The Test This Passage Poses

A reader who finishes this article should be able to do three things. First, reconstruct how a bill pronounced dead at least three times became law by lopsided margins in both chambers. Second, name the procedural defeat that nearly ended it in the House of Representatives. Third, explain the single strategic decision in a Senate committee that converted a preference-laden bill into the most radical base-broadening measure in modern history. The telling must be procedural rather than anecdotal, because the argument of this article is that procedure determined the text, and that in this rare case the procedural crisis improved the statute it nearly killed.

/assets/images/blog/blog-01.webp

The vehicle was H.R. 3838 of the 99th Congress, introduced on December 3, 1985, by Representative Dan Rostenkowski, Democrat of Illinois, the chairman of the House Ways and Means Committee. Less than a year later, on October 22, 1986, President Ronald Reagan signed it as Public Law 99-514, recorded at 100 Stat. 2085. Between those two dates the measure was declared finished by commentators so often that Reagan himself told the Tax Reform Coalition it had been declared dead so many times by the pundits. Three of those deaths were real procedural events with dates, vote tallies, and mechanisms. The first was the rejection of the rule for floor debate in the House on December 11, 1985, by 223 to 202, with only fourteen Republicans voting in favor. The second was the suspension of the Senate Finance Committee markup on April 18, 1986, when Chairman Bob Packwood, Republican of Oregon, adjourned a committee that stood twenty-nine billion dollars off its mark and announced there would be no more sessions. The third was a June 1986 floor fight over amendments affecting individual retirement accounts, which Packwood survived 51 to 48 and afterward described in stark terms, testifying that had he lost a couple of votes he would have lost on the IRAs and lost the bill. Each crisis had a distinct procedural character, and each was resolved by a distinct procedural move. The substance of the statute is treated in the companion complete guide to the act, while this article is concerned entirely with the machinery that carried H.R. 3838 from introduction to enrollment.

What makes the passage worth studying as procedure is the shape of the final votes. The Senate passed the bill on June 24, 1986, by 97 to 3, with Republicans voting 53 to 0 and Democrats 44 to 3. The conference report was adopted in the House on September 25, 1986, by 292 to 136, with Republicans 116 to 62 and Democrats 176 to 74, and in the Senate on September 27, 1986, by 74 to 23, with Republicans 41 to 11 and Democrats 33 to 12. Those are not the margins of a measure dragged across the finish line. They are the margins of a measure that had become, by the end, the only available answer to a question both parties had agreed to ask. The puzzle is how a bill that could not get a rule out of the House in December became, by the following September, a bill that majorities of both parties in both chambers were willing to own. The answer lies in the constraint under which the bill was written, the crises that the constraint produced, and the escalation that the crises forced.

The margins themselves are the first piece of procedural evidence. A 97 to 3 Senate vote does not happen for a controversial revenue bill by accident, and it does not happen because ninety-seven senators independently reached the same policy conclusion. It happens because the legislative process has been arranged so that voting no is costlier than voting yes. The task of this article is to show how that arrangement was built, crisis by crisis, and why each crisis made the arrangement stronger rather than weaker. The conventional telling treats the near-deaths as obstacles the bill overcame on its way to passage. The procedural telling treats them as the passage itself, because each death eliminated a weaker version of the bill and each revival installed a stronger one.

Consider the sequence as a procedure rather than a story. The House produced a bill under the constraint but with its coalition intact only on paper. The rule defeat killed the assumption that the House bill could pass as written. The Senate committee then tried to repair the bill by ordinary bargaining and killed the assumption that ordinary bargaining could work. The substitute that replaced it was not a compromise between the House bill and its critics. It was a different kind of bill, one whose logic made the old objections irrelevant. The floor fight tested whether that new logic could survive contact with the amendment process, and the conference tested whether two chambers could agree on the final price. At no point did anyone persuade anyone of anything in the deliberative sense. At every point, someone changed the procedural conditions under which the next vote would be taken. That is what procedure determining text looks like from the inside.

The lopsided final votes are therefore not the puzzle’s solution but its confirmation. By September 1986 the bill had been through three deaths and three revivals, and each revival had broadened the coalition by narrowing the alternatives. Voting against the conference report meant voting against rate cuts that both parties had spent two years promising, in favor of a status quo that no organized constituency would defend in public. The 292 to 136 House vote and the 74 to 23 Senate vote recorded the absence of an alternative, not the presence of enthusiasm. Procedure had done its work so thoroughly that the final tallies look inevitable in retrospect, which is exactly why the near-deaths must be recovered in detail. Nothing about this passage was inevitable while it was happening.

A note on method is required, because this article promises procedure rather than anecdote and must honor the distinction. Anecdote explains events by the vividness of their participants. Procedure explains events by the structure of the choices the participants faced. The beer, the corridor, the rare presidential trip to the Capitol, and the twenty-nine-billion-dollar shortfall are all vivid, and a lesser account would organize itself around them. This account uses them as evidence for structural claims. The beer is evidence for the substitute. The corridor is evidence for the constraint’s neutralization of lobbying. The presidential trip is evidence for the rule vote’s cost structure. The shortfall is evidence for the incremental approach’s arithmetic failure. Wherever this article tells a story, the story is doing procedural work, and wherever it makes a procedural claim, the claim is anchored in a dated event, a numbered report, or a recorded tally. That discipline is what separates a procedural telling from an anecdotal one, and it is the standard to which every section that follows is held.

The organization of the article follows the bill’s procedural sequence. It begins with the design constraint, because the constraint preceded the legislation and governed it throughout. It then traces the three blueprints that taught Congress the constraint’s logic, the House phase that produced the first death, the Senate committee phase that produced the radicalism escape, the Senate floor phase that tested the substitute’s indivisibility, and the conference that settled the final rates. It then tabulates the four near-deaths, develops the complication that the constraint rather than the personalities did the work, records the bipartisan authorship, describes the lobbying environment, and closes with presentment and signature. Each section is procedural in the sense defined above. Each explains what the participants could choose by explaining what the procedure permitted.

The Constraint That Wrote the Bill

Every number in this story follows from a design constraint that was stated before a single line of legislative text existed. The reform had to satisfy three demands at once. It had to raise the same revenue as existing law, so that the Treasury would lose nothing in the exchange. It had to preserve the existing distribution of the burden across income groups, so that no class of earners could claim the revision shifted the load onto its shoulders. And it had to cut marginal rates, sharply, which was the political purpose of the entire exercise. Those three demands together created a closed system. Because the measure could not lose revenue and could not shift the burden between groups, the only remaining source of funds for lower rates was the base itself. Every percentage point of rate reduction had to be purchased with an eliminated preference. This rate-for-preferences bargain was not a slogan. It was an arithmetic identity, and it governed everything that followed.

The constraint explains why ordinary bargaining kept failing. In a normal markup, members trade: a rate cut here for a new preference there, a concession to one industry offset by a concession to another. Under the three-part constraint, that style of bargaining was mathematically self-defeating. Each preference protected or restored narrowed the base, which meant the rates had to rise to keep revenue constant, which destroyed the political value of the exercise. A committee working incrementally therefore produced, with each amendment, a bill that was slightly less worth passing. The Finance Committee learned this the hard way in the spring of 1986, when weeks of conventional markup left it twenty-nine billion dollars off its mark with a bill nobody could defend. The constraint did not merely limit the options. It punished the incremental approach itself, and that punishment is the hinge on which the entire passage turned.

Why did the design constraint punish incremental bargaining?

Incremental bargaining trades one preference for another, and each preference kept shrinks the base that must carry the revenue load. A smaller base forces higher rates to hold revenue constant, which erases the rate cuts that justified the exercise. Members discovered that every conventional deal made the product worse, until abandoning incrementalism became the only way to meet the mark.

The three demands deserve separate attention, because each did distinct political work. Revenue neutrality, the requirement that the bill raise the same revenue as existing law, protected the Treasury and deprived opponents of the simplest attack, that reform was a giveaway. It also created the mark against which the Finance Committee measured itself, the target that Packwood found himself twenty-nine billion dollars short of in April 1986. Without a revenue target, a committee can always close a gap by accepting a smaller rate cut or a larger deficit. With one, the gap is a public fact, and the twenty-nine-billion-dollar figure became the procedural trigger for everything that followed. Distributional neutrality, the requirement that the burden across income groups remain as it was, protected the coalition from the most corrosive form of division. A bill that shifted the load visibly from one class to another would have given the losers a grievance around which to organize, and the reform would have died in the familiar trench warfare of class politics. By holding distribution constant, the designers denied every faction the language of victimhood and forced the debate onto the structure of the system rather than its incidence.

Rate reduction was the demand that made the other two politically bearable. Nobody endures the repeal of a cherished preference for the abstract satisfactions of a broader base. Members endure it for a lower marginal rate they can defend to their constituents as the price of a simpler and fairer system. The rate cut was the benefit that justified the pain, and its size determined how much pain the coalition would tolerate. This is why the arithmetic was so unforgiving. A small rate cut could not justify large repeals, and large repeals were the only way to fund a large rate cut. The system had exactly one stable equilibrium, the maximum feasible base broadening funding the maximum feasible rate reduction, and every partial version of the bill was unstable by comparison.

The simultaneity of the three demands is what made the constraint binding. Any two of the three, taken alone, would have left room for ordinary bargaining. Revenue neutrality plus rate reduction, without distributional neutrality, would have allowed the burden to shift between income groups, opening the class-warfare dimension the designers wanted closed. Revenue neutrality plus distributional neutrality, without rate reduction, would have produced a revenue-neutral simplification with no political payoff, a bill nobody would have fought for. Rate reduction plus distributional neutrality, without revenue neutrality, would have allowed the bill to lose revenue, turning the reform into a deficit-financed cut that the fiscal context would not support. Only all three together eliminated the compromise region. The designers may not have reasoned it out in these terms, but the procedure enforced the logic regardless. Every attempt to relax one demand reintroduced the bargaining space that the other two were meant to close, and every such attempt failed, until the process converged on the single point where all three held. Committees do not normally legislate at equilibria. They legislate at compromises. The 1986 process is unusual because the constraint eliminated the compromise region and left only the extreme.

It is worth pausing on why the constraint took this particular form. The demand for revenue neutrality came from the fiscal context and from the need to keep the bill off the deficit debate. The demand for distributional neutrality came from the political need to prevent the reform from being captured by either party’s class narrative. The demand for rate reduction came from the President, who had made lower marginal rates the centerpiece of his economic program and who needed a deliverable achievement. None of these demands was negotiable within the process, because each was owned by a different veto player. The Treasury would not accept revenue loss. Neither party would accept a distributional shift it could be blamed for. The President would not accept a bill without rate cuts. A constraint owned by no one can be bargained away. A constraint owned by three different someones cannot, and that is why the arithmetic held through every crisis.

Distributional neutrality performed a quieter but equally essential procedural function. By requiring that the burden across income groups remain unchanged, it deprived every participant of the most potent weapon in revenue politics, the claim that the other side was shifting the load. In an ordinary tax bill, members argue about who pays, and the argument organizes the chamber along class lines that are difficult to bridge. In the 1986 process, the distributional question was settled before the legislating began, which meant the chamber could not organize around it. The debate was forced onto the structure of the system, onto the question of which preferences to repeal and how low the rates could go, rather than onto the incidence of the burden. This procedural effect is easy to overlook because it consists of an argument that did not happen. But the absence of class warfare from the 1986 debate is one of its most remarkable features, and distributional neutrality is the reason for the absence. The constraint did not merely limit the bill’s substance. It structured the debate itself, channeling disagreement into the one dimension where agreement was possible.

Understanding the constraint also clarifies what the famous personalities in this story actually did. Ronald Reagan supplied the demand for lower rates and the willingness to spend presidential capital. Dan Rostenkowski supplied the House vehicle and the willingness to mark up a bill his own party’s interest groups opposed. Bob Packwood supplied the decisive procedural innovation in the Senate. But none of them invented the arithmetic. The constraint channeled their efforts the way a riverbed channels water, and the same individuals, operating later without such a constraint, could not reproduce the result. That is the complication this article will develop at length: the design did most of the work, and the heroes were its instruments rather than its authors.

Three Blueprints and a Starting Vehicle

The reform did not begin with H.R. 3838. It began with a presidential request and three competing blueprints, each of which taught the eventual bill something about the constraint. In his January 1984 State of the Union address, Reagan asked the Treasury Department to study the federal revenue system and recommend changes. The result was Treasury I, formally titled Tax Reform for Fairness, Simplicity, and Economic Growth: The Treasury Department Report to the President, released in November 1984, with a transmittal signed by Treasury Secretary Donald T. Regan. Treasury I proposed a sweeping simplification, and its reception demonstrated both the appetite for reform and the ferocity of the interests it threatened. After the 1984 election, the White House produced a revised plan, The President’s Tax Proposals to the Congress for Fairness, Growth, and Simplicity, dated May 29, 1985, known as Treasury II. Between the two Treasury reports, the locus of responsibility shifted: James A. Baker III became Treasury Secretary in February 1985, succeeding Regan, and Deputy Treasury Secretary Richard Darman became a key internal advocate for the reform effort. Darman would later back the Packwood substitute in the Senate, in the account given by Jeffrey H. Birnbaum and Alan S. Murray in their 1987 history of the passage.

Congress had its own blueprints. The Fair Tax Act, associated with Senator Bill Bradley, Democrat of New Jersey, and Representative Dick Gephardt, Democrat of Missouri, was first introduced in 1982 during the 97th Congress and reintroduced in 1983 as S.1421 in the 98th Congress, titled the Fair Tax Act of 1983 and introduced on June 8, 1983. Bradley-Gephardt demonstrated that base broadening for rate reduction could attract a bipartisan constituency and gave the concept a legislative pedigree outside the executive branch. The competing Republican blueprint was the Fair and Simple Tax Act, known as FAST, associated with Representative Jack Kemp, Republican of New York, and Senator Robert Kasten, Republican of Wisconsin. Kemp-Kasten appeared in the 98th Congress as H.R. 6165 and S. 2948 and returned in the 99th Congress as H.R. 777 and S. 325. Between Treasury I, Bradley-Gephardt, and Kemp-Kasten, the idea of trading preferences for rates had been tested in every forum that mattered, and each version confirmed the same lesson: the politics worked only if the trade was kept clean, with no side payments that violated the arithmetic.

The Treasury reports deserve a closer look, because they show the executive branch learning the constraint in public. Treasury I, released in November 1984 with Secretary Regan’s transmittal, was the product of the study Reagan had requested in his January 1984 State of the Union. It proposed sweeping simplification and demonstrated that the administration was serious about restructuring rather than merely cutting. But Treasury I also demonstrated the political cost of specificity. A Treasury report can propose the repeal of any preference at no political cost to any member of Congress, which is precisely why its proposals alarmed the interests that would later fill Gucci Gulch. Treasury II, issued from the White House on May 29, 1985, as the President’s Tax Proposals to the Congress for Fairness, Growth, and Simplicity, was the revised and politically seasoned version. The interval between the two reports coincided with the change at Treasury, as James A. Baker III succeeded Donald T. Regan in February 1985, bringing a different set of political instincts to the management of the effort. Deputy Secretary Richard Darman’s role as an internal advocate for the reform, later extending to backing the Packwood substitute in the Senate, shows how the executive branch’s commitment survived the personnel change and carried into the legislative phase.

The congressional blueprints played a different procedural role. Bradley-Gephardt, the Fair Tax Act, first introduced in 1982 and reintroduced in 1983 as S.1421 in the 98th Congress with introduction on June 8, 1983, proved that base broadening for rate reduction was not solely a Republican or executive project. Bradley in the Senate and Gephardt in the House gave the concept sponsors in both chambers and both parties’ reform wings, and their bill’s persistence across two Congresses kept the idea alive during the long interval before the administration acted. Kemp-Kasten, the Fair and Simple Tax Act, performed the mirror function on the Republican side. With Kemp in the House and Kasten in the Senate, and with bill numbers H.R. 6165 and S. 2948 in the 98th Congress and H.R. 777 and S. 325 in the 99th, FAST gave Republican members a reform vehicle of their own and prevented the effort from being typecast as a Democratic redistribution scheme. The existence of competing blueprints is often treated as a sign of disarray. In this case it was a procedural asset. By the time Rostenkowski introduced H.R. 3838, every faction in the debate had already invested in some version of the trade, which meant every faction had a reason to keep the process alive even when its own version was not the one moving.

The timing of the 99th Congress framed the entire effort. With the presidential election of 1984 settled and the next congressional elections not until November 1986, the window for controversial legislation was open but narrowing. The House acted in December 1985, the Senate committee in the spring of 1986, the Senate floor in June, and the conference in September, each phase consuming the time its procedural work required while the electoral calendar ticked. Had any phase slipped by a few months, the bill would have collided with the pre-election period, when members avoid difficult votes, and the reform would have been postponed to a future Congress with no guarantee of revival. The calendar was thus a silent veto point throughout, and the fact that each phase finished in time was itself a procedural achievement. Speed mattered not only on December 17 but at every stage, because the electoral clock was running from the day H.R. 3838 was introduced.

The choice of vehicle mattered procedurally. H.R. 3838 was a House bill from the tax-writing committee, which meant it carried the Origination requirement cleanly and gave Rostenkowski control of the first markup. The Ways and Means markup ran for twenty-six days, from September 18 to December 3, 1985, an unusually long session that reflected the difficulty of assembling a bill under the constraint. On December 3 the committee ordered the bill reported, and on December 7 it was reported to the House as H. Rept. 99-426. The mechanics of committee markups of this kind are examined in the companion piece on the committee system and markups. What the markup produced was a bill that satisfied the constraint on paper but protected enough preferences to leave its coalition fragile, a fragility the House would expose within days.

Rostenkowski’s role in this phase illustrates what a committee chairman can and cannot do procedurally. As chairman of Ways and Means, he controlled the markup’s agenda, the order in which provisions were considered, and the timing of the vote to order the bill reported. Those powers were sufficient to produce H. Rept. 99-426, a reported bill that met the constraint. They were not sufficient to protect the bill on the floor, because the rule belonged to a different procedural jurisdiction. The chairman’s authority ends where the floor’s begins, and the December 11 defeat demonstrated the boundary. Rostenkowski could deliver a bill out of committee. He could not deliver the terms on which the chamber would consider it. That limitation is structural, not personal, and it explains why the House phase required a presidential rescue. The actor who controlled the markup did not control the rule, and the actor who could move the rule was the President.

The House Phase: Markup, Rule Defeat, and Voice Vote

The House phase of the passage contains the procedural defeat that nearly ended the entire effort, and it is the event the One Test requires every reader to name. On December 11, 1985, the House rejected the rule for considering H.R. 3838, H.Res. 336, by a vote of 223 to 202. Only fourteen Republicans voted for the rule. The numbers deserve attention because they reveal the character of the defeat. This was not a vote on the bill. It was a vote on the terms of debate, and its rejection meant the bill could not reach the floor at all. A rule defeat is in some ways more dangerous than a bill defeat, because it strands the legislation without a recorded judgment on its substance, leaving opponents free to claim the measure lacked support while supporters have no vote to rally around.

The standard route of revenue legislation through Congress, from introduction through committee markup, floor action, and conference, is described in how tax bills move through Congress. The 1986 House phase followed that route in outline and broke it at the critical joint.

How did a rule defeat differ from a bill defeat in House procedure?

A bill defeat records the chamber’s judgment on substance and ends the matter cleanly. A rule defeat denies the terms of debate, stranding the bill without any verdict on its merits. The December 11 rejection therefore left H.R. 3838 in limbo: unjudged, unamendable, and procedurally homeless until its supporters could assemble a new majority for a new rule.

The vote count itself is worth parsing as procedural evidence. A 223 to 202 division is close, but the composition matters more than the margin. Only fourteen Republicans voted for the rule, which meant the bill’s own nominal allies had abandoned the procedural motion even though many of them supported reform in principle. The rule vote thus functioned as a costless protest. Members could signal to organized interests that they had stood against the bill without ever voting against its substance, because there was no substantive vote to cast.

The fourteen Republicans who voted for the rule are the most telling detail in the tally. In a chamber where the Republican president had made tax reform his initiative, only fourteen members of his party would vote to bring the bill to the floor. The figure measures the distance between the administration’s commitment and the congressional party’s, and it shows how isolated the reformers were within their own coalition. The fourteen were not enough to carry the rule, but they were enough to keep the possibility alive, because they demonstrated that a pro-reform Republican position existed and could be expanded with sufficient pressure. Reagan’s lobbying did not create those fourteen votes. It multiplied them, by giving the wavering members a reason to join a position that already existed rather than to invent one. The fourteen are thus the procedural seed of the December 17 revival, the evidence that the rule could pass if the cost structure changed. This is the structural reason rule defeats are so dangerous to complex legislation. They allow opponents to kill a bill while denying its supporters the clarifying power of a substantive roll call. Rostenkowski, who had just completed a twenty-six-day markup and secured H. Rept. 99-426, suddenly held a reported bill with no path to the floor and no procedural vehicle for getting one.

The revival mechanism is equally instructive as procedure. Reagan’s personal lobbying of the defectors, including his rare trip to the Capitol, worked not by changing minds about the bill but by changing the cost structure of the rule vote. A member who votes against a rule to please a lobbyist faces little consequence when the issue is procedural and obscure. The same member faces a different calculation when the President of his own party has personally asked for the vote and will remember the answer. Presidential lobbying is a scarce resource, and its deployment on a rule vote, normally a matter of party routine, signaled that the administration treated the procedural motion as a substantive test. The six days between December 11 and December 17 thus illustrate a general principle of legislative procedure. Rules are not neutral traffic management. They are substantive votes in procedural dress, and they can be moved by the same instruments of pressure that move substantive votes, provided someone is willing to spend the capital.

What did the Ways and Means markup accomplish before the rule vote?

The twenty-six-day markup from September 18 to December 3, 1985, converted the Treasury and congressional blueprints into reportable legislative text, H. Rept. 99-426. It proved the constraint could be met on paper. But the reported bill preserved enough preferences to leave its coalition brittle, which the rule vote four days after reporting exposed.

The revival of the bill after December 11 came from the one actor who could supply what the House could not generate internally: presidential pressure applied personally. Reagan lobbied the defectors himself, making a rare trip to the Capitol to work the members who had voted against the rule. The exact date of that trip is not verified in the record, and this account asserts none, but the lobbying campaign it anchored is well documented. The mechanism of revival is worth naming precisely, because it illustrates a recurring pattern in the passage: at each death, the rescuer was whoever could change the procedural calculus rather than whoever had the best policy argument. Reagan could not rewrite the bill, but he could make voting against the rule more costly than voting for it, and that was sufficient.

On December 17, 1985, the House passed the rule, and on the same day it passed H.R. 3838 itself by voice vote. The voice vote deserves emphasis for what it signals procedurally. A recorded tally would have forced every member to take a public position on a bill whose preferences still offended organized interests in both parties. A voice vote allowed the chamber to advance the measure without manufacturing hundreds of individual targets for retaliation. The House thus moved the bill while minimizing the political cost of moving it, a choice that reflected how fragile the coalition remained. The bill that left the House was alive, but it was not yet the radical measure it would become. That transformation required the second death and the second revival, both of which belonged to the Senate.

There is a further procedural point about the December 17 sequence that is easy to miss. The House passed the rule and then passed the bill on the same day, compressing into hours a process that normally spans days or weeks. The compression was not incidental. Once the rule was secured, the leadership had every incentive to move the bill before the coalition could decay, because the same members who had needed presidential pressure to support the rule might reconsider if given time to hear from organized interests. Speed is itself a procedural instrument, and the same-day sequence shows it being used deliberately. The voice vote compounded the effect. By avoiding a recorded tally, the House denied opponents a target list and denied wavering members a moment of maximum exposure. The combination of speed and voicelessness moved the bill through the chamber with the minimum possible friction, which was exactly what a fragile coalition required. The House phase thus ended not with a triumph but with an escape, and the escape was engineered through procedure at every step.

The Senate Committee Phase: The Substitute

The Senate Finance Committee received the House bill and began its own markup under Chairman Packwood, with Bill Diefenderfer serving as the committee’s chief of staff and the chairman’s counsel. What followed was a demonstration of the constraint’s cruelty. For weeks the committee bargained in the conventional incremental style, restoring this preference and protecting that one, and with each deal the arithmetic deteriorated. On April 18, 1986, Packwood suspended the markup. His words to the committee are a matter of record: “We are now $29 billion off of our mark… there will be no votes this morning.” He later described the moment bluntly: “I finally just adjourned the committee and said, ‘We are done’… ‘we will have no more sessions’.” The committee that had been charged with improving the House bill had produced, through normal bargaining, a bill that could not meet its own revenue target and that nobody could defend. This was the second death, and it was the most instructive of the three, because it was caused not by opponents of reform but by the reform’s own friends practicing ordinary legislative craft.

The mark itself deserves explanation as a procedural instrument. When Packwood said the committee was twenty-nine billion dollars off its mark, he was referring to the revenue target the bill had to hit to satisfy the neutrality constraint. The mark is not an aspiration. It is a number produced by the revenue estimators, and it functions as the committee’s binding instruction. A bill that misses its mark cannot be reported as revenue neutral, and a bill that cannot be reported as revenue neutral cannot keep the coalition that revenue neutrality assembled. The twenty-nine-billion-dollar figure is therefore not color. It is the quantitative measure of the incremental approach’s failure, the distance between what ordinary bargaining had produced and what the constraint required. Packwood’s announcement of the figure to the committee was a procedural act. By stating the shortfall publicly, he made it impossible for the committee to pretend the bill was close enough, and he created the record on which the suspension of the markup would be justified. The mark is the constraint made visible, and the twenty-nine billion dollars is the constraint’s verdict on incrementalism.

What happened next is the single strategic decision the One Test requires every reader to explain. Packwood and Diefenderfer met at the Irish Times, a pub near the Capitol, and over two famous pitchers of beer sketched on a napkin a substitute that abandoned the incremental approach entirely. Instead of bargaining over which preferences to keep, the substitute eliminated them wholesale and used the recovered base to buy dramatically lower rates. The conventional account, which this article follows, is that of Birnbaum and Murray, whose 1987 book Showdown at Gucci Gulch: Lawmakers, Lobbyists, and the Unlikely Triumph of Tax Reform remains the canonical history of the passage. The napkin has become the most famous anecdote in modern legislative lore, but the procedure is what matters: Packwood returned to the committee not with a repaired version of the failed bill but with a replacement that changed the nature of the choice before the members.

What did an amendment in the nature of a substitute let the Finance Committee do?

A substitute replaces the entire text referred to the committee while preserving the bill’s procedural identity and number. Packwood could therefore discard months of failed incremental bargaining without losing H.R. 3838’s place in the legislative pipeline. The committee voted on a new bill wearing the old bill’s number.

The procedural device deserves a fuller explanation, because it is the hinge of the entire passage. An amendment in the nature of a substitute does not amend the referred text. It replaces it entirely, striking everything after the enacting clause and inserting new language, while the bill retains its original number, its place in the legislative calendar, and its procedural history. For H.R. 3838, this meant that months of House action, the introduction by Rostenkowski, the Ways and Means markup, H. Rept. 99-426, the rule fight, and the voice vote, all remained attached to a bill whose substantive content had been completely rewritten in the Senate. The device is normally a convenience, a way to avoid reintroducing a heavily amended bill. Packwood turned it into an instrument of radicalism, using the continuity of the vehicle to smuggle a discontinuity of substance through the process. The Senate was not voting on the House’s bill as amended. It was voting on Packwood’s bill wearing the House bill’s clothes.

The personnel of the revival matter as procedure too. Bill Diefenderfer, the Finance Committee’s chief of staff and Packwood’s counsel, was not a public figure, but his role in the Irish Times meeting illustrates how committee staff function as procedural engineers. The chairman supplies the authority to act. The staff supplies the capacity to draft, to calculate, and to translate a napkin sketch into legislative language that can survive a markup. Richard Darman’s backing of the Packwood plan, recorded by Birnbaum and Murray, supplied the executive branch’s acquiescence, which mattered because a substitute that the administration opposed would have faced a veto threat and died in committee. The coalition for the substitute was thus assembled before the committee ever voted: the chairman’s authority, the staff’s drafting capacity, and the administration’s tolerance, all aligned behind a text that did not yet formally exist.

The 20 to 0 vote that followed is one of the most remarkable procedural facts in modern revenue legislation, and it requires explanation rather than admiration. Unanimous committee votes are normally produced by compromise, by sanding down the controversial edges until no member has a reason to dissent. This unanimity was produced by the opposite method, by sharpening the bill until every member’s objection was answered by the same logic. A member who wanted a particular preference restored had to explain why the rate should rise to pay for it. A member who wanted a higher rate had to explain which preference should be repealed to fund it. The constraint made every objection self-defeating, and the substitute’s comprehensiveness made the self-defeat visible. The committee did not agree on the substance. It agreed that no alternative substance could survive the arithmetic, and unanimity was the procedural expression of that agreement. The order to report on May 6, 1986, and the report to the Senate on May 29 as S. Rept. 99-313, formalized a consensus that had been manufactured by the constraint rather than discovered through deliberation.

The logic of the substitute is the heart of this article, and it must be stated in the exact formulation the brief requires: The radicalism escape: when incremental bargaining produced a bill nobody could defend, the only viable move was to become more radical rather than less, because only a sweeping elimination of preferences could buy a rate low enough to be worth voting for, and the 1986 act is the clearest documented case of escalation as a legislative survival strategy.

How could a more radical substitute win votes that a moderate draft could not?

A moderate draft spread pain widely while concentrating benefits nowhere, giving every affected interest a reason to fight and no member a reason to champion it. The radical substitute bought a rate cut deep enough that the broad middle of each party could defend the trade it required.

Consider why the escalation worked procedurally where moderation had failed. A bill that keeps half the preferences can buy only a modest rate cut, and a modest rate cut is not worth the political pain of repealing the other half. Members asked to vote for such a bill bear all the costs of reform and receive few of the benefits. A bill that eliminates nearly all preferences, by contrast, can buy a rate cut large enough to change the political calculation: the pain is greater, but the reward is transformative, and the breadth of the base broadening means no single interest can plausibly claim it was singled out. Packwood’s substitute thus solved a collective action problem by making the action more collective. The committee approved the substitute by 20 to 0, ordered it reported on May 6, 1986, and reported it to the Senate on May 29, 1986, as an amendment in the nature of a substitute under S. Rept. 99-313. Reagan’s comment on the unanimous report captured the procedural significance: “getting a tax reform bill reported in your committee by a unanimous vote.” A committee that had been twenty-nine billion dollars off its mark six weeks earlier had become unanimous, not by compromising but by escalating.

The 20 to 0 vote also illustrates the series thesis that procedure determined the text. The substitute’s text was not the product of deliberation converging on the best policy. It was the product of a procedural crisis that eliminated every option except radicalism. Had the committee been only five billion dollars off its mark, it might have patched the bill and reported a moderate measure. Because it was twenty-nine billion off, patching was arithmetically hopeless, and only the wholesale approach could close the gap. The crisis improved the statute. That is a rare claim to make about legislative procedure, and the Finance markup of April and May 1986 is the rare case that supports it.

The Senate Floor: Ninety-Seven to Three

The full Senate took up the Finance Committee’s substitute in June 1986, and the floor phase supplied the third death. The threat came from amendments affecting individual retirement accounts.

The move from committee to floor is the most dangerous transition in the legislative process, because it moves the bill from a controlled environment to an uncontrolled one. In committee, the chairman sets the agenda, the staff controls the drafting, and the members are a known quantity. On the floor, any senator can offer amendments, the press watches every vote, and the organized interests mobilize their full strength. The Finance Committee’s 20 to 0 unanimity was achieved in the controlled environment. The question of the floor phase was whether that unanimity could survive the uncontrolled one, and the IRA amendment fight was the test. The managers’ strategy was to reproduce the committee’s discipline on a larger stage, and the 51 to 48 margin shows how narrowly the reproduction succeeded. The floor phase thus confirms the general principle that committee action is necessary but not sufficient. A bill must survive both the room where it is written and the chamber where it is voted on, and the two survivals require different procedural skills. The IRA provisions were among the most politically sensitive in the bill, touching the retirement savings of millions of voters, and amendments to alter them threatened to unravel the rate structure the substitute had so carefully purchased. Packwood defeated the threatening amendment 51 to 48, a margin of three votes, and his subsequent testimony stated the stakes without embellishment: had he lost a couple, he would have lost on the IRAs and lost the bill. The 51 to 48 fight is the clearest illustration in the passage of how fragile the radical bargain remained even after the committee’s unanimous report. A single amendment, aimed at a single popular preference, could have reopened the entire rate-for-preferences exchange and collapsed the bill back into the incremental bargaining that had already failed once.

The Senate passed H.R. 3838 on June 24, 1986, by 97 to 3, recorded as Record Vote No. 148. The party breakdown was Republicans 53 to 0 and Democrats 44 to 3. The lopsidedness of the final tally, set against the 51 to 48 near miss on the IRA amendment, captures the essential dynamic of the floor phase. The coalition for the bill as a whole was enormous, but the coalition for any particular piece of it was narrow, and the bill survived because its managers kept the chamber voting on the whole rather than the pieces. That is a procedural achievement, not a substantive one. The Senate did not pass the bill because ninety-seven senators agreed on every preference eliminated. It passed because the substitute’s structure made the package indivisible in practice: to defend any single preference was to attack the rates, and to attack the rates was to attack the bill.

The IRA amendment fight that tested this indivisibility deserves reconstruction as procedure, because it shows how close the bill came to reverting to the incremental pattern. Individual retirement accounts were among the most widely held preferences in the code, touching voters in every state and every income group that saved. An amendment altering the IRA provisions therefore had a natural constituency far broader than the organized interests in Gucci Gulch. Had the amendment passed, it would not merely have changed one provision. It would have demonstrated that the package was divisible after all, that a sufficiently popular preference could be carved out without consequence. Every other protected interest would have demanded the same treatment, the base would have narrowed amendment by amendment, the rates would have risen to compensate, and the bill would have collapsed back into the twenty-nine-billion-dollar shortfall that had killed the incremental version in April. Packwood’s 51 to 48 victory, a margin of three, held the line at the exact point where the line was thinnest. His testimony afterward, that losing a couple of votes would have meant losing the IRAs and losing the bill, was not rhetoric. It was a precise description of the procedural mechanics. The bill’s survival required winning every divisibility fight, because losing one would have licensed all the others.

The final Senate vote of 97 to 3, Record Vote No. 148, with Republicans 53 to 0 and Democrats 44 to 3, should be read against that 51 to 48 near miss rather than in isolation. The contrast between the two tallies is the procedural story of the floor phase in miniature. On the amendment, where the question was whether the package could be divided, the coalition held by three votes. On final passage, where the question was whether the package should become law, the coalition held by ninety-four. The difference measures the power of the indivisibility strategy. Senators who would have voted to carve out their favorite preference would not vote to kill the rate cuts, and the managers’ achievement was to ensure that the only vote that mattered was the second one. The three Democratic nays are not named in this account, in keeping with the record’s emphasis on the coalition rather than its exceptions, but their existence is itself procedural evidence. Even at 97 to 3, the bill was not consensual. It was merely unbeatable.

The floor management that produced this result deserves attention as procedure. As Senate Majority Leader, Bob Dole controlled the scheduling of floor business, the allocation of debate time, and the sequencing of amendments, subject to the chamber’s rules and the minority’s rights. Those scheduling powers were the instruments by which the indivisibility strategy was implemented. By structuring the floor consideration so that the package was voted on as a whole and divisibility amendments were defeated in sequence, the leadership ensured that no single amendment could become the vehicle for unraveling the substitute. Dole’s presence as a conferee on the final bill confirms that his role extended beyond scheduling into the substance of the reconciliation. The majority leader is often described as a traffic manager, but in this passage the traffic management was the substance. The order in which the Senate voted determined what the Senate could vote for, and the 97 to 3 result was manufactured as much by the schedule as by the persuasion.

The Conference: Settling the Rates

With the House and Senate having passed different versions, the bill went to a conference committee, the House-Senate panel charged with reconciling the two texts. The conference report was filed in the House on September 18, 1986, as H. Rept. 99-841. The conferees named from the House were Rostenkowski, Pickle, Rangel, Stark, Gephardt, Russo, Pease, Duncan, Archer, Vander Jagt, and Crane. The conferees named from the Senate were Packwood, Dole, Roth, Danforth, Chafee, Wallop, Long, Bentsen, Matsunaga, Bradley, and Moynihan. The rosters themselves tell a procedural story: both tax-writing committees were fully represented, both parties were fully represented, and the principal authors of the competing blueprints, Bradley and Gephardt for the Fair Tax Act, sat alongside the chairmen who had carried the bill. The mechanics of such panels are treated in the companion explainer on conference committees.

The conference settled the question the entire two-year effort had been organized around: the rates. The final agreement set individual brackets at 15 percent and 28 percent and the corporate rate at 34 percent. The Senate bill had set the top individual rate at 27 percent, so the conference’s 28 percent represented a small but meaningful concession to the arithmetic of reconciliation.

What procedural tools did the conference use to settle the rate dispute?

The conference traded the Senate’s 27 percent top individual rate up to 28 percent against House preferences the Senate had resisted, staying inside the revenue-neutral box. Because conference reports cannot be amended on the floor, the conferees could balance the last dollars knowing the package would face a single up-or-down vote.

That the final negotiation concerned a single percentage point on the top individual bracket, after two years of struggle over the structure of the entire revenue system, shows how completely the constraint had done its work. Everything else had been decided by the base-broadening logic. Only the final price remained to be haggled.

The House adopted the conference report on September 25, 1986, by 292 to 136, recorded as Record Vote No. 413, with Republicans voting 116 to 62 and Democrats 176 to 74. The Senate adopted it on September 27, 1986, by 74 to 23, recorded as Record Vote No. 296, with Republicans 41 to 11 and Democrats 33 to 12. Both margins were lopsided, both were bipartisan, and both reflected the same procedural reality that had produced the 97 to 3 Senate vote in June. By September, the bill had become the only vehicle capable of delivering the rate cuts both parties had promised, and voting against it meant voting against the achievement itself.

The conference procedure itself is worth examining, because it is the stage of the process where the constraint operated most quietly and most completely. A conference committee exists to reconcile the differences between the House and Senate versions of a bill, and its members, the conferees, are drawn from the committees of jurisdiction in both chambers. The rosters for H.R. 3838 show the procedure working as designed. From the House came Rostenkowski, Pickle, Rangel, Stark, Gephardt, Russo, Pease, Duncan, Archer, Vander Jagt, and Crane, the leadership of Ways and Means with both parties represented and with Gephardt present as the co-author of the Fair Tax Act blueprint. From the Senate came Packwood, Dole, Roth, Danforth, Chafee, Wallop, Long, Bentsen, Matsunaga, Bradley, and Moynihan, the leadership of Finance with both parties represented, with Dole present as majority leader, Long as ranking Democrat, and Bradley as the Senate co-author of the Fair Tax Act. The presence of the blueprint authors alongside the committee chairmen meant that the conference contained, in miniature, the entire intellectual history of the reform. No faction could claim it had been excluded from the room where the final text was written.

The rate settlement that emerged, individual brackets of 15 percent and 28 percent and a corporate rate of 34 percent, with the Senate’s 27 percent top individual figure giving way to 28, illustrates how the constraint had narrowed the negotiable space. In an ordinary conference, the members haggle over dozens of provisions, trading concessions across titles and sections. In this conference, the base-broadening logic had already settled nearly everything. The only remaining variable was the final price of the trade, the exact rates that the broadened base could fund. That the negotiation concerned a single percentage point on the top individual bracket is not a sign that the conference was trivial. It is a sign that the constraint had done the conference’s work in advance, leaving the members to ratify an arithmetic result rather than to bargain over a political one.

The movement from 27 to 28 percent on the top individual bracket is worth pausing over, because it is the smallest decision in the passage and in some ways the most revealing. The Senate bill had set the top individual rate at 27 percent. The conference settled on 28. One percentage point, applied to the highest bracket, is a rounding error in the context of a bill that restructured the entire revenue system. But the point had to be negotiated because the constraint required the numbers to balance exactly, and the conference was the last place where the balance could be adjusted. The 27 to 28 movement shows the constraint operating at its finest resolution. Even after the wholesale base broadening, even after the indivisible package had survived the floor, the arithmetic still demanded its due, and the conferees paid it. A procedure that can enforce a one-point adjustment at the end of a two-year struggle is a procedure that has governed everything in between, and the final rates, 15 and 28 percent for individuals and 34 percent for corporations, are the constraint’s signature on the statute. The conference report filed in the House on September 18, 1986, as H. Rept. 99-841, was therefore less a compromise than a calculation, and the lopsided adoption votes, 292 to 136 in the House and 74 to 23 in the Senate, recorded the chamber’s acceptance of a result that procedure had made inevitable.

Veto Points: Where the Bill Could Have Died

Every bill that becomes law must pass a series of veto points, moments at which a single actor or a single vote can stop it permanently. Most bills die at the first veto point they encounter, which is why most bills never become law. The 1986 passage is unusual not because the bill avoided veto points but because it encountered an extraordinary number of them and survived each by a different method. Mapping the veto points shows the procedure the way an engineer would, as a sequence of gates, each with its own mechanism and its own key.

The first veto point was the House rule. In the House, major legislation reaches the floor under a rule reported by the Rules Committee and adopted by the chamber, and the rule sets the terms of debate: how long debate will last, which amendments will be allowed, and whether the bill can be amended at all. A rule that fails leaves the bill without access to the floor, which is functionally equivalent to killing it, since a bill that cannot be debated cannot be passed. The December 11 rejection of H.Res. 336 was therefore a veto exercised by the chamber against its own leadership’s scheduling decision. The key that opened this gate was presidential pressure applied to individual members, a key available only because the President chose to spend it. Had Reagan declined to lobby, or had the lobbying failed, H.R. 3838 would have remained a reported bill with no rule, a common enough fate for controversial measures, and the reform would have ended in December 1985.

Behind the rule vote stood an earlier gatekeeper that the record mentions only indirectly. The House Rules Committee reports the rules under which major bills are considered, and its decision to grant a rule, and the terms of the rule it grants, shape everything downstream. A restrictive rule that bars amendments protects a fragile coalition. An open rule that permits them invites the divisibility fights that killed the incremental approach in the Senate. The rule for H.R. 3838, H.Res. 336, was the product of negotiations between the leadership, the Rules Committee, and the bill’s managers, and its rejection by the chamber was a rejection of that negotiated product. The Rules Committee is thus the unmarked first veto point, the gate before the gate, and its work illustrates a general feature of House procedure. The terms of debate are themselves debated, and the debate over the terms can kill a bill before the debate over the substance begins. The December 11 vote is usually described as a defeat for the bill. More precisely, it was a defeat for the rule, which is a distinct procedural event with its own causes and its own remedies.

The second veto point was the Senate Finance Committee markup. A committee chairman who suspends a markup and announces no more sessions has exercised a veto as absolute as any floor vote. No rule requires a chairman to continue marking up a bill, and no appeal lies from the decision to stop. Packwood’s April 18 adjournment was thus a veto exercised by a single actor against his own committee’s product, and it is the most paradoxical of the three deaths, because the veto was deployed not to kill the reform but to kill the version of it that could not work. The key that opened this gate was the substitute, a new text that gave the chairman a reason to reconvene. The veto and the key were held by the same person, which is why the second death and the second revival form a single procedural event. Packwood killed the bill he could not defend so that he could revive the bill he could.

The third veto point was the Senate floor amendment process. Unlike the House, where the rule structures amendments in advance, the Senate’s floor procedures allow a wide range of amendments, and each amendment is a potential veto point, a moment at which the bill’s coalition can be fractured. The IRA amendment fight of June 1986 was the veto point that came closest to succeeding, because the amendment targeted the precise joint in the bill’s armor, the popular preference whose protection would have demonstrated the package’s divisibility. The key that held this gate was the 51 to 48 defeat of the amendment, a margin so narrow that Packwood’s testimony about losing a couple of votes reads as a literal description of the procedural margin rather than as dramatic emphasis. Three votes separated the bill from the cascade that would have followed.

The fourth veto point was the conference, or more precisely the adoption of the conference report in each chamber. A conference report cannot be amended on the floor. It must be accepted or rejected as a whole, which makes the adoption votes a final veto point at which either chamber can kill months of work with a single no. The 292 to 136 House vote and the 74 to 23 Senate vote show this gate opening widely, but the width of the opening should not obscure the gate’s existence. Had the conference deadlocked on the rates, or had either chamber rejected the report, the bill would have died in September after surviving everything else. The key here was the narrowness of the remaining disagreement. Because the constraint had settled everything except the final rates, the conference had little to fight over, and the report that emerged was acceptable to majorities of both parties in both chambers.

The fifth and final veto point was presentment, the President’s decision to sign or veto. This gate opened without resistance on October 22, 1986, because the President who received the bill was the President who had demanded it.

The veto points also differ in who can exercise them, and the difference shaped the revival strategies. The rule veto was exercised by the chamber as a whole, 223 members voting no, which meant the revival had to move dozens of votes and required presidential pressure. The markup veto was exercised by a single chairman, which meant the revival had to move one mind and required a new text that the chairman could embrace. The amendment veto was exercised by a potential majority of the Senate, which meant the revival had to hold a coalition together vote by vote and required the discipline that produced the 51 to 48 margin. The conference veto could have been exercised by either chamber, which meant the revival had to satisfy both simultaneously and required a report that split the remaining difference. The presentment veto belonged to the President alone, and it was never exercised because the President’s preferences aligned with the bill’s. Each veto point’s revival strategy was thus tailored to its gatekeeper. Mass pressure for the mass veto, a new text for the single veto, coalition discipline for the majoritarian veto, and a balanced report for the dual veto. The passage is a catalog of matched strategies, each fitted to the procedural lock it had to open. But the formality matters procedurally. A bill that has passed both chambers is not yet law, and the signature is the act that converts legislative procedure into legal reality. The quiet of the final veto point, after the noise of the first four, is the procedural signature of a bill whose fights were all fought upstream.

The Four Near-Deaths

The passage is best remembered through its crises, and the brief for this article requires them to be tabulated. Three of the near-deaths were discrete procedural events. The fourth was the standing verdict of the commentariat, which Reagan acknowledged when he told the Tax Reform Coalition the bill had been declared dead so many times by the pundits. The table records each moment, the mechanism that produced it, the actor who reversed it, and what had to be conceded to secure the revival.

The four near-deaths table

moment declared dead	mechanism that killed it	who revived it	what was conceded to do so
December 11, 1985: the House rejects the rule for H.R. 3838, H.Res. 336, by 223 to 202, with only fourteen Republicans in favor	The rule’s defeat denied the bill any path to floor debate and left it stranded without a verdict on its substance	President Ronald Reagan, through personal lobbying of the defectors, including a rare trip to the Capitol	Time and momentum: the bill sat in limbo until the rule passed on December 17, 1985, compressing the House schedule
April 18, 1986: Chairman Packwood suspends the Senate Finance markup, declaring the committee twenty-nine billion dollars off its mark with no more sessions	Weeks of incremental bargaining had produced a bill that could not meet its revenue target and that no member could defend	Chairman Bob Packwood and chief of staff Bill Diefenderfer, with the radical base-broadening substitute sketched over two pitchers of beer	The incremental bill itself: its accumulated preference protections were abandoned for wholesale base broadening
June 1986: the Senate floor fight over amendments affecting individual retirement accounts	A single amendment threatened to reopen the rate-for-preferences exchange and unravel the substitute’s structure	Chairman Packwood, prevailing 51 to 48 and testifying that losing a couple of votes would have lost the IRAs and the bill	Nothing structural: the coalition held at 51 to 48 and the rate framework survived intact
1985 to 1986: the pundits’ serial obituaries across the two-year effort	Commentators repeatedly pronounced the reform finished, a verdict Reagan himself quoted to the Tax Reform Coalition	The procedural pipeline: House passage by voice vote, a 20 to 0 Finance report, and a 97 to 3 Senate passage	Nothing: the obituaries extracted no legislative price, because procedure outlasted commentary

Two observations about the table deserve emphasis. First, the mechanisms of death were procedural in every case: a rule, a markup suspension, an amendment, a narrative. None of the three real crises was a substantive defeat on the merits of the bill. Second, the concessions diminished as the process advanced. The first revival cost time. The second cost the incremental bill. The third and fourth cost nothing, because by June the structure had become self-sustaining. That progression is itself evidence for the series thesis. Procedure did not merely carry the text. At the decisive moment, procedure improved it.

Why the Bill’s Enemies Could Not Coordinate

The passage poses a puzzle that the near-deaths table alone does not resolve. The bill threatened every organized interest with a preference to protect, and those interests were well funded, well represented in Gucci Gulch, and fully aware of what was at stake. Why did they not coordinate to kill it? The answer is procedural, and it follows from the constraint. The interests could not coordinate because the constraint made their interests mutually exclusive. Each preference protected required a higher rate, and the higher rate harmed every other interest’s constituents. A coalition of preference defenders would have had to agree on whose preferences to sacrifice to fund whose protection, which is the same bargaining problem the Finance Committee had already failed to solve. The lobbyists were numerous, but they were numerous as individuals, not as a coalition, and the arithmetic prevented them from becoming one.

The structure of the legislative process compounded the coordination problem. The bill moved through multiple veto points in sequence, and defeating it required winning at every point while its supporters needed to win only once at each. An interest that defeated the rule in December would have had to defeat the substitute in May, the package in June, and the conference report in September. The supporters, by contrast, needed only to hold each gate once. Sequential veto points favor the defense when the defense is organized around a single indivisible proposition, because the attackers must solve their coordination problem repeatedly while the defenders solve theirs once. The 20 to 0 committee vote and the 97 to 3 Senate vote were the procedural expressions of a solved coordination problem on the defense. The corridor never produced an equivalent expression on the offense, because the constraint denied it the common ground on which a coalition could have been built.

There is a further reason, and it concerns the distribution of the pain. The substitute’s comprehensiveness meant that no single interest could plausibly claim to have been singled out. In an incremental bill that repeals half the preferences, the losers can identify themselves, organize, and demand restoration. In a wholesale bill that repeals nearly all of them, the losers are everyone, which means the grievance is universal and therefore politically diffuse. Universal grievances do not organize. They dissipate into the general unhappiness that accompanies all major legislation and that no coalition can weaponize. Packwood’s escalation thus solved the opponents’ coordination problem by dissolving it. The more radical the bill, the harder it was to assemble a coalition against any particular part of it, because every part was everyone’s part. This is the political complement to the arithmetic argument, and it is why the radicalism escape worked as politics as well as procedure.

Gucci Gulch: the lobbying environment

No account of the procedure is complete without the hallway. “Gucci Gulch” was the nickname for the corridor outside the Senate Finance Committee hearing room, where leading tax lobbyists plied their trade during the markup. The name came from their footwear: expensive Italian shoes, the uniform of the Washington tax bar at its most prosperous. The term was popularized by the book that made the legislative story famous, “Showdown at Gucci Gulch: Lawmakers, Lobbyists, and the Unlikely Triumph of Tax Reform” by Jeffrey H. Birnbaum and Alan S. Murray, published in 1987. The book’s title fixed the phrase in the political vocabulary, and its reporting remains the canonical narrative account of the passage.

The lobbying environment should be described factually, not moralistically. The lobbyists in Gucci Gulch were doing what the system invited them to do: representing clients with intense interests in particular provisions, supplying technical arguments and political intelligence to committee members, and seeking to preserve the preferences on which their clients relied. Their presence was a constant of the process, not a scandal within it. What is procedurally significant is that they lost. The preference defenders had every structural advantage, concentration, organization, expertise, access, and the incremental draft of April was their natural habitat, because incremental bargaining gave each of them a separate negotiation to win. The Packwood substitute defeated them not by out-arguing them but by changing the structure: wholesale base broadening denied them the individual negotiations in which they were strongest, and the rate cut gave members a reason to vote against them that no lobbyist could neutralize. Gucci Gulch is remembered because the lobbyists were visible there. It matters because the final bill was written as if they were not.

The canonical record

The narrative details of the passage, the pitchers, the napkin, the hallway, the $29 billion declaration, come principally from one source: “Showdown at Gucci Gulch: Lawmakers, Lobbyists, and the Unlikely Triumph of Tax Reform,” by Jeffrey H. Birnbaum and Alan S. Murray, published in 1987. The book is the canonical account of the legislative history, and its reporting underlies the standard telling of every episode reconstructed in this article. It is the source for the Irish Times scene, for Darman’s backing of the Packwood plan, and for the popularization of “Gucci Gulch” itself, a phrase that entered the political vocabulary through the book’s title.

The reliance on a single journalistic account deserves a procedural note. Legislative history is reconstructed from multiple kinds of evidence: the Congressional Record, committee reports, vote tallies, and contemporary reporting. The vote tallies and report numbers cited in this article, the 223 to 202 rule vote, the 20 to 0 committee vote, the 97 to 3 Senate passage, the 292 to 136 and 74 to 23 conference votes, the report numbers 99-426, 99-313, and 99-841, are matters of public record. The narrative texture, what was said in the pub, what the hallway looked like, who backed whom behind the scenes, comes from Birnbaum and Murray’s reporting, which has stood as the accepted account for decades. A procedural history told from the record alone would be accurate but bloodless; the canonical narrative supplies the human mechanism without which the procedure is hard to follow. This article uses both, distinguishing where it matters between the documented vote and the reported scene.

The Complication: Constraints, Not Heroes

The heroic version of this story is tempting and false. In that version, Reagan’s will, Rostenkowski’s courage, and Packwood’s daring carried the reform past the lobbyists and the cynics. The personalities were real, and their contributions are recorded elsewhere in this article by name and role. But the heroic version mistakes the instruments for the cause. The cause was the constraint. Rate-for-preferences, revenue neutrality, and distributional neutrality together created a situation in which the only bill that could pass was a radical one, and the participants, whatever their motives, were channeled toward radicalism by the arithmetic. Packwood did not choose the substitute because he was bold. He chose it because the committee was twenty-nine billion dollars off its mark and nothing else could close the gap. Rostenkowski did not mark up the bill because he loved its substance. He marked it up because the vehicle existed and the constraint made the exercise meaningful. Reagan did not lobby the rule defectors because he cherished procedure. He lobbied because without a rule there would be no bill and without a bill no rate cuts.

The test of the constraint’s primacy is what happened when it was abandoned. The same personalities, operating later without the three-part discipline, could not repeat the feat. This article draws no conclusions about revenue politics after 1986, and it offers no verdict on later efforts, but the structural point stands on the record of 1986 alone. Remove revenue neutrality, and the pressure to broaden the base disappears. Remove distributional neutrality, and the coalition fragments along class lines. Remove the rate-reduction imperative, and there is nothing to buy with the pain of repeal. The 1986 passage worked because all three demands operated simultaneously, and it is the clearest documented case of a design constraint doing the legislative work that is usually attributed to leadership. A comparison with the later 2017 effort is available for readers who want to test that proposition against a different constraint regime.

This is also why the anecdote of the beer must not be mistaken for the explanation. The Irish Times meeting is vivid, and Birnbaum and Murray were right to make it famous, but the napkin did not create the arithmetic. It recorded a solution the arithmetic had already dictated. Had Packwood and Diefenderfer ordered coffee instead of beer, the substitute would have looked the same, because the mark was still twenty-nine billion dollars off and only wholesale base broadening could close it. Procedure determining text is the thread of this series, and here the thread runs unusually straight: the crisis was procedural, the solution was procedural, and the statute was better for both.

The heroic version also fails a comparative test that the record of 1986 itself supplies. Consider the distribution of boldness across the participants. Rostenkowski, who carried the House bill, was operating within the conventional incremental framework, and his product was the fragile text that died on the rule vote. Packwood, who authored the substitute, was operating under the pressure of a twenty-nine-billion-dollar shortfall that left him no incremental option. If boldness were a personal trait, it would have manifested consistently. Instead it manifested exactly where the constraint demanded it and nowhere else, which is what one would expect if the constraint, not the character, were doing the work. The same logic applies to Reagan. His personal lobbying in December 1985 is often cited as evidence of presidential leadership, but the lobbying was effective because the rule vote was the binding constraint at that moment, not because presidential attention is generally sufficient to pass legislation. Deployed a month earlier or a month later, the same effort would have had nothing to act on.

There is a final reason to resist the heroic telling, and it concerns the lobbyists. The heroic version casts Gucci Gulch as the dragon and the reformers as the knights, which flatters the participants but misdescribes the mechanism. The advocates in the corridor did not lose because they were outmatched as individuals. They lost because the constraint made their product, the targeted carve-out, arithmetically toxic. Every preference they sought had to be funded by a higher rate, and the rate was the one number every member had promised to cut. The lobbyists were defeated by a spreadsheet, not by a hero, and acknowledging that is not cynicism. It is the precondition for understanding why the victory could not be repeated at will. Spreadsheets can be reconstructed. Heroes cannot be summoned on demand, and the 1986 passage did not require them to be.

The executive branch’s supporting cast illustrates the same point from the other direction. Donald T. Regan signed the transmittal of Treasury I, James A. Baker III succeeded him at Treasury in February 1985 and managed the administration’s legislative strategy, and Richard Darman, as deputy secretary, backed the Packwood substitute in the Senate. Three different officials, with three different temperaments and three different institutional positions, all ended up serving the same constraint. Regan launched the study that defined the problem. Baker managed the politics that kept the effort alive through the House phase. Darman supported the radical solution that the Senate phase required. If the passage had depended on any one of them personally, the personnel change of February 1985 would have disrupted it. Instead the effort continued seamlessly, because what continued was not a personal project but a procedural one, governed by the arithmetic that all three men, in their different roles, were implementing. The interchangeability of the personnel is the strongest evidence that the constraint was the constant and the personalities the variables.

Why did revenue neutrality make the politics harder and the policy better?

Revenue neutrality barred buying votes with unpaid-for benefits, since every concession needed a matching offset with a visible loser. That made coalition-building harder. It also made the result better, because the only surviving path was genuine base broadening, which is why the April crisis produced a stronger statute than the incremental draft.

Two Parties, One Signature

Both parties claim credit for the 1986 act, and both claims are correct. The passage cannot be told as the achievement of one party without falsifying the record, because at every stage the decisive actors came from both sides of the aisle. The presidential demand came from a Republican, Ronald Reagan, who requested the Treasury study in his January 1984 State of the Union, received Treasury I under Donald T. Regan’s transmittal, issued Treasury II from the White House on May 29, 1985, after James A. Baker III had succeeded Regan at Treasury in February 1985, and personally lobbied the House rule defectors in December 1985. The House vehicle came from a Democrat, Dan Rostenkowski of Illinois, whose Ways and Means Committee conducted the twenty-six-day markup and reported H. Rept. 99-426. The Senate transformation came from a Republican, Bob Packwood of Oregon, whose Finance Committee produced the 20 to 0 substitute report, S. Rept. 99-313, with the assistance of chief of staff Bill Diefenderfer and, by Birnbaum and Murray’s account, the backing of Deputy Treasury Secretary Richard Darman.

The intellectual blueprints were bipartisan in the same way. Bradley, a Democrat from New Jersey, and Gephardt, a Democrat from Missouri, authored the Fair Tax Act that gave base broadening its congressional pedigree, and both sat as conferees on the final bill. Kemp, a Republican from New York, and Kasten, a Republican from Wisconsin, authored the competing FAST proposal that kept Republican reformers inside the coalition. On the Senate floor, the majority leader who managed the process was Bob Dole, Republican of Kansas, himself a conferee, while the ranking Democrat on the Finance Committee, Russell B. Long of Louisiana, also served as a conferee and helped hold the Democratic side of the unanimous committee vote. The final tallies confirm the bipartisanship quantitatively: 97 to 3 in the Senate with both parties overwhelmingly in favor, 292 to 136 and 74 to 23 on the conference report with majorities of both parties voting yes in both chambers.

The bipartisanship was procedural as well as numerical, and the distinction matters. Numerical bipartisanship means members of both parties voted yes. Procedural bipartisanship means members of both parties performed indispensable functions that the other party could not have performed alone. Reagan, a Republican president, supplied the demand for rate cuts and the rescue of the rule vote, functions no Democrat could have performed because the defectors who needed lobbying were disproportionately Republican. Rostenkowski, a Democratic chairman, supplied the House vehicle and the markup, functions no Republican could have performed because Ways and Means was under Democratic control. Packwood, a Republican chairman, supplied the substitute, which no Democrat could have supplied because Finance was under Republican control. Bradley and Gephardt, Democrats, supplied the congressional blueprint that gave the effort its legislative pedigree, while Kemp and Kasten, Republicans, supplied the competing blueprint that kept Republican reformers invested. Dole managed the floor as majority leader. Long held the Democratic side of the Finance Committee as ranking member. Remove any one of these contributions and the procedure stalls at the stage that contribution unlocked. The bill was bipartisan not in the sentimental sense that everyone agreed, but in the structural sense that the procedure required both parties’ institutional positions at every stage.

This structural bipartisanship also explains why the credit is correctly claimed by both parties. Each party’s contribution was necessary and neither was sufficient, which is the definition of a joint product. The Democrats can point to the House vehicle, the Fair Tax Act blueprint, and the ranking member’s cooperation. The Republicans can point to the presidential initiative, the Treasury reports, the FAST blueprint, the Finance substitute, and the floor management. Neither account is complete without the other, and the attempt to assign sole authorship to either party would require ignoring the stages at which the other party’s actors were indispensable. The neutrality of this article on that question is not evasion. It is fidelity to a record in which the procedure itself was bipartisan by construction.

The conference rosters make the structural point concrete. Eleven conferees from each chamber, drawn from the tax-writing committees, with both parties represented on each side and with the blueprint authors seated alongside the chairmen. A conference constituted on purely partisan lines would have produced a report that the minority could disavow, and the adoption votes would have been narrower. The bipartisan rosters produced a report that both parties had to own, because both parties’ institutional representatives had written it. The adoption tallies reflect this ownership. In the House, 116 Republicans and 176 Democrats voted yes. In the Senate, 41 Republicans and 33 Democrats voted yes. Those are not the numbers of a partisan imposition. They are the numbers of a joint product, and the jointness was built into the procedure before a single vote was cast. The conferee lists are therefore not ceremonial. They are the documentary evidence of how the procedure distributed authorship, and distributed authorship is what made the final votes lopsided.

The roll call of the builders

The passage history names a specific set of builders, and the series brief requires that members of both parties be credited by name and role. The record, drawn from the fact base of this article, is as follows. Ronald Reagan, the president, requested the Treasury study, issued the 1985 proposals, lobbied the House rule fight, and signed the act. Dan Rostenkowski, Democrat of Illinois, chaired Ways and Means, introduced H.R. 3838, managed the 26-day markup, and led the House conferees. Bob Packwood, Republican of Oregon, chaired the Finance Committee, adjourned the failed markup, co-authored the substitute, won the IRA floor fight, and led the Senate conferees.

Bill Bradley, Democrat of New Jersey, co-authored the Fair Tax Act with Gephardt, supplied the Democratic intellectual foundation for base broadening, and served as a Senate conferee. Dick Gephardt, Democrat of Missouri, co-authored the Fair Tax Act and served as a House conferee. Jack Kemp, Republican of New York, and Robert Kasten, Republican of Wisconsin, co-authored the FAST blueprint that gave reform its Republican pedigree. Bill Diefenderfer, the Finance Committee’s chief of staff and Packwood’s counsel, co-architected the substitute and translated the strategy into legislative text. James A. Baker III became Treasury Secretary in February 1985 and presided over the department during the congressional phase; Donald T. Regan signed the transmittal of Treasury I as the earlier secretary. Richard Darman, the Deputy Treasury Secretary, backed the Packwood plan. Russell B. Long, Democrat of Louisiana, served as ranking Democrat on Finance, partnered in the unanimous report, and served as a Senate conferee. Bob Dole, Republican of Kansas, served as Senate Majority Leader, managed the floor, and served as a conferee.

This roll call is the factual basis for the article’s neutrality: the contributions are distributed across the parties, the chambers, and the branches, and no single name explains the outcome.

The Counterfactuals: What Failure Would Have Looked Like

A procedural account is incomplete without the counterfactuals, the failures that each crisis nearly produced. Reconstructing what would have happened if each revival had failed clarifies what the revivals actually accomplished, and it shows how narrow the path to passage was at every stage. The counterfactuals are not speculation about personalities. They are deductions from the procedural facts, and each follows from the mechanism of the corresponding near-death.

If Reagan had not lobbied the rule defectors in December 1985, H.R. 3838 would have remained a reported bill without a rule. The Ways and Means markup, the twenty-six days of work, and H. Rept. 99-426 would have become legislative history without a legislative future, the common fate of reported bills that cannot reach the floor. The reform effort would not have died in any formal sense. It would have decayed, as the 99th Congress turned to other business and the coalition that Rostenkowski had assembled dispersed. A future Congress might have revived the idea, but it would have started from the blueprints rather than from a live vehicle, and the two years of executive and congressional investment would have been written off. The rule defeat was thus a potential killing not by verdict but by neglect, and Reagan’s lobbying mattered because it interrupted the neglect before it became permanent.

If Packwood had not produced the substitute after the April 18 suspension, the Finance Committee would have faced two options, both fatal. It could have resumed the incremental markup and reported a bill that missed its mark, which the Senate would have had no reason to pass and the House no reason to accept. Or it could have declined to report at all, leaving the House bill to die in committee. In either case, the reform would have ended in the spring of 1986, and the obituaries the pundits had already written would have been confirmed. The counterfactual is instructive because it shows that the substitute was not one option among several. It was the only option that satisfied the constraint, which means the April crisis had exactly one survivable outcome. Packwood’s decision looks like creativity in retrospect. In prospect, it was the recognition of a forced move.

If the IRA amendment had passed 51 to 48 instead of failing by that margin, the cascade described earlier would have followed. Each protected preference would have licensed the next demand, the base would have narrowed, the rates would have risen, and the bill reaching final passage would have been the incremental measure the substitute was designed to replace. Whether that measure could have passed at 97 to 3 is doubtful. The lopsided Senate vote was a vote for the radical bargain, for rates low enough to justify the pain. An incremental bill would have offered the pain without the reward, and the coalition for it would have been the narrow one that had already failed in committee. The most likely outcome is that the bill would have died on the floor, not by formal defeat but by the slow withdrawal of support as the rate structure deteriorated. The 51 to 48 vote was thus the moment at which the entire passage balanced on the smallest number of votes, and the counterfactual shows how much depended on three senators.

If the conference had deadlocked or either chamber had rejected the report, the bill would have died in September within sight of the finish. The conference could not have been reconvened easily, because the 99th Congress was approaching adjournment and the political calendar was closing. A rejected conference report would have sent the bill back to a process with no time left to run it. The counterfactual here is the least dramatic but the most final. The earlier deaths each left a procedural path to revival. A death in September would have left none, because the calendar itself would have been the killer. That the conference produced agreement on the rates, 15 and 28 percent for individuals and 34 percent for corporations, is therefore not a footnote. It is the closing of the last gate through which the bill had to pass.

Reading the Paper Trail

Legislative procedure leaves a paper trail, and the 1986 passage can be reconstructed from its documents alone. Three reports and three recorded votes carry the entire procedural history, and each document performed distinct work in moving the bill forward. The House report, H. Rept. 99-426, filed on December 7, 1985, was the product of the Ways and Means markup and the formal vehicle by which the committee’s text reached the floor. A House report does more than transmit text. It explains the committee’s intent, records the votes by which the text was approved, and provides the legislative history that courts and administrators will later consult. H. Rept. 99-426 thus froze the House’s version of the constraint bargain at a particular moment, with its particular set of surviving preferences, and made that version the baseline against which all subsequent changes would be measured.

The Senate report, S. Rept. 99-313, filed on May 29, 1986, performed the more dramatic documentary function. Because the Finance Committee reported its text as an amendment in the nature of a substitute, the Senate report transmitted not a revised House bill but a replacement, and the report number marks the procedural moment when Packwood’s radicalism became the official Senate position. The interval between the order to report on May 6 and the filing on May 29 represents the staff work of converting the substitute’s logic into legislative language, the unglamorous drafting without which the napkin sketch would have remained a sketch. The conference report, H. Rept. 99-841, filed in the House on September 18, 1986, closed the documentary sequence. A conference report is the only text both chambers vote on without amendment, an up-or-down proposition that prevents the reconciliation from unraveling on the floor. The three report numbers together tell the procedural story in compressed form: a House bill reported in December, a Senate substitute reported in May, a conference agreement reported in September.

The reports also performed a function that extends beyond the legislative process itself. A reported bill carries the committee’s explanation of its provisions, and that explanation becomes part of the legislative history that administrators consult when implementing the statute and that courts consult when interpreting it. H. Rept. 99-426, S. Rept. 99-313, and H. Rept. 99-841 thus did double duty. They moved the bill through Congress, and they documented what Congress thought it was doing for the benefit of everyone who would later have to apply the law. The precision of the reports mattered procedurally because imprecision would have created implementation disputes that could have unraveled the coalition retroactively. A member who voted for the bill on the understanding recorded in the report could not later claim to have been misled, and the reports’ specificity was part of what made the lopsided final votes sustainable. Procedure does not end at passage. It extends into the documentary record that governs what passage means.

The recorded votes are the complementary trail. Record Vote No. 148, the Senate’s 97 to 3 passage on June 24, 1986, with its 53 to 0 Republican and 44 to 3 Democratic breakdown, is the quantitative signature of the indivisibility strategy. Record Vote No. 413, the House’s 292 to 136 adoption of the conference report on September 25, 1986, with Republicans 116 to 62 and Democrats 176 to 74, shows the coalition holding through the final price settlement. Record Vote No. 296, the Senate’s 74 to 23 adoption on September 27, with Republicans 41 to 11 and Democrats 33 to 12, closes the sequence. Recorded votes matter procedurally because they fix responsibility. A voice vote, like the House’s December 17 passage, allows members to advance a bill without individual exposure. A recorded vote attaches each member’s name to a position, which is why the lopsided recorded tallies of June and September are stronger evidence of the coalition’s breadth than the voice vote of December. The paper trail thus confirms what the narrative asserts. Procedure did not merely accompany the bill. At every stage, a document or a roll call performed the work of moving it.

Recorded and Unrecorded Votes

The bill’s votes divide into two kinds, and the division is procedurally significant. The House passed H.R. 3838 on December 17, 1985, by voice vote, which means no member’s position was recorded and no tally was taken. The Senate passed it on June 24, 1986, by 97 to 3 as Record Vote No. 148. The conference report was adopted by 292 to 136 as Record Vote No. 413 in the House and 74 to 23 as Record Vote No. 296 in the Senate. The IRA amendment was defeated 51 to 48. Each of these voting methods did different procedural work, and the choice among them was itself a procedural decision.

The voice vote of December 17 is the most strategically interesting. A voice vote is normally used for uncontroversial matters, and its use for a major revenue bill was a deliberate choice to minimize exposure. In a voice vote, the presiding officer asks for the ayes and nays and judges which side is louder, without recording individual positions. For a bill whose coalition was fragile and whose preferences offended organized interests, this was protective. Members who had been lobbied intensely against the bill could allow it to pass without casting a vote their opponents could use against them. The voice vote thus functioned as a procedural anesthetic, moving the bill through the chamber while dulling the political pain of moving it. The contrast with the rule vote six days earlier is instructive. The rule required a recorded vote because rules are adopted by roll call as a matter of course, and the recorded 223 to 202 defeat exposed every member’s position. The bill itself, passed by voice, exposed no one’s. The House leadership understood that the bill could survive obscurity but not scrutiny, and chose the voting method accordingly.

The recorded votes of the later phases did the opposite work. By June, the substitute’s logic had made the bill defensible, and the managers wanted exposure rather than obscurity. Record Vote No. 148, with its 97 to 3 tally and its 53 to 0 Republican and 44 to 3 Democratic breakdown, was a public demonstration that the coalition was real and bipartisan. A voice vote in June would have suggested the managers were still hiding. The recorded vote announced that they were not. The same logic governed the conference report votes. Record Vote No. 413’s 292 to 136 and Record Vote No. 296’s 74 to 23, each with majorities of both parties voting yes, were designed to be quotable, to fix in the public record the breadth of the support. Recorded votes create accountability, and the managers sought accountability once the bill could withstand it.

The 53 to 0 Republican vote on final passage is the detail that completes the floor picture. Not a single Republican senator voted against the bill that a Republican president had demanded, a Republican chairman had rewritten, and a Republican majority leader had managed. The unanimity on the Republican side, set against the 44 to 3 Democratic division, shows where the bill’s core support lay and how completely the party had been brought along. Six months earlier, only fourteen House Republicans would vote for the rule. By June, fifty-three Senate Republicans voted for the bill itself. The transformation from fourteen to fifty-three measures what the substitute accomplished politically. It did not merely satisfy the arithmetic. It gave Republican members a bill they could vote for without reservation, and the party that had been the obstacle in December became the engine in June. Procedure had converted the skeptics, not by persuading them but by changing what they were being asked to support.

The 51 to 48 IRA amendment vote sits between these poles. It was recorded, because Senate amendment votes are, and its narrowness was therefore public knowledge. The public narrowness served a procedural purpose of its own. It warned every subsequent would-be amender that the managers would fight divisibility amendments to the last vote, and it warned every member that defecting on such amendments meant joining a losing cause by a handful of votes. A 51 to 48 win is more intimidating to future challengers than a comfortable win would have been, because it demonstrates both the managers’ willingness to fight and the coalition’s discipline under maximum pressure. The voting methods of the passage thus form a coherent procedural strategy. Obscurity when the bill was weak, exposure when it was strong, and a public display of discipline at the moment of maximum danger.

The numbering of the recorded votes is itself part of the procedural record. Record Vote No. 148 in the Senate, No. 413 in the House, and No. 296 in the Senate again are the chamber’s sequential tallies of roll-call votes in the 99th Congress, and their sequence locates each decision in the legislative calendar. No. 148 in June, No. 413 in late September in the House, No. 296 in late September in the Senate. The numbers are not analytically significant, but they are evidentially significant, because they allow any reader to retrieve the exact division, the exact question put, and the exact position of every member. Procedure leaves this kind of trace deliberately. The recorded vote is the legislature’s way of making its decisions legible to the future, and the three numbers that anchor this passage are the points at which the bill’s support was measured under conditions of maximum accountability. The voice vote of December, which left no such trace, is the exception that proves the rule. Where the managers wanted history to see the coalition, they recorded it. Where they wanted history to look away, they did not.

Presentment and Signature

The final procedural steps were swift. The conference report, H. Rept. 99-841, filed in the House on September 18, 1986, was adopted by the House on September 25 and by the Senate on September 27. The enrolled bill was presented to the President on October 20, 1986. Reagan signed it on October 22, 1986, as Public Law 99-514, recorded at 100 Stat. 2085. Presentment and signature are the least dramatic steps in the legislative process, and in this case their very uneventfulness is the point. A bill that had died three times, that had required a presidential rescue in the House and a radical substitution in the Senate, completed its final steps without incident, because by October the procedural battles were over and only the formalities remained.

The formalities themselves are worth describing, because they complete the procedural map. After both chambers adopt the conference report, the bill is enrolled, which means it is prepared in its final form, certified as having passed both chambers in identical text, and signed by the presiding officers. Enrollment is the process by which the legislative branch attests that the text is the text, and it is the precondition for presentment. Presentment is the delivery of the enrolled bill to the President, which occurred on October 20, 1986. The President then has ten days, Sundays excepted, to sign the bill, veto it, or allow it to become law without signature. Reagan signed on October 22, two days after presentment, which indicates that the decision required no deliberation. The signature converted H.R. 3838 from a bill into Public Law 99-514, and the Statutes at Large citation, 100 Stat. 2085, fixed its place in the permanent record of federal legislation.

The speed of the final steps is itself procedural evidence of the coalition’s solidity. A President who harbored doubts would have used the ten days. A Congress uncertain of its work would have delayed enrollment. Neither happened, because by late October the bill’s passage was the least controversial fact in Washington. The three deaths and three revivals had produced a text that no veto player had a reason to stop, and the enrollment, presentment, and signature moved at the pace of a process with nothing left to decide. The quiet end is the procedural reward for a noisy passage. Bills that die loudly in committee never reach enrollment. Bills that survive to enrollment have already won every fight that matters, and the signature is the clerk’s confirmation of a verdict the procedure rendered weeks before.

The Calendar as Procedure

The timeline of the passage repays attention as a procedural artifact in its own right. From introduction on December 3, 1985, to signature on October 22, 1986, the bill moved through four distinct phases, each with its own tempo, and the tempo of each phase reflected the procedural work being done. The House phase was compressed into fourteen days, from introduction on December 3 to passage on December 17, with the markup having already consumed the preceding eleven weeks. The compression was deliberate, as the same-day rule and passage votes show, and it reflected the leadership’s judgment that the coalition was perishable. The Senate committee phase stretched across nearly five months, from the receipt of the House bill to the 20 to 0 report on May 6 and the filing of S. Rept. 99-313 on May 29, with the April 18 suspension marking the midpoint crisis. The length of this phase reflected the difficulty of the work. Replacing an incremental bill with a radical substitute cannot be rushed, because the arithmetic must be verified and the coalition must be rebuilt around the new text.

The floor and conference phases moved at the tempo the earlier phases had earned. The Senate passed the bill on June 24, less than a month after receiving the committee report, because the substitute’s indivisibility made extended floor debate unnecessary and dangerous in equal measure. The conference filed on September 18 and secured adoption in both chambers within nine days, because the constraint had reduced the negotiable space to a single percentage point. The final month, from the Senate’s adoption on September 27 to presentment on October 20 and signature on October 22, was pure formality, the enrollment and presentment steps that convert a congressional agreement into a presidential act. The calendar thus traces the arc of the procedure. Slow where the text was being remade, fast where the coalition was being protected, and still at the end where only formalities remained. A bill that moved at a uniform pace through all four phases would have been a bill whose procedure was not responding to its substance. This bill’s varying tempo is evidence that the procedure was.

Procedure determined text, and the crisis improved it

The series thesis finds in this history a rare clean demonstration: procedure determined the text, and the procedural crisis improved the statute. Consider the counterfactual that the record invites. Had the Finance Committee’s incremental markup somehow staggered across the finish line in April 1986, closing the $29 billion gap with a final round of trades, the resulting bill would have been a modest measure: some preferences trimmed, rates cut slightly, every bargain visible in the text. It might have passed. It would not have been the most radical base-broadening measure in modern legislative history. The collapse of the markup, the second near-death, forced the substitution, and the substitution produced a bill that the incremental method could never have generated. The crisis did not merely delay the statute; it authored it.

This is why the heroic-personality version, though tempting, misleads. Packwood’s decisiveness, Diefenderfer’s drafting, Rostenkowski’s bargaining, Reagan’s lobbying, Bradley’s advocacy, Kemp’s agitation: all real, all documented, all necessary. None of them was sufficient, and the proof is the negative case. When later Congresses abandoned the design constraint, when revenue neutrality and distributional neutrality ceased to bind, the same kind of talented legislative personalities produced tax legislation of the familiar kind: targeted benefits, deficit finance, and incremental bargains. The procedural contrast with later efforts is the point, not any judgment about later policy. The personalities did not lose their skill. The constraint that had channeled their skill was gone, and without it the skill produced different results. The constraint did the work. The people did the work the constraint assigned them.

That is the durable procedural lesson, and it is stated without reference to later tax politics, about which this article draws no conclusion. A binding design constraint, publicly measurable and enforced at every stage, can convert a legislative process from a bazaar of individual bargains into a machine for producing coherent statutes. The 1986 act is the clearest case in which the machine worked: the rule fight tested whether the House wanted the bill, the markup collapse tested whether incrementalism could satisfy the constraint, the floor fight tested whether the coalition would hold, and the conference tested whether the two chambers could agree on the price. At each test the procedure, not the personalities, determined what survived.

What Procedure Determined

The 1986 act is the series’ clearest demonstration that procedure can determine text, and the rare case in which the determination improved the law. The House rule defeat forced the bill to find a presidential rescuer. The Finance markup collapse forced the bill to abandon incrementalism for the radical substitute. The IRA amendment fight forced the bill’s managers to defend the package as indivisible. At each point, the procedural crisis did not merely threaten the text. It selected a better one. Readers who want to trace each vote, each report number, and each near-death in their own notes can do so in the legislation study notebook.

The pattern generalizes beyond the particulars of 1986, and stating the generalization is the final work of this article. Legislative procedure is usually described as the machinery by which preferences are aggregated, a neutral conveyor belt carrying bills from introduction to signature. The 1986 passage shows procedure doing something more active. At each veto point, the procedure did not merely register the preferences of the participants. It altered the set of available options, eliminating the incremental bill in April, protecting the radical bill in June, and narrowing the conference to a single percentage point in September. The participants’ preferences were relatively constant throughout. What changed was what the procedure permitted them to choose. That is what it means to say that procedure determined the text. The text was not the output of a bargaining process that happened to use procedural forms. It was the output of the procedural forms themselves, operating on a constraint that left only one stable result.

The One Test, restated at the end: the bill pronounced dead three times became law by 97 to 3, 292 to 136, and 74 to 23 because a design constraint made radicalism the only survivable strategy. The procedural defeat that nearly ended it in the House was the December 11, 1985, rejection of H.Res. 336. The single strategic decision in a Senate committee was Packwood’s substitution of the preference-laden incremental bill with the wholesale base-broadening plan sketched over two pitchers of beer. Everything else, the lobbyists in the corridor, the unanimous committee, the voice vote, the conference arithmetic, was commentary on those facts. A reader who can reconstruct the sequence, name the defeat, and explain the decision has understood not only how the 1986 act passed but how legislative procedure, at its most consequential, actually works.

Frequently Asked Questions

Q: How did the Tax Reform Act pass Congress?

The bill, H.R. 3838, was introduced by Ways and Means Chairman Dan Rostenkowski on December 3, 1985, and reported as H. Rept. 99-426 after a twenty-six-day markup. The House rejected its rule on December 11, 223 to 202, then passed the rule and the bill by voice vote on December 17 after President Reagan personally lobbied defectors. In the Senate, Finance Chairman Bob Packwood suspended his markup on April 18, 1986, twenty-nine billion dollars off the revenue mark, then returned with a radical base-broadening substitute drafted with chief of staff Bill Diefenderfer. The committee reported it 20 to 0 on May 29 as S. Rept. 99-313. The Senate defeated a dangerous IRA amendment 51 to 48 and passed the bill 97 to 3 on June 24. A conference committee reconciled the versions, filing H. Rept. 99-841 on September 18; the House adopted it 292 to 136 on September 25 and the Senate 74 to 23 on September 27. Reagan signed it October 22 as Public Law 99-514.

Q: What was Gucci Gulch during the Tax Reform Act debate?

Gucci Gulch was the nickname for the hallway outside the Senate Finance Committee, where leading tax lobbyists plied their trade during the 1986 debate. The name referred to their expensive Italian shoes, a visible marker of the well-funded interests seeking to protect preferences in the bill. The term was popularized by the title of Jeffrey H. Birnbaum and Alan S. Murray’s 1987 book, Showdown at Gucci Gulch: Lawmakers, Lobbyists, and the Unlikely Triumph of Tax Reform, the canonical history of the passage. The corridor’s fame captures something real about the process: the lobbyists worked the members at the exact point where the text was being written. But the design constraint neutralized them structurally. Because every preference had to be funded by a higher rate, no carve-out could be won without visibly raising someone else’s rate, and the committee’s 20 to 0 vote showed how completely the arithmetic disarmed the hallway.

Q: Who were the main authors of the Tax Reform Act?

No single member authored the act; its authorship was distributed across the procedure. President Ronald Reagan initiated it, requesting the Treasury study in his January 1984 State of the Union and issuing Treasury II on May 29, 1985. Representative Dan Rostenkowski, Democrat of Illinois, introduced H.R. 3838 and chaired the Ways and Means markup as H. Rept. 99-426. Senator Bob Packwood, Republican of Oregon, authored the decisive Senate substitute with Finance chief of staff Bill Diefenderfer, reported 20 to 0 as S. Rept. 99-313. The intellectual blueprints came from Senator Bill Bradley and Representative Dick Gephardt, authors of the Fair Tax Act, and Representative Jack Kemp and Senator Robert Kasten, authors of the Fair and Simple Tax Act. Treasury Secretaries Donald T. Regan and James A. Baker III, Deputy Secretary Richard Darman, Majority Leader Bob Dole, and ranking Democrat Russell Long each performed indispensable roles.

Q: How did two pitchers of beer save the Tax Reform Act?

On April 18, 1986, Senate Finance Chairman Bob Packwood suspended his committee’s markup, announcing it was twenty-nine billion dollars off its revenue mark with no more sessions. The incremental bargaining had produced a bill nobody could defend. Packwood and his chief of staff, Bill Diefenderfer, then met at the Irish Times, a pub near the Capitol, where over two famous pitchers of beer they sketched on a napkin a substitute that abandoned incrementalism entirely. Instead of protecting preferences, the substitute eliminated them wholesale and used the recovered base to buy dramatically lower rates. The committee approved the substitute 20 to 0 and reported it May 29 as S. Rept. 99-313. The napkin is the most famous anecdote in modern legislative lore, but the procedure is what mattered: the substitute gave Packwood a new text that satisfied the constraint, which was the only thing that could restart the markup.

Q: How many times was the Tax Reform Act declared dead?

Three times in documented procedural fact, and countless times in commentary. The first death was December 11, 1985, when the House rejected the rule for H.R. 3838, H.Res. 336, by 223 to 202, stranding the bill without a path to the floor until President Reagan’s personal lobbying revived it. The second was April 18, 1986, when Chairman Packwood suspended the Senate Finance markup, twenty-nine billion dollars off its mark, and announced no more sessions, until the beer-napkin substitute gave him a text worth reconvening for. The third was the June 1986 Senate floor fight over IRA amendments, which Packwood won 51 to 48, testifying that losing a couple of votes would have lost the IRAs and the bill. Beyond these three, pundits pronounced the effort finished so often that Reagan told the Tax Reform Coalition it had been declared dead so many times by the pundits.

Q: What was the final Senate vote on the Tax Reform Act?

The Senate passed H.R. 3838 on June 24, 1986, by 97 to 3, recorded as Record Vote No. 148. Republicans voted 53 to 0 and Democrats 44 to 3. The lopsided tally is the quantitative signature of the substitute’s indivisibility strategy: senators who might have carved out individual preferences would not vote against the rate cuts, because the managers kept the chamber voting on the package as a whole. The Senate later adopted the conference report on September 27, 1986, by 74 to 23, Record Vote No. 296, with Republicans 41 to 11 and Democrats 33 to 12. Both votes were bipartisan, and both reflected the same procedural reality. By the summer of 1986 the bill had become the only vehicle capable of delivering the rate cuts both parties had promised, and voting against it meant voting against the achievement itself.

Q: Why did House Republicans vote down the Tax Reform Act rule?

On December 11, 1985, only fourteen Republicans voted for H.Res. 336, the rule to bring H.R. 3838 to the floor, and the rule failed 223 to 202. The Republicans voted no because the rule vote was a costless protest. Members could signal to organized interests that opposed the bill’s preference repeals that they had stood against the measure, without ever voting against its substance, since no substantive vote was on offer. The bill that emerged from the Ways and Means markup protected enough preferences to be fragile but repealed enough to alarm the interests, leaving its own party’s members exposed. Voting against the rule let Republicans please the lobbyists while avoiding a recorded position on reform itself. President Reagan had to lobby the defectors personally, including a rare trip to the Capitol, to change the cost structure of the vote before the rule passed on December 17.

Q: Was the Tax Reform Act bipartisan?

Yes, structurally as well as numerically. The presidential initiative came from Republican Ronald Reagan; the House vehicle from Democrat Dan Rostenkowski; the Senate substitute from Republican Bob Packwood; the blueprints from Democrats Bill Bradley and Dick Gephardt and Republicans Jack Kemp and Robert Kasten; the floor management from Republican Majority Leader Bob Dole; the Democratic cooperation from ranking member Russell Long. The votes confirm it: 97 to 3 in the Senate with Republicans 53 to 0 and Democrats 44 to 3; 292 to 136 in the House on the conference report with Republicans 116 to 62 and Democrats 176 to 74; 74 to 23 in the Senate with Republicans 41 to 11 and Democrats 33 to 12. Both parties claim credit, and both claims are correct, because at every stage the decisive actors came from both sides and neither party’s contribution was sufficient alone.

Q: What happened during the Senate floor fight over IRA amendments in June 1986?

The most dangerous moment of the Senate floor phase came over amendments affecting individual retirement accounts. The IRA provisions touched the retirement savings of millions of voters, and an amendment altering them threatened to demonstrate that the substitute’s package was divisible after all. Had the amendment passed, every other protected interest would have demanded the same treatment, the base would have narrowed amendment by amendment, the rates would have risen to compensate, and the bill would have collapsed back into the incremental failure of April. Chairman Packwood defeated the amendment 51 to 48, a margin of three votes, and testified afterward that had he lost a couple he would have lost on the IRAs and lost the bill. The vote held the line at the exact point where it was thinnest, and the bill’s managers kept the chamber voting on the package as a whole rather than on its pieces.

Q: Who served as conferees when the House and Senate conferenced the bill?

The House conferees were Rostenkowski, Pickle, Rangel, Stark, Gephardt, Russo, Pease, Duncan, Archer, Vander Jagt, and Crane, the leadership of the Ways and Means Committee with both parties represented. The Senate conferees were Packwood, Dole, Roth, Danforth, Chafee, Wallop, Long, Bentsen, Matsunaga, Bradley, and Moynihan, the leadership of the Finance Committee, also bipartisan. The rosters are procedurally significant. Both tax-writing committees were fully represented, both parties were fully represented, and the principal authors of the competing blueprints sat alongside the chairmen who had carried the bill: Gephardt and Bradley for the Fair Tax Act, Packwood and Rostenkowski for the vehicle, Dole as majority leader, Long as ranking Democrat. No faction could claim exclusion from the room where the final text was written, and the bipartisan rosters produced a report both parties had to own.

Q: Which report numbers tracked the bill through Congress?

Three reports carried the bill’s procedural history. H. Rept. 99-426 was the Ways and Means Committee’s report, filed December 7, 1985, after the twenty-six-day markup, transmitting the House’s version of the constraint bargain with its particular set of surviving preferences. S. Rept. 99-313 was the Senate Finance Committee’s report, filed May 29, 1986, transmitting Packwood’s radical substitute as an amendment in the nature of a substitute; the report number marks the moment the substitute became the official Senate position. H. Rept. 99-841 was the conference report, filed in the House on September 18, 1986, the unamendable text both chambers voted up or down. Together the three numbers compress the passage: a House bill reported in December, a Senate substitute reported in May, a conference agreement reported in September.

Q: What roles did Donald Regan, James Baker, and Richard Darman play?

Donald T. Regan was Treasury Secretary when Treasury I was produced; his transmittal signed the November 1984 report, Tax Reform for Fairness, Simplicity, and Economic Growth, that launched the administration’s effort pursuant to Reagan’s January 1984 State of the Union request. James A. Baker III succeeded Regan at Treasury in February 1985 and managed the administration’s legislative strategy through the House phase, including the issuance of Treasury II, the President’s Tax Proposals to the Congress for Fairness, Growth, and Simplicity, on May 29, 1985. Richard Darman, the Deputy Treasury Secretary, served as an internal advocate for the reform and, in the account of Birnbaum and Murray, backed the Packwood substitute in the Senate. The three men illustrate the complication this article develops: different temperaments and positions, all serving the same constraint, which is why the personnel change of February 1985 did not disrupt the effort.

Q: Why did the House pass the bill by voice vote on December 17, 1985?

The House passed H.R. 3838 by voice vote on December 17, 1985, the same day it passed the rule, as a deliberate procedural choice to minimize exposure. In a voice vote the presiding officer judges the ayes and nays without recording individual positions, so no member’s stance becomes a target. For a bill whose coalition was fragile and whose preference repeals offended organized interests in both parties, this was protective. Members who had needed presidential pressure to support the rule could allow the bill to advance without casting a vote their opponents could use against them. The voice vote functioned as a procedural anesthetic, moving the bill with minimum friction. The contrast with the recorded rule vote six days earlier is the point: the House exposed members on the procedural motion but shielded them on the substance, because the bill could survive obscurity but not scrutiny.

Q: What did Bob Dole do as Senate Majority Leader during the reform fight?

As Senate Majority Leader, Bob Dole, Republican of Kansas, controlled the scheduling of floor business, the allocation of debate time, and the sequencing of amendments. Those powers were the instruments of the indivisibility strategy: by structuring floor consideration so the package was voted on as a whole and divisibility amendments were defeated in sequence, most critically the IRA amendment at 51 to 48, the leadership ensured no single amendment could unravel the substitute. Dole also served as a Senate conferee on the final bill, extending his role from scheduling into the reconciliation of the two chambers’ versions. The majority leader is often described as a traffic manager, but here the traffic management was the substance. The order in which the Senate voted determined what it could vote for, and the 97 to 3 result was manufactured as much by the schedule as by persuasion.

Q: What was Russell Long’s part as the Finance Committee’s ranking Democrat?

Russell B. Long, Democrat of Louisiana, was the ranking Democrat on the Senate Finance Committee during the 1986 markup, which made him the senior minority member and his party’s institutional representative on the tax-writing panel. In that role he helped hold the Democratic side of the committee’s unanimous 20 to 0 vote for the Packwood substitute, a bipartisan result that would have been impossible without the ranking member’s cooperation. Long also served as a Senate conferee on the final bill, carrying the Democratic position into the reconciliation with the House. His presence illustrates the article’s structural bipartisanship: the procedure required both parties’ institutional positions at every stage, and the ranking member’s office was the Democrats’ position on Finance. The substitute needed Democratic votes to be credible as a national reform rather than a partisan project, and Long’s role was to deliver them.

Q: How did the bill move from a rejected rule to passage in six days?

On December 11, 1985, the House rejected H.Res. 336, the rule for H.R. 3838, by 223 to 202, with only fourteen Republicans in favor. The bill was stranded without a path to the floor. President Reagan then lobbied the defectors personally, making a rare trip to the Capitol, which changed the cost structure of the rule vote: opposing the rule now meant defying a president of one’s own party who would remember the answer. On December 17 the House passed the rule, and on the same day it passed the bill itself by voice vote. The six-day turnaround illustrates a general principle: rules are substantive votes in procedural dress, and they can be moved by the same pressure that moves substantive votes. Once the rule was secured, the leadership moved the bill the same day, before the coalition could decay, using speed and the voiceless vote to minimize friction.

Q: What is the public law number and Statutes at Large citation of the act?

The act is Public Law 99-514 of the 99th Congress, recorded at 100 Stat. 2085 in the United States Statutes at Large. The designations encode the bill’s procedural history. The number 99-514 means the 514th public law enacted by the 99th Congress; the citation 100 Stat. 2085 locates the statute at page 2085 of volume 100 of the Statutes at Large, the permanent official record of federal legislation. The enrolled bill was presented to President Reagan on October 20, 1986, and he signed it on October 22, 1986, two days later, indicating the decision required no deliberation. The signature converted H.R. 3838 from a bill into a public law, and the citations fix its place in the permanent record. A reader who wants to retrieve the statute as enacted looks for 100 Stat. 2085; a reader tracking its congressional history looks for Public Law 99-514.

Q: How did the Bradley-Gephardt and Kemp-Kasten proposals shape the final statute?

The two congressional blueprints gave the reform its legislative pedigree before the administration acted. The Fair Tax Act of Bradley, Democrat of New Jersey, and Gephardt, Democrat of Missouri, first introduced in 1982 and reintroduced in 1983 as S.1421, proved that base broadening for rate reduction could attract a bipartisan constituency and kept the idea alive across two Congresses. The Fair and Simple Tax Act of Kemp, Republican of New York, and Kasten, Republican of Wisconsin, introduced as H.R. 6165 and S. 2948 in the 98th Congress and H.R. 777 and S. 325 in the 99th, gave Republican members a reform vehicle of their own and prevented the effort from being typecast as Democratic redistribution. By the time Rostenkowski introduced H.R. 3838, every faction had invested in some version of the trade, which gave every faction a reason to keep the process alive. Both Bradley and Gephardt served as conferees on the final bill.

Q: How did revenue neutrality shape the Tax Reform Act?

Revenue neutrality required the bill to raise the same revenue as the law it replaced, so every rate cut had to be matched by base broadening that raised an equal amount. The matching was scored by staff, published, and debated publicly, leaving no room to hide shortfalls. This discipline transformed bargaining: members could not buy votes with unpaid-for benefits, since each concession needed an offset with a visible loser. The Finance Committee’s incremental markup failed because its trades accumulated into a $29 billion shortfall. The Packwood substitute succeeded because wholesale preference elimination raised revenue on a scale piecemeal bargaining could not, financing the 15 and 28 percent individual brackets and the 34 percent corporate rate. Distributional neutrality added a second lock, barring burden shifts across income classes that the published distribution tables would have exposed. Together the neutralities closed off the easy paths and left only genuine base broadening, which is why the constraint produced radicalism and why the procedural crisis improved the statute.

Q: Why did H.R. 3838 need a House rule to reach the floor?

Major legislation reaches the House floor under a special rule from the Rules Committee that sets the terms of debate: its length, which amendments are in order, and whether members may rewrite the bill or must take it as written. For a tax bill, the rule is typically restrictive, protecting the committee’s balanced package from being picked apart by floor amendments. House Resolution 336 was the rule for H.R. 3838. When the House rejected it 223 to 202 on December 11, 1985, the bill had no path to floor debate under controlled conditions. Voting against the rule is the classic way to kill a bill without opposing its substance, since the vote is procedural and deniable but its effect is substantive. Only 14 Republicans supported the rule, showing how the Ways and Means bargains had alienated the floor. After President Reagan personally lobbied defectors, the rule passed on December 17, 1985, and the bill passed by voice vote the same day. The episode shows the rule as the procedural key to the floor.