughout the decade. The statute also accommodated, through its own text and through later amendments to its scorecards, the largest revenue measure of the decade without offsets, sequestration, or repayment. Both halves belong in the assessment, and either half alone flatters a prior position. The deeper judgment concerns the 1990 act itself, whose profile this section closes, and candor requires listing what the earlier statute plainly failed to accomplish even during its most effective years. The Budget Enforcement Act of 1990 did not control the growth of mandatory programs already in law, since pay-as-you-go reached new legislation and left the benefit formulas, demographic pressures, and health cost growth inside Social Security, Medicare, and Medicaid to compound untouched outside the scorecard. A generation of beneficiaries entered those programs between 1990 and 2002 under rules the enforcement regime never reviewed, and the fiscal weight of that growth dwarfed the debits and credits that the scorecard policed at the margin of new enactments.

The 1990 act also failed to bind a future Congress determined to legislate around it, a limitation demonstrated inside its own lifespan and repeated, in the scorecard shifts of 2018 and 2019, under its successor. Emergency designations, directed scorekeeping, and end of session adjustments had all appeared before the 2002 expiration, and the techniques available in 2018 were refinements of methods visible in the 1990s rather than inventions of a less disciplined age. Finally, the 1990 act did not prevent the return of deficits once its instruments lapsed on September 30, 2002. The unified surpluses of 1998 to 2001 gave way to deficits from fiscal year 2002 onward, driven by tax legislation enacted in 2001 and 2003, by war spending after September 11, 2001, by a new Medicare prescription drug benefit enacted in 2003, and by economic downturns, in combinations that no surviving enforcement statute was positioned to block because none survived. The sequence supplies the closing verdict on the whole enforcement era that opened at Andrews Air Force Base in 1990. Instruments that constrain the action, not the outcome, can protect a settlement already negotiated and can slow the erosion of a fiscal position already achieved. Instruments of that kind cannot manufacture the settlement, compel the underlying choices about taxation and entitlement growth, or survive the determined opposition of the institution that created them. The record after 2010, taken up in the companion profiles in this series, presented the same dilemma in harder form, through a renewed experiment in discretionary caps, sequestration machinery that actually fired, and repeated budget acts through 2019 that renegotiated its limits.

A reader who wants to organize the citations, votes, dates, and accounting terms traced in this profile into a working research record can keep your statute notes, citations, and case chronologies together free on VaultBook as a companion to the statutory text and to the Congressional Research Service and Congressional Budget Office reports named in the sections above.

Frequently Asked Questions

Q: What did the Budget Enforcement Act change?

The act replaced fixed deficit targets with controls on legislative action. Title XIII of the Omnibus Budget Reconciliation Act of 1990, Public Law 101-508, placed statutory limits on discretionary appropriations and created a pay-as-you-go rule for enacted legislation affecting direct spending and revenues. The design preserved the deficit reduction in the reconciliation act, which the Congressional Research Service projected at about 482 billion dollars over five years, rather than promising a deficit figure for a future year. Automatic sequestration backed both instruments when appropriations breached a limit or the scorecard recorded a debit.

Q: What are discretionary caps in the Budget Enforcement Act?

Discretionary caps were statutory ceilings on the budget authority and outlays provided through annual appropriations acts. For fiscal years 1991 to 1993, separate caps governed defense, international, and domestic spending, and statutory walls prevented savings in one category from financing spending in another. From fiscal year 1994, the categories were consolidated into a single discretionary total. The Office of Management and Budget reported the limits in preview, update, and final reports. Enacted appropriations above an adjusted cap triggered sequestration confined to the breached category, applied as a uniform percentage reduction across nonexempt accounts.

Q: What is PAYGO in the Budget Enforcement Act?

Pay-as-you-go was a deficit neutrality rule for enacted legislation affecting direct spending and revenues. The Office of Management and Budget recorded the estimated effects of each enacted law on a rolling five year scorecard. Costs from benefit expansions or revenue reductions had to be matched by savings from direct spending reductions or revenue increases inside the recorded period. A cumulative balance of zero or less satisfied the rule. A net debit at the end of a session triggered sequestration of nonexempt direct spending programs, calculated to eliminate the debit, with Social Security and Medicaid exempt and Medicare reductions limited by statute.

Q: Why did the Budget Enforcement Act stop using deficit targets?

Fixed deficit targets punished Congress for results that legislation alone did not control. The deficit depended on economic growth, unemployment, interest rates, inflation, asset prices, and forecasting error, alongside tax and spending law. A recession could widen the deficit and trigger cuts after Congress had enacted a restrained budget, while a revenue surge could conceal costly legislation. Targets also invited optimistic assumptions, timing shifts, and revisions when compliance became painful, as the 1987 amendments demonstrated by moving the balance date outward. The 1990 act therefore measured appropriations and enacted legislation, actions traceable to recorded votes.

Q: Did a summit produce the Budget Enforcement Act?

A budget summit produced the settlement that the act enforced. President George H. W. Bush and congressional leaders convened negotiations in May 1990, and the talks moved to Andrews Air Force Base in Maryland during September 1990. On June 26, 1990, Bush issued a written statement accepting that deficit reduction would require tax revenue increases. Negotiators announced agreement on September 30, 1990, promising about 500 billion dollars of deficit reduction over five years, including about 134 billion dollars in taxes. Renegotiation after the House defeat reshaped the package before final passage and signature on November 5, 1990.

Q: Did the House defeat the first Budget Enforcement Act package?

The House defeated the first summit package in the early morning of October 5, 1990, by 179 to 254. Conservative Republicans led by Minority Whip Newt Gingrich opposed the tax increases and the abandonment of the no new taxes pledge, while liberal Democrats opposed the Medicare savings and the distribution of the burdens. President Bush then vetoed a further continuing resolution, and a funding lapse ran from Saturday, October 6 through Monday, October 8, 1990, closing national parks and Smithsonian museums. Leaders renegotiated the tax and Medicare provisions, and the revised reconciliation act passed later in October.

Q: When did the Budget Enforcement Act caps expire?

The discretionary caps and statutory pay-as-you-go expired on September 30, 2002, the final day of fiscal year 2002. The Omnibus Budget Reconciliation Act of 1993 had extended the regime through fiscal year 1998, and the Balanced Budget Act of 1997 had extended the regime through fiscal year 2002. Congress did not renew the controls at the 2002 deadline. The pay-as-you-go scorecard balances for later years were set to zero by statute, closing the account without a sequester. After expiry, each chamber relied on internal pay-as-you-go points of order rather than statutory enforcement.

Q: What is credit reform in the Budget Enforcement Act?

Credit reform was the Federal Credit Reform Act of 1990, enacted by section 13201 of the reconciliation act and placed as Title V of the Congressional Budget Act of 1974. Effective beginning in fiscal year 1992, the reform replaced annual cash flow scoring for federal direct loans and loan guarantees with an estimate of subsidy cost recorded at commitment. Subsidy costs of discretionary credit programs required appropriations in advance and competed inside the discretionary caps. Program accounts recorded the budgetary cost, financing accounts handled loan cash flows outside the budget totals, and agencies reestimated each cohort over its life.

Q: What is sequestration under the Budget Enforcement Act?

Sequestration was the automatic cancellation of budgetary resources that enforced a breach under the deficit control act. The Office of Management and Budget calculated the excess in a final report, and the President issued an order applying a uniform percentage reduction across nonexempt accounts sufficient to eliminate the excess. A cap breach produced a sequester confined to the breached discretionary category during the years of separate categories. A pay-as-you-go debit produced a sequester of nonexempt direct spending at the end of a session. The cancellation was permanent, and neither a committee vote nor administrative discretion could waive a reported breach.

Q: How did the Budget Enforcement Act treat emergency spending?

Spending designated as an emergency requirement received treatment outside both enforcement instruments. Appropriations with an emergency designation, made by Congress in law with the concurrence of the President, raised the discretionary caps by the designated amount. Legislation designated as an emergency by both branches stayed off the pay-as-you-go scorecard, so the estimated effects entered neither the cost column nor the savings column. Disaster relief and military operations used the channel during the 1990s. Use widened late in the decade, when the fiscal 2000 census and farm assistance packages carried emergency designations inside regular appropriations measures.

Q: What is the difference between discretionary caps and PAYGO?

The caps governed annual appropriations, while pay-as-you-go governed enacted legislation affecting direct spending and revenues. Caps measured budget authority and outlays against statutory limits, sorted first into defense, international, and domestic categories and later into a consolidated total. Pay-as-you-go measured the cumulative estimated deficit effect of new mandatory spending and tax law on a five year scorecard. The statute kept a firewall between the two ledgers. Savings under the caps created no scorecard credit, and scorecard savings raised no discretionary ceiling. Each instrument carried a separate sequester confined to the accounts within its own field.

Q: Did the Budget Enforcement Act cause the budget surpluses?

The enforcement regime contributed to the surpluses of fiscal years 1998 to 2001 without causing the surpluses alone. The reconciliation packages of 1990 and 1993 supplied direct tax increases and spending restraint. A revenue surge driven heavily by capital gains realizations during the equity boom carried receipts beyond forecasts through the late 1990s. Defense outlays fell from roughly five percent of gross domestic product in 1990 to roughly three percent in 2000 after the Cold War. The caps and pay-as-you-go guarded those gains by raising the cost of legislation that would have surrendered revenue or added unfinanced spending.

Q: When were the Budget Enforcement Act rules extended?

Congress extended the caps and pay-as-you-go twice before the final expiry. The Omnibus Budget Reconciliation Act of 1993, Public Law 103-66, extended both instruments through fiscal year 1998 and paired the extension with a second large package of revenue increases and spending restraint. The Balanced Budget Act of 1997, Public Law 105-33, extended both instruments through fiscal year 2002 and paired the extension with Medicare and Medicaid restraint and companion tax legislation. Each extension forced a fresh legislative decision because each carried a defined end date, and no further extension followed the 1997 act.

Q: How does the statutory PAYGO law differ from the Budget Enforcement Act?

The Statutory Pay-As-You-Go Act of 2010 restored the scorecard without restoring the discretionary caps. President Barack Obama signed the act on February 12, 2010, as Title I of Public Law 111-139, codified at sections 931 to 939 of Title 2. The Office of Management and Budget kept two scorecards, covering five years and ten years, recording average annual effects of enacted revenue and direct spending legislation. The act carried no expiration date. Social Security and Medicaid remained exempt, Medicare reductions remained capped at 4 percent, and specified policy exclusions and emergency designations kept designated effects off the scorecards.

Q: What happened to the deficit targets of Gramm-Rudman-Hollings?

The deficit targets of the Balanced Budget and Emergency Deficit Control Act of 1985 were missed, revised, and then abandoned as an enforcement design. The original schedule promised balance by fiscal year 1991, and the 1987 revision pushed balance to fiscal year 1993 while resetting intermediate targets. Projected sequestration for fiscal year 1991 reached a scale that threatened deep defense and domestic reductions during the Persian Gulf deployment. The Budget Enforcement Act of 1990 replaced the target schedule with discretionary caps and pay-as-you-go, shifting enforcement from a forecast deficit outcome to recorded congressional action on legislation.

Q: Which programs were exempt from a PAYGO sequester?

Section 255 of the deficit control act exempted major programs from a pay-as-you-go sequester. Social Security benefits stood outside sequestration apart from administrative expenses, and Medicaid stood exempt as well. Railroad retirement benefits, veterans compensation, and veterans pensions were exempt, as were payments of net interest and payments under the earned income tax credit. Federal retirement benefits and a range of programs directed at people with low incomes received exemption or special rule protection. Medicare occupied an intermediate position, exposed to reduction but protected by a statutory limit of 4 percent on the percentage reduction applied.

Q: How did credit reform change the cost of student loans?

Credit reform changed student lending by recording an estimated lifetime subsidy cost at commitment rather than a sequence of annual cash flows. Before fiscal year 1992, a direct student loan looked expensive in the disbursement year even when repayments were expected, and a guaranteed student loan looked free until defaults arrived years later. Under reform, estimators projected repayments, interest, fees, deferment, and defaults, discounted the stream at Treasury rates, and recorded the result once. Small changes in assumed default or discount rates moved estimated costs for annual lending volume measured in tens of billions of dollars.

Q: What is a subsidy cost in a federal credit program?

A subsidy cost is the estimated long term cost to the government of a direct loan or a loan guarantee, calculated on a net present value basis and recorded at the time of commitment. Estimators project disbursements, repayments of principal, interest, fees, recoveries, and default claim payments, then discount expected cash flows using rates on Treasury securities of comparable maturity. The definition in section 502 of the Federal Credit Reform Act excludes administrative costs, which agencies record separately on a cash basis. Agencies reestimate the subsidy for each annual cohort as defaults, recoveries, and interest conditions change.

Q: Why did Congress let the budget rules expire without renewal?

The political conditions that had sustained renewal in 1993 and 1997 had dissolved by 2002. The tax act of 2001 committed projected surpluses to rate reductions and rebates, the recession from March to November 2001 reduced receipts and raised automatic spending, and the attacks of September 11, 2001 brought military operations in Afghanistan and a new homeland security apparatus financed largely through emergency supplementals. President George W. Bush had proposed extending enforcement in the budget for fiscal year 2002, but no extension legislation gathered a coalition before the September 30, 2002 expiry.

Q: What is the public law number of the Budget Enforcement Act?

The Budget Enforcement Act of 1990 carried no separate public law number, because the act formed Title XIII of the Omnibus Budget Reconciliation Act of 1990, Public Law 101-508, printed at 104 Stat. 1388. President George H. W. Bush signed the reconciliation act on November 5, 1990. The House agreed to the conference report on October 27, 1990, by 228 to 200, and the Senate agreed to the conference report on the same day by 54 to 45. The enforcement title amended the deficit control act of 1985 and the Congressional Budget Act of 1974.