The story of the 1986 tax reform is usually told as a story about rates. The top individual rate fell from half of every marginal dollar to a little more than a quarter, the corporate rate fell from 46 percent to 34 percent, and a fourteen-bracket statutory schedule collapsed into two clean lines. That telling is not wrong, but it misleads, because the part of the law that filled the headlines was the part that disappeared fastest. The two-rate schedule that defined the act in the public mind survived less than five years in its enacted form, and within seven years Congress had restored a top rate higher than the one the reform had replaced. What endured was quieter and, in the judgment of the research literature, far more consequential: a wider base, a new set of limits on the use of paper losses, and a natural experiment that public finance scholars have been mining ever since.

WHAT THE 1986 TAX REFORM MEASURED UP AGAINST

Every assessment of the act has to begin with what its authors said it was trying to do, because a statute can only be judged against its own aims if those aims are stated plainly. The reformers of 1986 set themselves three objectives. The first was fairness, meaning that taxpayers in similar economic circumstances should face similar burdens, which in practice meant attacking the shelters, preferences, and exclusions that let high-income filers pay effective rates far below the statutory schedule. The second was efficiency, meaning that the code should distort economic decisions as little as possible, which in practice meant lowering marginal rates while removing the preferences whose removal paid for the cuts. The third was simplicity, meaning a shorter, plainer statute that ordinary filers could understand and the revenue service could administer. Around these three aims sat a hard fiscal constraint. The package was designed and scored by the Joint Committee on Taxation as approximately revenue-neutral over the five-year budget window, so every rate reduction had to be matched by base broadening or preference repeal, and no provision could ride free.

That constraint shaped the entire architecture of the law. The famous two-rate individual schedule, 15 percent and 28 percent, applied to tax years beginning in 1988, with a transitional five-bracket schedule running from 11 percent to 38.5 percent covering tax year 1987. The 60 percent exclusion for long-term capital gains was repealed, so gains were taxed as ordinary income subject to a 28 percent maximum, and a 5 percent surcharge phased in over a middle income range created an effective 33 percent bubble rate that the drafters accepted as the price of holding the two-rate shape. The corporate rate dropped to 34 percent, leaving the top individual rate below the corporate rate for the first time in the postwar period, a relationship that would turn out to matter more than almost anyone anticipated. The standard deduction rose sharply, to 5,000 dollars for joint filers in 1988, and the personal exemption climbed from 1,080 dollars to 1,900 dollars for 1987 on its way to 2,000 dollars by 1989. The earned income credit grew from 11 percent to 14 percent of the first 5,714 dollars of earnings. The individual alternative minimum tax was widened and its rate raised. Secondary accounts drawing on Congressional Research Service figures put the number of low-income filers removed from the rolls near six million, a figure that belongs to the fairness objective rather than to the rate story.

The provisions that paid for the rate cuts were the ones that would last. Section 469 imposed the passive activity loss rules, limiting the use of losses from activities in which the taxpayer did not materially participate. The investment tax credit was repealed outright. Depreciation schedules were lengthened, the consumer interest deduction was phased out, the state and local sales tax deduction was eliminated while the income tax deduction survived, and a long list of smaller preferences fell away. This is the distinction that organizes everything that follows. The rate schedule was the headline. The base broadening was the substance. One vanished within a decade. The other, in its essentials, never left.

The three aims deserve a closer look, because the reformers meant specific things by them and the specificity matters for the scorecard. Fairness meant horizontal equity, the principle that taxpayers with the same economic income should pay the same tax. The pre-1986 code violated that principle flagrantly. Two households with identical salaries could face radically different tax bills depending on their appetite for shelters, and the difference had nothing to do with ability to pay or with any policy purpose. The reformers’ fairness agenda was therefore not primarily about redistribution between rich and poor. It was about ending the arbitrary variation in burdens among taxpayers at the same income level, variation created by differential access to tax planning.

Efficiency meant neutrality, the principle that the tax system should influence economic decisions as little as possible. Every preference in the code was a thumb on the scale, steering investment toward the favored activity and away from alternatives with higher pre-tax returns. The efficiency cost of those thumbs was the lost output from misallocated capital, and the reformers believed, with good reason, that the cost was large. Lowering marginal rates while removing preferences served efficiency twice over. Lower rates reduced the penalty on earning additional income, and fewer preferences reduced the distortions in how that income was earned and invested. The two halves of the design were not separate ambitions. They were the same ambition expressed in two provisions.

Simplicity was the aim the reformers stated least precisely and achieved least fully. In the rhetoric of 1985 and 1986, simplicity meant a code ordinary taxpayers could understand without professional help, with fewer forms, fewer worksheets, and fewer traps. The fourteen-to-two rate compression was simplicity’s showcase. But the reformers never developed a theory of what made a tax system complex, and the omission showed. They treated complexity as a function of the number of provisions, when it was at least as much a function of the difficulty of the remaining ones. The passive loss rules were a single provision, but they were harder to apply than the dozen preferences they replaced. The lesson, which Slemrod’s 1992 assessment drove home, was that simplification by subtraction works only when the subtractions do not require complex anti-abuse machinery to enforce, and the 1986 reform’s subtractions required a great deal of it.

The three aims also traded off against one another, and the tradeoffs shaped the final law. Fairness and efficiency pointed in the same direction, toward base broadening, but simplicity sometimes pointed the other way, toward the intricate rules needed to police the broadened base. Revenue neutrality constrained all three, forcing the rate cuts to be smaller than their advocates wanted and the base broadening to be larger than its opponents could bear. Understanding the reform as a negotiated settlement among these aims, rather than as the triumph of any one of them, is the precondition for judging it on the evidence. The sections that follow apply that judgment aim by aim, provision by provision, following the evidence where it leads rather than the rhetoric where it points.

DID THE 1986 ACT PAY FOR ITSELF?

Did the 1986 act pay for itself?

No. The act was designed and scored by the Joint Committee on Taxation as approximately revenue-neutral over the five-year budget window, meaning the rate cuts were matched by offsetting base broadening within the scoring period. It was never designed to raise revenue through faster growth, and the research finds no significant growth effect attributable to the statute.

That question matters because it is the one the reform is most often misremembered as having answered. The legislative bargain was explicit. Lower rates would be purchased with a broader base, and the purchase price would be paid in full within the budget window. The Joint Committee’s revenue estimates treated the two sides of the ledger as equal, and the political coalition held together only because neither side could be accused of adding to the deficit. Whatever growth the act produced, and the evidence on that point is examined below, was never part of the financing plan. When later commentators credit the reform with revenues it did not collect or blame it for deficits it did not create, they are measuring the statute against a promise it never made.

The deeper point is that revenue neutrality was itself an achievement with consequences. Because the rate cuts had to be paid for, the coalition that wanted lower rates became the coalition that had to abolish the preferences, and that alignment is what made the base broadening politically possible. A rate cut financed by borrowing would have required no sacrifice from anyone. A rate cut financed by repeal required the shelter industry, the real estate syndication business, and a long parade of preference holders to surrender the provisions that had made their business models work. That is why the revenue-neutral design belongs in any account of what the act accomplished. It was the mechanism by which the ambition of lower rates was converted into the reality of a wider base.

THE REVENUE-NEUTRALITY BARGAIN

The design constraint that made the 1986 act possible is also the constraint that explains its shape. The package was built to be approximately revenue-neutral over the five-year budget window as scored by the Joint Committee on Taxation, which meant that every dollar of rate reduction had to be matched by a dollar of base broadening or preference repeal inside that window. That sounds like a technicality. In practice it was the political keystone of the entire enterprise, because it converted the ambition of lower rates into the machinery of preference repeal.

The scoring process deserves a brief explanation, since the estimates that held the coalition together were products of a specific method. The Joint Committee estimated the revenue effect of each provision by comparing projected receipts under the proposed law with projected receipts under then-current law, holding taxpayer behavior largely fixed for the mechanical calculation and applying standard assumptions about the timing of income and deductions. Those estimates were not forecasts of what would happen in practice. They were disciplined comparisons, and their discipline was the point. When the committee reported that the rate cuts and the base broadeners offset each other over five years, it gave every member of the coalition a number to cite against the charge of fiscal irresponsibility. The coalition could claim, with the committee’s imprimatur, that the reform would not add to the deficit.

That claim was load-bearing in a way that is hard to reconstruct from a distance. The mid-1980s were years of large federal deficits, and any major tax bill that appeared to enlarge them would have faced an uphill fight in both chambers. Revenue neutrality neutralized the deficit objection before it could be raised. It also neutralized a second objection, subtler but equally dangerous: the suspicion that rate cuts were a stalking horse for revenue loss that would later be recouped from someone else. By binding the rate cuts to offsetting base broadeners in the same bill, scored in the same window, the design made the distributional bargain transparent. The people who got the lower rates were, in large measure, the people who lost the preferences.

Why did revenue neutrality make the base broadening possible?

Because it forced rate-cut advocates to become preference repealers. A deficit-financed rate cut requires no sacrifice from anyone. A revenue-neutral rate cut requires its supporters to abolish the preferences that pay for it, which turns the natural enemies of base broadening into its legislative foot soldiers.

The five-year window introduced its own distortions, and an honest account of the reform must acknowledge them. Provisions could be timed to concentrate their revenue gains inside the window and their revenue losses outside it, and phase-ins could push costs beyond the scoring horizon. The passive loss rules, with their five-year glide path for pre-enactment interests, illustrate the technique: the full revenue gain from the provision arrived gradually, which meant the early-year scores understated the long-run yield. None of this violated the scoring conventions. But it meant that revenue neutrality over five years was a different promise than revenue neutrality over twenty, and the difference would matter when later Congresses inherited the base the reform had broadened and the rates it had cut.

The scoring conventions behind the neutrality claim deserve explanation, because they determine what the claim can and cannot prove. The Joint Committee on Taxation scored the 1986 act under the static conventions then standard: the estimate counted the mechanical revenue effects of the rate and base changes against a fixed economic baseline, without adding a feedback loop for any growth the reform might induce. This was not a statement that growth effects were zero. It was a methodological choice to keep the score verifiable, and it meant the neutrality verdict described the legislation’s arithmetic rather than its economic consequences. The five-year budget window compounded the convention’s importance. Provisions that phased in slowly, like the passive loss disallowance reaching full effect in 1991, contributed less to the scored total than provisions effective immediately, which gave drafters an incentive to back-load the base broadeners and front-load the rate cuts. The window was a political artifact, but it shaped the statute’s architecture.

The later debate over dynamic scoring, the practice of adding estimated growth feedback to official scores, grew partly out of dissatisfaction with this convention, and the 1986 literature supplied ammunition to both sides. Advocates of dynamic scoring cited the large early elasticity estimates as evidence that static scores understated the revenue feedback from rate cuts. Defenders of the static convention cited the timing contamination and the later consensus range as evidence that the feedback was smaller and less certain than the early numbers suggested. The 1986 episode thus became the empirical battleground for a methodological dispute that outlived the statute’s own rate schedule, another instance of the research outliving the policy. What the scoring history establishes for this article’s purposes is narrower: the revenue neutrality of the 1986 act was a scored property under stated conventions, not a discovered fact about the economy, and it should be cited with the conventions attached.

The deeper significance of the bargain is the one the rate reversals of 1990 and 1993 exposed. Revenue neutrality bound the Congress that enacted the reform. It did not bind any subsequent Congress. A later legislature was free to keep the broadened base and raise the rates, pocketing the revenue the original bargain had used to finance the cuts, and that is precisely what the two reconciliation acts did. The base broadening survived because no coalition formed to repeal it. The rate cuts did not survive because a coalition formed to reverse them. The asymmetry was not an accident of the design. It was the predictable consequence of financing visible, politically exposed rate cuts with invisible, politically protected base broadeners, under a neutrality constraint that expired with the Congress that imposed it.

WHAT SURVIVED: THE END OF THE MASS-MARKET SHELTER

If the reform is remembered for cutting rates, it should be remembered for what it destroyed. The single most durable achievement of the 1986 act was the effective end of the mass-market tax shelter industry, and that achievement has never been reversed. Section 469, the passive activity loss provision, drew a line the code had never drawn before. Losses from activities in which the taxpayer did not materially participate could no longer be used to offset wage, salary, or portfolio income. The provision struck at the heart of the shelter business model, which had been built on selling high-income professionals passive losses they could deduct against their salaries while putting little or no economic capital at risk.

To understand the scale of what changed, it helps to recall what the shelter market looked like before the reform. Through the early 1980s, the combination of high marginal rates, accelerated depreciation, the investment tax credit, and generous loss rules had made tax-motivated investing into a retail industry. Partnerships were marketed not on their economic merits but on their first-year write-offs, with prospectuses advertising the multiple of each invested dollar that would return as a tax deduction. Cattle feeding, oil and gas drilling funds, equipment leasing, and above all real estate syndications were packaged for doctors, lawyers, and executives who would never visit the underlying asset. The losses were real on paper and fictitious in substance, and the revenue loss ran into the billions. Congress had tried narrower fixes before, including the at-risk rules of the late 1970s, but the shelter promoters had routed around every one of them.

The leverage at the heart of the shelter deserves emphasis, because it explains why the industry was so large and why section 469 killed it so completely. The typical shelter partnership borrowed most of the purchase price, which meant the investor’s deductions were computed on the full cost of the asset while the investor’s cash at risk was a fraction of that cost. Depreciation on the full value, interest on the full borrowing, and in equipment deals the investment credit on the full price, all flowed through to partners whose actual economic exposure was limited to their equity contribution. The earlier at-risk rules had attempted to limit loss claims to the amount the investor genuinely stood to lose, but promoters engineered around them with nonrecourse structures and guarantees that satisfied the letter while defeating the purpose. Section 469 succeeded where the earlier rules failed because it attacked the use of the loss rather than its measurement. It did not matter how the loss was computed or how much was at risk; the loss could not leave the passive basket. No restructuring of the financing could change that result, which is why the industry collapsed rather than adapted.

Section 469 ended the routing. By suspending passive losses and carrying them forward against future passive income only, the provision removed the economic engine that had driven the retail shelter market. A taxpayer could still lose money on a passive investment, but the loss could no longer shelter unrelated income, which meant the investment had to make sense on its own economic terms. The investment tax credit repeal and the lengthened depreciation schedules compounded the effect, but the passive loss rules were the structural change. The shelter industry did not adapt to them. It collapsed. Within a few years the market for retail tax shelters, the limited partnership offerings marketed in newspaper advertisements and broker pitches, had effectively vanished, and no subsequent Congress has restored it.

What made the passive loss rules so difficult to reverse?

The rules attacked the mechanism rather than any single investment, so no industry coalition could defend a specific preference. The phase-in gave existing investors time to adjust without creating a lobby for repeal, and the provision generated revenue every year, making any repeal bill carry a visible cost.

The phase-in deserves attention because it explains how a provision this severe became law at all. Congress did not apply the new limits to pre-enactment interests all at once. Instead it allowed a declining share of passive losses from existing investments to offset other income: 35 percent in 1987, 60 percent in 1988, 80 percent in 1989, 90 percent in 1990, and the full limitation from 1991 onward. That glide path defused the transition-equity objection, the argument that investors who had acted under old rules deserved old treatment, without surrendering the principle. It also meant the provision’s full bite arrived gradually, which may explain why the political resistance never organized. By the time the rules applied in full, the shelter industry that would have fought them had already been dismantled by the anticipation of their arrival.

Two qualifications to the survival story need to be stated, because the brief for this article requires precision about what was weakened as well as what endured. First, the original act itself contained a 25,000 dollar offset for rental real estate, section 469(i), available to taxpayers with adjusted gross income below 100,000 dollars and phased out entirely at 150,000 dollars. That offset was part of the enacted compromise, not a later weakening, and it remains in the code. Second, the Omnibus Budget Reconciliation Act of 1993 added the real estate professional exception, section 469(c)(7), which treats rental real estate losses as nonpassive for taxpayers who spend more than 750 hours a year and more than half of their personal services in real property trades or businesses. That exception was a genuine narrowing, and it was aimed at a specific constituency. But it left the core architecture intact. Section 469 has never been repealed, no statute has restored the pre-1986 treatment of passive losses, and the retail shelter industry has never returned.

The stranded investments of the phase-in period created a secondary market that illustrated the rules’ bite. Partnership interests acquired before the act, carrying suspended passive losses that could be released only against passive income or on disposition of the entire interest, traded at deep discounts to their stated asset values, because the tax benefits that had justified the original prices were evaporating on the statutory schedule. Buyers of these interests were purchasing the underlying real estate or equipment with the tax attributes largely stripped out, which repriced an entire asset class toward its economic value. The disposition rule added a lock-in of its own: holders of losing passive interests had an incentive to sell to release the suspended losses, while holders of winning interests had no such urgency, so the market for passive interests thinned unevenly. The repricing was the market’s verdict on the shelter era. Assets bought for their tax losses were worth less without them, and the difference measured the subsidy’s size.

The aftermath reorganized the tax-advisory business around the new rules. Where promoters had once sold partnerships as tax products, advisers now sold compliance: classifying activities, tracking suspended losses across years, applying the phase-in percentages to pre-enactment interests, and testing clients against the material-participation standards. The real estate professional exception added by the 1993 act created a further specialty, documenting the 750-hour test for clients who claimed it. This was a smaller, duller, and more honest industry than the one it replaced, and its existence is itself evidence of the structural change. An anti-abuse rule that generates a compliance practice has reshaped behavior. A rate cut that generates no practice has altered a number without altering behavior.

The ledger for this provision is therefore the cleanest in the entire act. Change made: passive losses suspended against nonpassive income. Still operative: yes, with the two qualifications noted. Statute that modified it: none for the core rule; the 1993 reconciliation act added the professional exception. Measured effect: the collapse of the mass-market shelter industry, an outcome visible in the disappearance of retail shelter offerings and in the swing of reported partnership and S corporation net income from negative to positive territory that researchers documented in the years after enactment.

THE ANATOMY OF THE BASE BROADENING

The base broadening that paid for the rate cuts was not a single provision. It was a long list of repeals, curtailments, and tightenings, each aimed at a preference that had accumulated its own constituency over the decades. Walking through the list is necessary, because the durability of the reform lives in these provisions rather than in the rate schedule, and because the pattern of what was repealed and what was spared reveals the political logic of the entire package.

The investment tax credit was the most prominent casualty. The credit had allowed businesses to subtract a percentage of the cost of new equipment directly from their tax liability, and it had been a centerpiece of investment incentives since the early 1960s. Its repeal removed one of the two pillars of the shelter-era equipment leasing business, the other being accelerated depreciation. The credit was not phased out or narrowed. It was abolished, and no Congress has reenacted it in its pre-1986 form. The repeal illustrated the reform’s method at its most uncompromising: a preference with an organized business constituency was eliminated outright because the revenue-neutral design required the money and the coalition held.

Depreciation was the second pillar. The reform lengthened the recovery periods for both equipment and structures, slowing the rate at which businesses could write off capital investments. The change reduced the present value of depreciation allowances and with them the tax motivation for churning assets through successive owners to harvest fresh depreciation schedules. Subsequent legislation adjusted the lives of specific asset classes over the years, but the direction of the 1986 change was never reversed. The code never returned to the accelerated schedules that had made depreciation harvesting a shelter staple.

The consumer interest deduction was phased out, ending the deductibility of interest on personal borrowing such as credit card debt and auto loans. The provision had been one of the code’s quieter preferences, claimed by millions of filers without much political organization behind it, which may explain why its repeal provoked less resistance than its revenue yield might have suggested. Business interest and investment interest survived in curtailed form, but the era of deducting the interest on a family car came to an end.

The treatment of state and local taxes drew one of the reform’s most politically charged lines. The deduction for state and local sales taxes was repealed. The deduction for state and local income taxes was retained. The distinction had no deep economic logic; a dollar of sales tax and a dollar of income tax burden a household in much the same way. It had a political logic. The income tax deduction was defended by high-tax states with powerful congressional delegations, while the sales tax deduction, claimed disproportionately in states without income taxes, had fewer champions in the rooms where the bill was written. The line the reform drew, repealing one and keeping the other, survived. The sales tax deduction was never restored as an itemized deduction in its pre-1986 form.

The repeal of the 60 percent capital gains exclusion belonged to the same base-broadening drive, though its fate diverged from the others. Taxing gains as ordinary income with a 28 percent maximum was one of the reform’s boldest strokes, eliminating a preference that had defined investment taxation for decades. But the exclusion’s repeal proved less durable than the other base broadeners, and the 1990 and 1993 acts restored a substantial preference, as the rate section above described. The episode illustrates a general point about the ledger. Provisions that were visible and valuable to organized constituencies faced repeal pressure. Provisions that operated invisibly, through loss limitations and timing rules, did not.

The individual alternative minimum tax was expanded to catch more of the preference income the regular tax no longer sheltered. The AMT had been created to ensure that high-income filers could not combine preferences to eliminate their liability entirely, but the proliferation of preferences had outrun it. The reform widened the AMT base and raised its reach, and the 1990 reconciliation act later raised the individual AMT rate from 21 percent to 24 percent rather than narrowing the base. The expanded minimum tax persisted as a parallel system, and its persistence is part of why the base broadening endured: even where the regular tax left openings, the minimum tax narrowed them.

Equally revealing as what the reform repealed is what it deliberately left alone. The home mortgage interest deduction survived, despite its cost and its concentration among higher-income itemizers, because homeowners were too numerous and too organized a constituency to confront. The deduction for state and local income taxes survived for the same reason. The exclusion for employer-provided benefits and the retirement saving provisions were left largely intact. These survivals were not oversights. They were the boundaries of the politically possible, and they marked the places where the coalition’s nerve failed. A complete account of the base broadening must include them, because they show that the reform was a negotiated settlement rather than a triumph of principle, and because every preference that survived became a precedent for the preferences that later Congresses would add.

At the bottom of the income scale, the reform narrowed the base in the other direction, removing income from tax rather than adding it. The earned income credit expansion, from 11 percent to 14 percent of the first 5,714 dollars of earnings, increased the credit’s antipoverty reach. The personal exemption rose from 1,080 dollars in 1986 to 1,900 dollars in 1987, 1,950 dollars in 1988, and 2,000 dollars in 1989. The standard deduction rose to 5,000 dollars for joint filers in 1988, with 3,000 dollars for single filers, 4,400 dollars for heads of household, and 2,500 dollars for married filers filing separately. Together these changes lifted millions of low-income households off the rolls entirely; secondary accounts drawing on Congressional Research Service figures put the number near six million. That narrowing was part of the fairness objective, and it was financed the same way the rate cuts were, from the base broadening above. The ledger of the reform therefore runs in both directions: broader at the top, narrower at the bottom, with the middle paying for neither and benefiting from both.

THE CAPITAL GAINS BARGAIN

Under the pre-1986 code, an individual excluded 60 percent of long-term capital gains from income and paid tax on the remaining 40 percent at ordinary rates, which produced an effective top rate of 20 percent when the top ordinary bracket was 50 percent. The exclusion was the single largest preference in the individual income tax, and it shaped investment behavior across the economy: it rewarded the realization of gains, penalized ordinary income relative to appreciation, and supplied the raw material for a family of conversion shelters that transformed wages and interest into gains. The 1986 act repealed the exclusion outright and taxed long-term gains as ordinary income, subject to a 28 percent maximum. The maximum rate on gains therefore rose from 20 to 28 percent at the same moment the top ordinary rate fell from 50 to 28 percent, compressing the two into a single schedule.

The bargain embedded in this change was explicit and, in retrospect, fragile. Investors who held appreciated assets accepted a higher tax on gains in exchange for lower taxes on everything else, and the 28 percent cap preserved a measure of the old preference for those at the top. But the bargain depended on the rest of the package holding. When the 1990 act raised the top ordinary rate to 31 percent while capping gains at 28 percent, it reopened a differential between gains and ordinary income that the 1986 act had closed, and the 1993 act widened the gap further by taking the ordinary top to 39.6 percent while leaving the gains cap in place. The exclusion itself was never restored, which is why the ledger marks this row as surviving in substance, but the rate relationship the 1986 act had established lasted only until the next deficit package. The episode demonstrates the general rule in a pure form. Repealing a preference is structural and durable; the rate applied to the newly broadened base is parametric and negotiable. Congress kept the broader base and renegotiated the price, and the investors who had accepted the 1986 bargain found themselves paying 1986-level taxes on gains alongside 1993-level taxes on everything else.

The behavioral research on gains deserves a separate note because it fed directly into the elasticity debate. Realizations of capital gains are among the most timing-sensitive margins in the tax system: investors can choose when to sell, and the 1986-1987 boundary produced one of the largest recorded shifts, with realizations accelerated into 1986 ahead of the higher 1987 treatment. Studies of the episode consistently found large short-run responses and much smaller permanent ones, the same pattern Goolsbee documented for executive compensation and Saez documented for top incomes generally. The gains evidence thus reinforced the literature’s central caution rather than qualifying it. Where the response is easiest to measure, it is also most contaminated by timing, and the cleanest-looking elasticities are the least informative about durable behavior.

THE CORPORATE BARGAIN

The statutory rate fell from 46 to 34 percent, the largest corporate rate cut in the history of the income tax to that point, and the business community supported the package on the strength of that number. The price was a corporate base broadened more aggressively than the individual base: the investment tax credit repealed, depreciation lives lengthened, the minimum tax extended to corporations, and a catalogue of industry preferences curtailed. For many firms the wider base offset much of the rate benefit, and for capital-intensive firms the offset was close to complete. The bargain’s political logic mirrored the individual side. Each industry accepted the repeal of its preferences in exchange for the lower rate, and the rate, being visible and universal, was easier to sell to shareholders than the base changes were to oppose.

The afterlife of the corporate bargain diverged from the individual one in a way that mattered for the organizational response. Because the corporate rate settled at 34 percent while the individual top settled at 28, the act created the rate inversion that drove the pass-through migration, and the corporate base broadeners applied to a shrinking population of C corporations as the migration proceeded. The firms that remained in C form were disproportionately the large public companies for which pass-through status was unavailable or impractical, which meant the corporate tax increasingly fell on a narrower set of taxpayers even as its statutory rate stayed fixed. Later Congresses adjusted the corporate rate again, but the structural consequence, a corporate tax paid by fewer firms on a broader definition of income, persisted. The corporate bargain thus illustrates the article’s central asymmetry from the other direction: the rate cut was the headline and the base was the substance, and the substance outlasted the headline’s political life.

THE MINIMUM TAX BACKSTOP

The logic was straightforward. A code that had closed the most popular shelters needed a backstop for the shelters it had missed and the ones that would be invented next, a parallel computation that added back preferences and applied a flat rate so that high-income filers could not reduce liability to zero. The 1986 act broadened the minimum tax substantially, expanding the list of adjustments and preferences caught by the parallel system at the same time the regular tax was being cleaned. The individual rate, set at 21 percent, was later raised to 24 percent by the 1990 act, a change that widened the backstop rather than narrowing it.

The minimum tax endured because its function was independent of the rate debate that consumed everything else. Brackets could be raised and lowered without affecting the need for a floor under the liability of preference-heavy filers, and no later Congress found a reason to dismantle the floor. If anything, the base broadening of 1986 made the minimum tax more important in the short run, because the taxpayers displaced from shelters needed somewhere to go and the parallel system determined how far they could go. Over time the minimum tax acquired its own controversies, as the unindexed exemption pulled in filers far below the high-income population it was designed to catch, but those controversies concerned the parameters, not the existence, of the backstop. The structure survived. Like the passive loss rules, the minimum tax belongs in the ledger’s durable column: a definitional change, embedded in the architecture of the code, that later Congresses adjusted at the margins without ever reversing.

THE AVOIDANCE EVIDENCE: WHAT THE DATA SHOWED

The collapse of the shelter industry is one of those claims that sounds rhetorical until it is confronted with the numbers, and the numbers are unusually clear. Roger Gordon and Jeffrey MacKie-Mason, in their 1990 study, documented the swing in reported business income that followed the reform. Net income from partnerships and S corporations reported on individual tax returns had averaged negative 2.2 billion dollars a year from 1981 through 1986. The negative sign was the signature of the shelter era: these entities existed, in large measure, to generate paper losses for their high-bracket owners. In 1987, the first full year under the new rules, that figure swung to positive 32 billion dollars. The swing did not mean that American business had suddenly become 34 billion dollars more profitable. It meant that the losses had been artificial and the new rules had stopped their manufacture, while genuine business income, pulled by the rate inversion, flowed into individual form.

The S corporation election data told the same story from the organizational side. Approximately 375,000 businesses filed S elections in the first six months of 1987, against a semiannual average of roughly 150,000 in the years 1983 through 1986. Some of that surge was reclassification, existing C corporations converting to capture the lower individual rate. Some of it was the mirror image of the shelter collapse, partnerships that had existed to generate losses reorganizing into forms suited to reporting genuine income. The two movements were opposite sides of the same coin. The reform simultaneously destroyed the incentive to manufacture losses and created the incentive to report income in individual form, and the filing data recorded both.

What the data cannot show, and what the honest literature does not claim, is the full counterfactual. How large would the shelter industry have grown without the reform? The retail shelter market was expanding through the mid-1980s, fed by high marginal rates and an inventive promotion industry, and there is no natural ceiling visible in the pre-1986 trend. The revenue cost of the shelters was rising, the economic distortions were widening, and Congress had tried and failed to contain the industry with narrower instruments. The at-risk rules of the late 1970s had been routed around within a few years. Against that baseline, the achievement of section 469 looks larger, not smaller, than the raw numbers suggest. It did not merely reduce sheltering from its 1986 level. It arrested a growth trend that showed no sign of stopping on its own.

There is a final subtlety in the avoidance evidence that bears on the elasticity debate. Some of the income that appeared on individual returns after the reform, the surge Saez documented in business income among the top 1 percent, represented activity that had previously been sheltered or deferred rather than activity newly created. When that income surfaced, it looked like a behavioral response to lower rates. In economic substance, it was the unwinding of past avoidance, a one-time level shift rather than a permanent change in the growth rate of reported income. Distinguishing the level shift from a change in trend was one of the central challenges of the early literature, and the failure to make the distinction cleanly was one reason the early elasticity estimates ran too high. The shelter collapse was a triumph of the reform’s design, but it was a triumph that contaminated the measurement of the reform’s other effects.

WHAT DID NOT SURVIVE: THE SHORT LIFE OF THE RATE SCHEDULE

The irony of the 1986 act is that the provision the public remembers best is the one that lasted least. The two-rate individual schedule, 15 percent and 28 percent, took effect for tax years beginning in 1988 and was gone in its enacted form before the decade turned. The mechanism of its undoing was fiscal, not ideological. The federal deficit, which the revenue-neutral design of the reform had been careful not to worsen, became the dominant fact of budget politics by the end of the 1980s, and the rate structure that the reform had built became the most convenient place to find new revenue.

The first blow came from the Omnibus Budget Reconciliation Act of 1990, Public Law 101-508, signed on November 5, 1990. The 1990 act eliminated the 33 percent bubble, the surcharge mechanism that had been the price of the two-rate shape, and added a 31 percent bracket on top of the 15 and 28 percent rates, effective for tax years beginning in 1991. The two-rate structure had become a three-rate structure. The same statute raised the individual alternative minimum tax from 21 percent to 24 percent, capped the capital gains rate at 28 percent, limited itemized deductions for higher-income filers through the provision known as Pease, applicable above 100,000 dollars of adjusted gross income, and began phasing out personal exemptions for joint filers between 150,000 and 275,000 dollars of income. Each of those changes moved away from the 1986 design, but the addition of the 31 percent bracket was the decisive break. The schedule no longer had the shape the reform had given it.

The bubble’s mechanics merit a closer look, because the surcharge was the provision that most damaged the act’s claim to simplicity while doing the most distributional work. The 5 percent surcharge applied across a phase-out range, and within that range each additional dollar of income was taxed at 33 percent: the 28 percent bracket rate plus the 5 percent clawback. The surcharge phased out the benefit of the 15 percent bracket and the personal exemption for higher earners, which meant the celebrated two-rate schedule was, for a band of upper-middle-income filers, a three-rate schedule with the highest marginal rate in the middle. Taxpayers in the bubble paid a higher marginal rate than taxpayers above it, a pattern that violated the progressivity the schedule appeared to promise and that no public explanation of the reform ever made intuitive. When the 1990 act eliminated the bubble, it removed the most confusing feature of the 1986 design, and the simplification came, ironically, from the statute that began dismantling the rates.

The second blow came from the Omnibus Budget Reconciliation Act of 1993, Public Law 103-66, signed on August 10, 1993. The 1993 act added 36 percent and 39.6 percent brackets, the latter constructed as a 36 percent rate plus a 10 percent surtax on income above 250,000 dollars, and it applied the new rates retroactively to January 1, 1993. It made the Pease limitation and the personal exemption phaseout permanent. With that statute, the top individual rate stood at 39.6 percent, above the 38.5 percent transitional top rate of 1987 and far above the 28 percent that had been the reform’s signature number. The rate schedule that the 1986 act had defined in the public mind, two low rates bought with a broad base, had been dismantled within seven years of its full implementation.

It is worth pausing over what this means and what it does not mean. It does not mean the reform failed. A statute can succeed at its structural aims while its headline provisions are reversed, and the evidence below suggests that is exactly what happened. The rate schedule was always the most politically exposed part of the design, because rates are the most visible part of any tax system and therefore the most tempting target when revenue is needed. The base broadening, by contrast, was politically protected by its own invisibility. Nobody campaigns on restoring the investment tax credit. Nobody holds a rally for the 60 percent capital gains exclusion. The provisions that were hardest to see were the hardest to repeal, and the provisions that were easiest to see were the first to go.

The corporate side of the story reinforces the point. The 1986 act cut the corporate rate from 46 percent to 34 percent, and that cut survived the 1990 and 1993 reconciliation acts intact. The corporate base broadening, the repeal of the investment credit, the longer depreciation lives, the tightened loss rules, survived as well. But the relationship between the two schedules did not. In 1986 the reform had placed the top individual rate below the corporate rate for the first time in the postwar period, 28 percent against 34 percent, and that inversion was the engine of one of the largest organizational changes in American business history. When the 1990 and 1993 acts pushed the top individual rate back above the corporate rate, the engine lost its fuel, but the organizational shift it had started kept running on momentum for decades. That story belongs to the section on business organization below.

The rate progression across the period tells the story in a single line. In 1987 the transitional top rate was 38.5 percent. From 1988 through 1990 the enacted top rate was 28 percent, with the 33 percent bubble applying across a middle income band. In 1991 and 1992 the top rate was 31 percent. From 1993 onward it was 39.6 percent. Seven years after the reform took full effect, the top individual rate was higher than the transitional rate of the reform’s first year. The base that had been broadened to pay for the 28 percent rate remained broad. That asymmetry is the core of the namable claim developed at the end of this article, and it is the reason the reform is better remembered as a base-broadening achievement than as a rate-cutting one.

A note on the capital gains treatment is necessary here, because the rate story is often told without it. The 1986 act repealed the 60 percent exclusion for long-term gains, taxing them as ordinary income with a 28 percent maximum. The 1990 reconciliation act capped the gains rate at 28 percent even as it raised the top ordinary rate to 31 percent, creating a preference the 1986 act had deliberately eliminated. The 1993 act left the 28 percent cap in place while the top ordinary rate rose to 39.6 percent, widening the gap further. So the capital gains preference that the reform had abolished was partially restored within four years and in large part restored within seven, while the passive loss rules that had done the heavier structural work remained untouched. The pattern repeats. The visible provisions moved. The structural provisions stayed.

The 1987 filing season captured the transition’s confusion in miniature. Taxpayers filed under the five-bracket transitional schedule while planning under the two-rate schedule that would take effect the following year, and the mismatch between the year they were reporting and the year they were anticipating produced a wave of timing responses that researchers would spend a decade disentangling. Income was accelerated into 1986 ahead of the exclusion repeal, deductions were deferred into the lower-rate years, and the revenue statistics for 1987 reflected the calendar more than the economy. Anyone who cites the 1987 revenue figures as evidence for or against the reform is citing the transition, not the steady state, and the steady state lasted only until the 1990 act changed the parameters again.

The politics of the reversals deserve attention, because they explain why the rate schedule proved so much more fragile than the base. The 1990 reconciliation act was a deficit-reduction measure, enacted under the budget procedures that gave it its name, and its revenue provisions were chosen for their yield rather than their elegance. The 31 percent bracket, the Pease limitation, the personal exemption phaseout, and the higher alternative minimum tax were all ways of raising revenue from the individual tax system without revisiting the base broadening that the 1986 act had accomplished. That was the path of least resistance. Reopening the preference fights of 1986 would have meant reassembling the coalition that had taken years to build. Adding brackets on top of the existing base required only a majority willing to raise rates.

The 1993 act followed the same logic on a larger scale. The 36 percent bracket and the 39.6 percent rate, constructed as a 36 percent rate plus a 10 percent surtax on income above 250,000 dollars, were deficit-reduction instruments, applied retroactively to the start of the year to maximize their revenue capture. Making Pease and the personal exemption phaseout permanent locked in the base-broadening overlay. Neither act touched the passive loss rules, the investment credit repeal, or the other structural provisions of 1986, because those provisions were raising revenue every year and no deficit-driven Congress had any reason to surrender them. The pattern was consistent across both episodes. When Congress needed money, it raised the rates the 1986 act had cut and kept the base the 1986 act had broadened.

This is the political economy of the survive-or-reverse ledger in its purest form. Base broadening is fiscally productive and politically invisible, which makes it durable. Rate cuts are fiscally costly and politically visible, which makes them reversible. A reform that finances the second with the first creates a ratchet. The base, once broadened, stays broad under Congresses of both parties, because broadening raises revenue without raising rates. The rates, once cut, return upward whenever deficits demand it, because raising rates on a broad base is the most efficient way to close a fiscal gap. The 1986 act built that ratchet into the code, whether or not its authors understood what they were building, and the 1990 and 1993 acts were the ratchet turning.

The deficit targets of the period sharpen the irony. The mid-1980s deficit-reduction framework had set declining deficit targets that the actual budgets kept missing, and each missed target increased the pressure for a revenue package that the 1986 act, by design, could not supply. The 1990 budget summit that produced the Omnibus Budget Reconciliation Act was convened precisely because the targets were failing, and the rate increases it enacted were the fiscal consequence of the 1986 act’s self-imposed neutrality. A reform that had been sold as fiscally responsible left the deficit untouched, and the deficit then consumed the reform’s most visible achievement. The sequence is worth remembering whenever revenue neutrality is proposed as a sufficient fiscal virtue. Neutrality is a property of the legislation. The deficit is a property of the budget. The two are related, but they are not the same, and the 1986 act demonstrates the distance between them.

ANTICIPATION, TIMING, AND THE TRANSITION

The 1986 act did not arrive as a surprise. It was debated publicly through 1985 and 1986, passed in the autumn, enacted as Public Law 99-514 and signed on October 22, 1986, with its principal provisions phasing in across 1987 and 1988. That long public gestation matters for every measurement that follows, because taxpayers who can see a rate change coming will rearrange their affairs to meet it, and the rearrangements show up in the data as responses to the law itself.

The phase-in schedule created a staircase of incentives. For tax year 1987, a transitional five-bracket schedule applied, running from 11 percent to 38.5 percent. For tax years beginning in 1988, the two-rate structure of 15 percent and 28 percent took full effect, with the 5 percent surcharge producing the 33 percent bubble across a middle income band. A taxpayer deciding when to realize income or claim a deduction in late 1986 faced a clear map of future rates, and the rational move was to push income into the low-rate years ahead and pull deductions into the high-rate year at hand. That is calendar management, not economic behavior, but in the tax return data the two look identical unless the researcher knows to look for the difference.

The passive loss rules had their own transition, and it was designed to defuse political resistance rather than to aid measurement. Pre-enactment interests were allowed a declining share of passive losses against other income: 35 percent in 1987, 60 percent in 1988, 80 percent in 1989, 90 percent in 1990, and full limitation from 1991 onward. The glide path gave existing investors time to restructure or exit, which muted the transition-equity objection that had killed earlier reform attempts. It also meant that the provision’s full effect unfolded gradually, so the data from 1987 and 1988 captured a partial treatment. Researchers studying the early years were measuring a law still arriving, not a law fully in force.

The anticipation problem cuts deeper than the phase-in. Because the reform’s broad outlines, lower rates financed by base broadening, were public knowledge well before enactment, some of the behavioral response preceded the statute. Taxpayers accelerated deductions into 1986, deferred income into 1987 and 1988, and restructured business organizations in anticipation of the rate inversion. When researchers later compared pre-reform and post-reform income, the pre-reform baseline was already contaminated by anticipation, and the post-reform observation was inflated by retiming. Both contaminations pushed the measured elasticity upward. This is one of the central reasons the early estimates were too high, and it is why Austan Goolsbee’s finding, that the large measured responses to the 1986 cuts were concentrated in the short run around the effective dates, carries such weight. The short-run concentration is exactly the signature of retiming.

There is a broader lesson in the transition mechanics. A tax reform is never a clean before-and-after experiment, because the “before” period includes the anticipation of the reform and the “after” period includes its phase-in. The 1986 act is the most studied natural experiment in public finance precisely because its changes were large and its dates were known, but even here the measurement is entangled with the transition. The later literature’s achievement was not to find a cleaner experiment. It was to model the contamination, to separate the retiming from the real response and the reclassification from both, and to report the range that remained once those separations were made.

THE BEHAVIORAL RESEARCH: ONE LAW, A GENERATION OF ESTIMATES

No tax statute in American history has been studied as intensively as the 1986 act, and the reason is straightforward. The reform changed marginal rates sharply, differentially across income groups, and at a precisely known date, which gave researchers the closest thing public finance gets to a controlled experiment. The question the literature set out to answer was how much taxable income responds to changes in tax rates, a quantity economists call the elasticity of taxable income. The answer matters for a reason that goes beyond academic curiosity. Every revenue estimate for every subsequent tax proposal depends on an assumption about how taxpayers will respond, and the official scoring of legislation, the subject of the series companion on how Congress estimates the cost of tax legislation, stands or falls on getting that assumption approximately right.

The first major estimate came from Martin Feldstein in 1995, in the Journal of Political Economy, volume 103, number 3. Feldstein examined the responses of high-income taxpayers to the 1986 rate cuts and reported a preferred elasticity estimate of approximately 1.04, meaning that a 10 percent reduction in the net-of-tax rate was associated with roughly a 10 percent increase in reported taxable income. The finding was electrifying. If taxable income responded that strongly to rate changes, then rate cuts would lose far less revenue than conventional estimates assumed, and the policy implications were immediate. Feldstein’s paper, circulated earlier as National Bureau of Economic Research working paper 4496, became the most cited and most contested estimate in the literature.

The challenges came quickly. Gerald Auten and Robert Carroll, in the Review of Economics and Statistics, volume 81, number 4, in November 1999, used a broader panel of tax returns and reported estimates ranging from 0.54 to 1.10, with a preferred estimate near 1.33 for the highest-income filers. Austan Goolsbee, in the Journal of Political Economy, volume 108, number 2, in April 2000, studied executive compensation responses to both the 1986 cuts and the 1993 increases and found that the large responses were concentrated in the short run, driven by the retiming of income across the effective dates of the rate changes rather than by lasting changes in economic behavior. Emmanuel Saez, in Tax Policy and the Economy, volume 18, in 2004, circulated as NBER working paper 10273, found that only the top 1 percent of the income distribution showed substantial behavioral responses to the 1986 reform, and that those responses combined short-term retiming with a surge in business income reported on individual returns, the organizational shifting discussed in the next section.

How did researchers separate real responses from retiming?

They compared income reported immediately before and immediately after the rate changes took effect, then tested whether the movement persisted. Transitory spikes around the effective dates indicated retiming. Lasting changes in reported income, after accounting for income shifting between corporate and individual forms, indicated real behavioral response.

The methodological critique that reshaped the field came from Joel Slemrod in 1998, in the National Tax Journal, volume 51, number 4, in December of that year. Slemrod argued that the early estimates, Feldstein’s above all, could not distinguish real economic responses, changes in labor supply or saving, from the reclassification of existing income. When the top individual rate fell below the corporate rate, business owners had a powerful incentive to shift income from corporate to individual form, and that shifting would show up in the data as a surge in individual taxable income without any change in underlying economic activity. Timing responses compounded the problem. Taxpayers who could accelerate income into low-rate years or defer deductions into high-rate years would produce large measured elasticities that reflected nothing more than calendar management.

The synthesis that emerged, and the one the profession largely accepts, came from Emmanuel Saez, Joel Slemrod, and Seth Giertz in 2012, in the Journal of Economic Literature, volume 50, number 1, in March of that year. Surveying the full body of work, they placed the consensus elasticity in a range of approximately 0.12 to 0.40, far below the early estimates, and they concluded that Feldstein’s estimates from the 1986 reform were biased upward by control-group problems, a critique developed in detail by Jeffrey Navratil in 1995. Jon Gruber and Emmanuel Saez, in the Journal of Public Economics, volume 84, number 1, in April 2002, had reached a compatible conclusion with a preferred estimate of 0.4, using a broader panel that covered multiple rate changes across the 1980s rather than the 1986 reform alone.

Why does the elasticity range matter for revenue scoring?

Because the revenue cost of any rate change is the mechanical loss minus the revenue recovered through behavioral response, and the size of that recovery depends entirely on the elasticity. A high elasticity means rate cuts lose little revenue. A low elasticity means they lose nearly the full mechanical amount.

To see why the estimates moved so much, it helps to look inside the research designs. Martin Feldstein’s 1995 study compared the tax returns of high-income filers in 1985, before the reform, with their returns in 1988, after the rate cuts had taken effect. The identifying assumption was that the change in reported income between those two years, adjusted for observable factors, reflected the response to the lower rates. The design was elegant and the data were the best available, but the comparison had a structural weakness. The taxpayers facing the largest rate cuts were, by construction, the taxpayers with the highest incomes in the base year, and high incomes in any single year are a mixture of permanent earning power and transitory fluctuation. A filer whose 1985 income was temporarily depressed would show rapid income growth to 1988 for reasons having nothing to do with taxes, a phenomenon statisticians call mean reversion, and that growth would be misattributed to the rate cut.

Jeffrey Navratil, in a 1995 critique that the later survey literature treats as decisive, showed that this control-group problem biased Feldstein’s estimates upward. The comparison required a group of similar taxpayers who did not receive a rate cut, against whom the treated group’s response could be measured, and no such group existed in the necessary form. The taxpayers with the largest cuts were systematically different from everyone else, in income level, in income composition, and in their access to shifting opportunities, and those differences, not the rate cuts alone, explained part of the measured response. When Saez, Slemrod, and Giertz concluded in 2012 that Feldstein’s estimates from the 1986 reform were biased upward, the Navratil critique was the mechanism they had in mind.

Gerald Auten and Robert Carroll’s 1999 study improved on the design by using a larger panel of returns covering more years, which allowed them to control for mean reversion and to separate the effects of the 1986 cuts from the surrounding tax changes. Their estimates, ranging from 0.54 to 1.10 with a preferred figure near 1.33 for the highest-income filers, were still large by later standards, but the wider range they reported was itself informative. It showed how sensitive the answer was to choices about the sample period, the income cutoff, and the treatment of taxpayers whose incomes crossed the thresholds between years. Sensitivity of that kind is a warning sign. When small changes in specification produce large changes in the estimate, the estimate is measuring the specification as much as the behavior.

Austan Goolsbee’s 2000 study took a different approach that sidestepped some of these problems. Instead of studying all high-income filers, he studied corporate executives, whose compensation was reported in detail and whose ability to time income was unusually well documented. He examined their responses to both the 1986 rate cuts and the 1993 rate increases, which gave him variation in both directions. The pattern he found was striking. Executives showed large responses, but the responses were concentrated in the years immediately around the rate changes, the classic signature of retiming. When rates were about to fall, executives deferred compensation into the low-rate years. When rates were about to rise, they accelerated it. The long-run response, the change in behavior that persisted after the timing games were exhausted, was far smaller. Goolsbee’s design could not measure the whole economy’s elasticity, but it could show, with unusual clarity, how much of the measured response was calendar management.

Emmanuel Saez’s 2004 study added the distributional dimension. Using a long time series of top income shares, he found that only the top 1 percent of the distribution showed substantial behavioral responses to the 1986 reform, and that those responses combined two elements: short-term retiming around the effective dates, and a surge in business income reported on individual returns. The second element was the organizational shifting, the S corporation conversions and income reclassification that the rate inversion had triggered. Saez’s finding mattered because it located the response where the shifting opportunities were. The taxpayers with the most ability to reclassify income showed the largest responses, which is what one would expect if reclassification, not real economic activity, were driving the numbers.

Jon Gruber and Emmanuel Saez, in 2002, broadened the lens beyond the 1986 reform to the full sequence of rate changes across the 1980s, which gave them more variation and less dependence on any single episode’s peculiarities. Their preferred estimate of 0.4 sat well below the early 1986-based figures and pointed toward the consensus the survey literature would later codify. The multi-reform design had a cost, which was that it blended episodes with different shifting opportunities, but it had the virtue of testing whether the large 1986 responses survived outside the 1986 context. They did not, or not fully, which suggested that something specific to the 1986 episode, the rate inversion, the anticipation effects, the one-time reorganization wave, had inflated the early numbers.

Joel Slemrod’s 1998 methodological critique tied these threads together. He argued that the literature needed a three-way distinction, not the two-way distinction between response and non-response that the early studies had used. Taxpayers could respond in real ways, changing labor supply, saving, or investment. They could respond through avoidance, legally reclassifying income across tax bases or organizational forms. And they could respond through retiming, shifting the recognition of income across periods. Only the first of the three changes the size of the economy. The second changes the measured tax base without changing the economy. The third changes neither, except in the year of the shift. The early elasticity estimates had blended all three into a single number and labeled it the behavioral response to rate cuts. Slemrod’s taxonomy showed why that blending was a mistake, and the subsequent literature, culminating in the 2012 survey’s consensus range of approximately 0.12 to 0.40, was in large measure the profession’s effort to unblend them.

The range exists for identifiable reasons, and the brief for this article requires stating them rather than collapsing them into a single number. Estimates vary with the period studied, because the 1986 reform offered unusual shifting opportunities that later rate changes did not. They vary with the income group, because the highest-income filers have the most shifting opportunities and the strongest timing incentives, which is why Saez found substantial responses only in the top 1 percent. They vary with the treatment of timing, because income moved across the 1986 and 1987 effective dates in ways that inflated short-run estimates, which is why Goolsbee’s executive compensation study found the responses concentrated in the short run. And they vary with the treatment of organizational shifting, because the movement of business income between corporate and individual form, the S corporation surge documented below, looked like new economic activity in the individual data while being mere reclassification in the aggregate.

The distinction between real response and reclassification is the single most important idea in this literature, and it must be stated explicitly. A real response means the tax change altered economic behavior: more work, more saving, more investment, more output. Reclassification means the tax change altered only the legal form in which existing activity was reported: income moved from a corporation to a pass-through, a deduction was timed into a different year, compensation was relabeled. The two have radically different implications. Real responses change the size of the economy. Reclassification changes only the measured tax base. The early elasticity literature, by attributing reclassification to real response, overstated the economic payoff of rate cuts and understated the revenue cost. The later literature, by separating the two, brought the estimates down to the 0.12 to 0.40 range and with them the implied growth dividend.

Two cautions from the neutrality flags in the brief apply here with special force. First, every estimate in this section carries a named author, a journal, and a period, because unattributed elasticities are the raw material of policy abuse. Second, no finding in this literature extends to any current proposal. The elasticities were estimated from a specific reform, in a specific tax system, with specific shifting opportunities, and they describe how taxpayers responded to that reform, not how they would respond to a different one. The 1986 act’s organizational shifting, in particular, depended on the rate inversion that the 1990 and 1993 acts eliminated, so even the reclassification channel cannot be assumed to operate the same way twice.

For all its density, the 1986 literature leaves several questions open, and an honest assessment names them. The first is the long-run elasticity purged of every timing and shifting margin. The consensus range of 0.12 to 0.40 describes the measured response over the windows the studies could observe, but the windows were short and the shifting margins were large, so the permanent real-response parameter, the number that would describe behavior in a settled system with no arbitrage available, remains estimated rather than known. Later reforms supplied additional episodes, and researchers have applied the same methods to them, but each episode carries its own confounders and none reproduces the 1986 combination of a large rate cut with a simultaneous base broadening. The parameter is therefore better identified than it was before 1986 and worse identified than the policy debate usually assumes.

The second open question concerns heterogeneity below the top. The literature’s concentration on the top 1 percent reflects where the variation was, not where the population is, and the behavioral margins of the median filer remain less studied because the 1986 act changed their incentives less dramatically. Whether the elasticity estimated at the top applies, even approximately, to the broader filing population is unknown, and the theoretical reasons to doubt it are strong: the shifting margins that dominate the top-bracket response are largely unavailable to wage earners. The third open question is the interaction between the base and the rates over long horizons. The 1986 act changed both at once, which means no study of the episode can fully separate the behavioral effect of lower rates from the behavioral effect of the broader base. A rate cut financed by base broadening is a different treatment from a rate cut financed by borrowing, and the literature’s estimates describe the combined treatment. These limitations do not invalidate the research. They define its proper use: as a disciplined range for the combined response to a particular historical package, not as a universal constant for the scoring of packages not yet written.

The third open question is the interaction between the base and the rates over long horizons. The 1986 act changed both at once, which means no study of the episode can fully separate the behavioral effect of lower rates from the behavioral effect of the broader base. A rate cut financed by base broadening is a different treatment from a rate cut financed by borrowing, and the literature’s estimates describe the combined treatment. These limitations do not invalidate the research. They define its proper use: as a disciplined range for the combined response to a particular historical package, not as a universal constant for the scoring of packages not yet written.

THE ORGANIZATIONAL RESPONSE: THE PASS-THROUGH REVOLUTION

The most consequential behavioral response to the 1986 act was not a change in how much Americans worked or saved. It was a change in the legal form their businesses took. By cutting the top individual rate to 28 percent while cutting the corporate rate only to 34 percent, the reform made the individual tax system the cheaper place to earn business income for the first time in the postwar period. Business owners responded with a speed that surprised nearly everyone, reorganizing existing enterprises and forming new ones in pass-through form, partnerships and S corporations whose income flows directly to the owners’ individual returns.

The numbers, documented by Roger Gordon and Jeffrey MacKie-Mason in 1990 in National Bureau of Economic Research working paper 3222, are startling. In the first six months of 1987, approximately 375,000 businesses filed elections to be taxed as S corporations, compared with a semiannual average of roughly 150,000 in the years 1983 through 1986. The flow of business income told the same story. Net income from partnerships and S corporations reported on individual returns, which had averaged negative 2.2 billion dollars a year from 1981 through 1986, the negative figure reflecting the shelter-era losses, swung to positive 32 billion dollars in 1987. That swing combined two effects of the reform: the passive loss rules shut down the artificial losses, and the rate inversion pulled genuine business income into individual form.

Why did the corporate-individual rate inversion matter?

Because it made pass-through taxation strictly cheaper than corporate taxation for distributed earnings. A dollar earned through a C corporation faced the 34 percent corporate rate plus the individual tax on dividends, while the same dollar earned through an S corporation faced only the 28 percent individual rate.

The composition of the shift mattered as much as its size. Roger Carroll and David Joulfaian, in a 1997 study, found that the surge in S corporation elections was concentrated among profitable firms, which is exactly what the incentive structure predicted. Firms with real earnings to shelter from the higher corporate rate had the strongest reason to convert. Firms with losses had less reason, since the corporate form preserved the value of loss carryforwards. The shift was therefore not a random reorganization. It was a targeted migration of profitable business activity toward the cheaper tax form, and it continued long after the rate inversion that triggered it had been reversed.

Alan Auerbach and Joel Slemrod, in their 1997 survey of the reform’s economic effects in the Journal of Economic Literature, volume 35, number 2, attributed the jump in S corporate activity to the act’s increase in the relative corporate tax burden, the precise mechanism the rate inversion created. Their assessment is worth quoting in substance: the organizational response was the clearest and most measurable behavioral effect of the reform, clearer than any change in labor supply or saving, because it showed up directly in the filing data rather than having to be inferred from aggregates.

The pass-through shift outlived the rate structure that caused it, and that fact deserves emphasis. The 1990 and 1993 reconciliation acts pushed the top individual rate back above the corporate rate, eliminating the inversion. But the businesses that had converted to S corporation or partnership form did not convert back. Organizational form is sticky. Converting back would have triggered tax costs, legal costs, and the loss of the accumulated advantages of the pass-through structure, and the professional infrastructure of accountants and lawyers that had grown up to service the new forms kept steering new businesses toward them. What began as a tax arbitrage became the default architecture of American small and medium business, and it stayed that way for decades. Few contemporaries anticipated it. The reform’s drafters had debated the rate inversion as a technical matter of horizontal equity, not as the trigger for a structural transformation of business organization.

The 1986 act thus produced a ratchet effect: a temporary rate differential caused a permanent organizational shift, because the costs of moving were asymmetric and the infrastructure, once built, lowered the cost of staying. Few provisions in the history of the income tax have shown that taxpayers respond to incentives with their lawyers as well as their labor.

The later history of pass-through taxation belongs to a separate article in this series, the companion on the 2017 tax act’s pass-through deduction, which describes a provision enacted three decades later in direct response to the organizational landscape the 1986 reform had created. The connection is structural, not incidental. By 2017 the pass-through sector was so large a share of American business income that any major tax legislation had to address it, and the 1986 act was the reason it was that large. The reform’s least anticipated consequence became one of its most permanent.

WHY THE ORGANIZATIONAL SHIFT STUCK

The pass-through migration deserves a closer look, because its persistence is the strongest evidence that the reform’s structural consequences outlasted its rate schedule. The mechanism that started the migration was the rate inversion, the top individual rate of 28 percent sitting below the corporate rate of 34 percent. The puzzle is why the migration continued after the inversion was reversed, and the answer lies in the stickiness of organizational form.

Converting a business from corporate to pass-through form involves legal costs, accounting costs, and in some cases tax costs on the conversion itself. Those are one-time frictions, but they are followed by a subtler lock-in. Once a business operates as an S corporation or partnership, its contracts, its financing arrangements, its accounting systems, and its owners’ estate plans are built around that form. Converting back would mean rebuilding all of it, and the tax saving from converting back, after the 1990 and 1993 acts restored the individual rate above the corporate rate, was uncertain and in many cases negative once the full set of provisions was considered. The rational choice for most converted businesses was to stay where they were.

The professional infrastructure reinforced the inertia. In the years after the reform, a generation of accountants and tax lawyers built their practices around pass-through planning, and their advice to new businesses reflected the world they knew. A new enterprise forming in 1995 or 2000 was not choosing between the 1986 rate schedules. It was choosing between organizational forms in a market where the pass-through form had deeper professional support, more standardized documents, and a longer track record. The tax arbitrage of 1987 had become the default architecture of American small and medium business by the mid-1990s, and defaults are powerful.

Roger Carroll and David Joulfaian’s 1997 finding, that the S corporation surge was concentrated among profitable firms, sharpens the picture. The migration was not a fad or a paperwork exercise. It was a targeted movement of real earnings toward the cheaper tax form, undertaken by the firms with the most to gain. Profitable firms had earnings to shelter from the higher corporate rate. Unprofitable firms, which could use corporate losses against future corporate income, had less reason to move. The selectivity of the migration is what makes it a genuine behavioral response rather than a statistical artifact. It followed the incentives exactly as theory predicted.

Alan Auerbach and Joel Slemrod’s attribution of the S corporate jump to the act’s increase in the relative corporate tax burden completes the causal chain. The reform did not merely lower both rates. It changed the relationship between them, and the relationship was what mattered. A parallel cut that had preserved the pre-1986 ordering, with the individual rate above the corporate rate, would have produced no such migration. The organizational transformation was therefore not a generic consequence of lower rates. It was a specific consequence of this reform’s specific rate structure, which is why the episode is so instructive. Small differences in statutory design can produce large differences in economic organization, and the differences persist long after the design is superseded.

The long tail of the shift is visible in the decades that followed. The pass-through sector kept growing as a share of American business income through the 1990s and 2000s, driven by forces beyond the original tax incentive, including the rise of limited liability companies under state law and the continued development of the professional infrastructure. By the time Congress took up business taxation again in major legislation, it was legislating for an economy in which pass-through income was a central fact, and the provisions it wrote reflected that fact. The 1986 reform’s least anticipated consequence had become one of the permanent background conditions of American tax policy.

THE SIMPLIFICATION QUESTION

Of the three stated aims of the reform, fairness, efficiency, and simplicity, the third fared worst, and the evidence on this point is unusually direct. Joel Slemrod, in 1992, published the definitive assessment in the Journal of Economic Perspectives, volume 6, number 1, in the winter of that year, under the title that asked the question outright: “Did the Tax Reform Act of 1986 Simplify Tax Matters?” His answer, after reviewing the compliance cost evidence, deserves to be quoted at length, because it is the most careful statement the literature produced: “compliance cost of the income tax system is probably higher now than it was in the early 1980s. This suggests that the Tax Reform Act achieved little, if any, simplification in the tax system, although it remains possible that the Act dampened what would have been an even greater increase in compliance cost.”

The paradox is worth unpacking. The reform eliminated dozens of preferences, collapsed fourteen statutory rates into two, and removed millions of low-income filers from the rolls. By every simple count of provisions and brackets, the code got shorter. Yet compliance costs, the time and money taxpayers spend figuring out what they owe, did not fall. Slemrod’s explanation was that the reform replaced simple preferences with complex anti-abuse rules. The passive loss provisions, the expanded alternative minimum tax, the at-risk rules, the new limitations on interest deductions, each required taxpayers and their advisers to master a new body of intricate law. The code had fewer lines but harder lines. A taxpayer who had once claimed a straightforward investment credit now had to determine whether an activity was passive, whether participation was material, whether the at-risk rules applied, and whether the alternative minimum tax produced a different answer, and each determination carried its own recordkeeping burden.

The lesson generalizes beyond this statute. Simplification by base broadening works only if the base, once broadened, is left alone, and the American legislative process does not leave tax bases alone. Every subsequent Congress found new reasons to add targeted provisions, and each addition carried its own compliance cost. The 1986 reform may have dampened what would otherwise have been a steeper rise in complexity, as Slemrod allowed, but that is a counterfactual no one can measure. What can be measured is the compliance burden, and the burden did not fall. The reform that is sometimes credited with simplifying the code should be credited instead with demonstrating how difficult simplification is: even the most aggressive base-broadening effort in postwar history, enacted with unusual political consensus, could not produce a lasting reduction in the cost of compliance.

The measurement problem behind Slemrod’s finding deserves emphasis, because it explains why the simplification debate never produced a clean answer. Compliance cost is not a line item anyone reports. It must be inferred from surveys of time spent, from the fees paid to preparers, and from the imputed value of taxpayer hours, and each of those measures carries its own error. The Internal Revenue Service’s own taxpayer burden models, which came later, confirmed the general direction of Slemrod’s result without resolving the attribution question. What portion of the compliance burden in 1992 belonged to the 1986 act, and what portion belonged to the ordinary accretion of new provisions, new regulations, and new reporting requirements that every Congress adds? The data cannot fully separate the two, which is why Slemrod framed his conclusion as a suggestion rather than a verdict.

There is also a conceptual difficulty that the compliance-cost numbers miss. A simpler code is not the same as a shorter code, and a shorter code is not the same as a code with fewer brackets. The 1986 act shortened the statute and reduced the number of rates, and by those measures it simplified. But it also introduced concepts, passive versus nonpassive, material participation, the alternative minimum tax as a parallel system, that required new kinds of thinking from taxpayers and preparers. The cognitive burden of a tax system depends on the difficulty of its hardest determinations, not on the count of its provisions, and by that measure the reform arguably made the system harder. The taxpayer who itemized under the old law faced a long list of simple questions. The taxpayer under the new law faced a shorter list of harder ones.

REAL ESTATE: THE SECTOR THAT FELT IT FIRST

No sector of the economy absorbed the reform’s impact more directly than real estate, and the research on that impact illustrates both the power and the limits of tax-driven analysis. Before 1986, commercial and rental real estate had been the preferred vehicle of the shelter industry, combining accelerated depreciation, the investment tax credit, generous interest deductions, and the ability to pass paper losses through to high-bracket investors. The reform attacked every one of those advantages simultaneously. Depreciation lives were lengthened, the investment credit was repealed, the passive loss rules trapped the paper losses, and the lower marginal rates reduced the value of every remaining deduction. The after-tax economics of tax-motivated real estate changed overnight.

James Poterba, in 1992, in National Bureau of Economic Research working paper 3963, “Taxation and Housing: Old Questions, New Answers,” provided the most careful accounting of what followed. On the owner-occupied side, Poterba found that the rate cuts reduced the user-cost distortion, the wedge the tax system drove between the cost of owning and the cost of renting, which was a genuine efficiency gain. But the increase in the standard deduction removed several million middle-income homeowners from the ranks of itemizers, and the mortgage interest deduction became a benefit concentrated at the top: in 1988, Poterba calculated, more than half of the tax losses from the mortgage interest deduction accrued to the top 8 percent of taxpayers. On the rental side, the reform reduced the tax incentives for rental housing investment, and Poterba connected that reduction to the collapse in new multifamily housing starts, which fell from 500,000 units a year in 1985 to under 150,000 a year by 1991. He also found a striking distributional pattern in the 1983 to 1986 period: trade-up homes appreciated 13 percent less than starter homes, consistent with tax-driven distortions in the relative demand for housing types.

The transmission from tax provision to housing market ran through three distinct channels, and keeping them separate clarifies what the reform did. The first channel was the user cost of owner-occupied housing, where lower marginal rates reduced the value of deductions and moved the effective price of housing services toward its economic cost. The second channel was the after-tax return on rental investment, where the passive loss rules, the investment credit repeal, and longer depreciation lives simultaneously reduced the tax advantages that had drawn capital into multifamily construction. The third channel was the financing environment, where the savings-and-loan crisis was contracting credit for real estate development for reasons independent of the tax code. The construction collapse reflected all three, and the research can allocate the decline among them only approximately. What is clear is that the tax channels moved in the same direction as the credit channel, which made the downturn deeper than any single cause would have produced.

The distinction between the price effects and the quantity effects matters for the efficiency assessment. The user-cost channel was an efficiency gain almost by definition: reducing a tax subsidy moves consumption toward the undistorted level. The rental-investment channel is harder to sign. To the extent that pre-1986 multifamily construction had been driven by tax considerations rather than by housing demand, its decline was a correction, capital moving to better uses. To the extent that the reform overshot, discouraging rental construction that demand would have supported, the decline was a cost. Poterba’s evidence suggests both were present, and the honest reading is that the data do not permit a clean decomposition. The reform ended tax-driven overbuilding. Whether it also caused market-driven underbuilding in the years that followed remains an open question in the literature, and the brief for this article requires leaving it open rather than resolving it by assertion.

Commercial real estate absorbed a parallel shock with its own dynamics. The lengthened depreciation schedules raised the after-tax cost of structures, the passive loss rules removed the shelter equity that had financed office and retail development, and the repeal of the capital gains exclusion reduced the after-tax proceeds of sale. Markets that had been built on the assumption of ever-rising values and ever-available tax-motivated capital faced both assumptions failing at once, and the distress showed up first in the regions where the 1980s building boom had been most exuberant. The interaction with the savings-and-loan crisis compounded the damage: thrifts that had funded commercial development on inflated appraisals failed as values fell, the failures tightened credit for the entire sector, and the credit contraction pushed values down further. Disentangling the tax contribution from the credit-cycle contribution is difficult, and careful accounts assign the larger share to the lending collapse, but the tax change determined which projects were viable at the margin and therefore which markets cleared first. The episode became a cautionary case in the interaction of tax incentives with leveraged asset markets: a preference that encourages borrowing against appreciating collateral creates fragility that the repeal of the preference then exposes.

The regional variation in the housing effects also deserves mention, because the national aggregates conceal it. Ling’s 1986 projection distinguished fast-growth markets, where value declines were expected not to exceed 4 percent, from no-growth markets facing declines up to 8 percent, and the realized pattern roughly followed that geography. Markets with strong employment growth absorbed the reduced tax subsidy through continued demand; markets without it saw the subsidy withdrawal translate directly into lower prices. The rental construction collapse was similarly uneven, concentrated in the Sun Belt metros where the shelter-financed apartment boom had been largest. The national story of the 1986 act’s housing effects is therefore a composite of local stories, and the local variation is a reminder that a uniform national tax change lands on fifty different housing markets with fifty different elasticities of supply.

The contemporaneous forecasts, it should be said, were gloomier than the outcomes. David Ling, writing in 1986 before the effects could be observed, projected residential rent increases of 10 to 15 percent and commercial rent increases of 5 to 10 percent, with market-value declines unlikely to exceed 4 percent in fast-growth markets and up to 8 percent in no-growth markets. Those were projections, not findings, and they must be read as such. The rental increases Ling foresaw did not materialize on the scale or schedule he projected, in part because the late-1980s economy and the savings-and-loan crisis were moving rents and values for reasons that had nothing to do with the tax code. The episode is a useful reminder that even careful contemporaneous analysis struggles to separate the tax effect from the cycle, a problem that runs through the entire literature on this reform.

What can be said with confidence is narrower but still substantial. The reform ended the era in which real estate investment was driven primarily by tax considerations, and that was a deliberate objective of the base broadening. Whether the multifamily construction collapse represented an efficient correction or an overcorrection remains debated, and the honest answer is that the data do not fully separate the tax channel from the credit cycle that was devastating real estate finance in the same years. The brief for this article requires that growth and sectoral claims carry their confounders, and the savings-and-loan crisis is the confounder that matters most here.

Poterba’s user-cost analysis deserves a fuller explanation, because it captures the efficiency logic of the reform in a single mechanism. The user cost of housing capital is the effective price a household pays for the services of an owner-occupied home, and the tax system enters that price through the deductibility of mortgage interest and property taxes, the nontaxation of imputed rent, and the marginal rate at which deductions are taken. Before the reform, high marginal rates made the deductions valuable, which subsidized owner-occupied housing relative to other investments and encouraged households to buy more housing than they otherwise would. The rate cuts reduced the value of the deductions and therefore reduced the subsidy, moving the user cost closer to the true economic cost. That was a textbook efficiency gain: less tax-driven overconsumption of housing, more capital available for investments chosen on their merits.

The distributional side of the same mechanism ran the other way. Because the value of a deduction rises with the marginal rate, and because itemizing requires deductions large enough to exceed the standard amount, the mortgage interest deduction had always been worth more to affluent households. The reform’s higher standard deduction pushed millions of middle-income homeowners out of itemizing entirely, which meant they lost the deduction’s benefit while keeping none of its complexity. What remained was a subsidy concentrated at the top, which Poterba’s 1988 calculation documented: more than half the tax losses from the provision accrued to the top 8 percent of taxpayers. The provision survived politically because it was defended as a middle-class benefit, but its incidence after the reform told a different story.

GROWTH AND THE AGGREGATES: WHAT DID NOT MOVE

The question readers most want answered, whether the reform grew the economy, is the question the research answers most firmly in the negative. Alan Auerbach and Joel Slemrod, in their 1997 survey in the Journal of Economic Literature, volume 35, number 2, in June of that year, “The Economic Effects of the Tax Reform Act of 1986,” examined the full range of aggregate evidence and concluded that the primary measured impact of the act was a shift in the composition of reported income, not a change in its level. Their central finding, stated on page 626 of the survey, was that “the aggregate values of labor supply and saving showed little measurable response” to the reform. Henry Aaron, reviewing the evidence for the Brookings Institution, reached the same conclusion in plainer language: the act had “little apparent and less than the expected effect on broad economic aggregates.”

That finding should surprise no one who has followed the elasticity literature, because the two bodies of work are consistent. If the large early elasticity estimates had been correct, if reported taxable income had surged in response to lower rates through genuine new economic activity, then the aggregates should have shown it. They did not. The later, lower elasticity estimates, the 0.12 to 0.40 consensus range, are consistent with an act whose main effects were reclassification and reorganization rather than new production. The composition of income changed. The amount of income did not change by any amount the data can distinguish from noise.

The confounders that make this finding difficult to state with more precision must be named, because the brief requires it. The reform phased in across 1987 and 1988, years when the American economy was already expanding strongly in the late stages of a long business-cycle upswing, and monetary policy was tight. The stock market crash of October 1987 landed in the middle of the phase-in period, scrambling asset values and expectations. The lingering effects of the Economic Recovery Tax Act of 1981, the large rate cuts of the early 1980s, were still working through the economy. And the 1990 and 1993 reconciliation acts began reversing the rate structure before researchers had finished measuring the effects of the original. There is no credible counterfactual, no parallel American economy that did not enact the reform against which the actual economy can be compared, and every estimate of the aggregate effect is therefore an exercise in attributing movement to causes among many competing candidates.

None of this means the reform was without economic value. An act can improve the allocation of resources without increasing their total, and the base broadening did exactly that. By removing the tax advantages that had steered capital into shelters and tax-motivated real estate, the reform let investment decisions respond more to economic merit and less to tax engineering. That is a real gain, even if it does not show up in the growth rate. But it is a gain in efficiency, not in output, and the distinction matters for how the reform is remembered. The claim that the 1986 act paid for itself through faster growth, a claim sometimes made by later advocates of rate cuts, finds no support in the research record. The act was scored as revenue-neutral, it was designed to be revenue-neutral, and the aggregates behaved as revenue-neutral legislation would be expected to behave.

The absence of a credible counterfactual is the deepest problem in the growth literature, and it deserves to be stated without euphemism. To know what the reform did to aggregate output, one would need to observe the American economy of 1987 through 1992 both with and without the statute, holding everything else equal. That observation is impossible. What researchers observe instead is a single economy moving through time, buffeted by monetary policy, by the stock market crash of October 1987, by the savings-and-loan crisis, by the lingering effects of the 1981 tax cuts, and by the 1990 and 1993 reversals that began undoing the rate structure before its effects could be fully measured. Attributing the economy’s movement to the tax reform, as opposed to any of these competing forces, requires assumptions about how the economy would otherwise have behaved, and those assumptions are doing most of the analytical work.

The composition finding survives this problem better than any aggregate claim could, because it rests on direct observation rather than on a counterfactual model. Researchers did not need to imagine an alternative economy to see that reported partnership and S corporation income swung from negative to positive, or that S elections tripled, or that the retail shelter market vanished. Those were facts about the observed world, not inferences about an unobserved one. The growth finding, by contrast, is an inference, and its confidence interval reflects the competition among causes. Auerbach and Slemrod’s conclusion, that labor supply and saving showed little measurable response, is best read as a statement about the limits of what the data can show: whatever the reform did to the aggregates, it was small enough to be lost among the larger forces moving the late-1980s economy.

THE DEPRECIATION COMPROMISE

The compromise on cost recovery illustrates the general principle in its most mechanical form. The pre-1986 code allowed rapid depreciation of equipment and structures, which, combined with the investment tax credit, made the effective tax rate on many new investments low or even negative. The 1986 act lengthened depreciation lives and repealed the credit, raising the effective burden on new capital even as the statutory rate fell. For the drafters this was a feature, not an oversight: the revenue lost to the rate cuts had to come from somewhere, and cost recovery was one of the largest available sources. But the choice had a predictable consequence for the growth debate. Every study that looked for an investment boom after 1986 was looking for the effect of the rate cuts while the depreciation changes worked against it, and the absence of a boom is therefore evidence about the package, not about the irrelevance of incentives. The depreciation compromise also shaped the distributional story, because capital-intensive industries bore more of the base broadening while labor-intensive services captured more of the rate benefit. The reform’s celebrated neutrality across activities was real in design and rough in execution, and the roughness is visible in the investment data’s failure to move.

THE DISTRIBUTIONAL RECORD: WHO PAID AND WHO STOPPED PAYING

The fairness objective, the first of the reform’s three stated aims, can be assessed in the same survive-or-reverse framework as the provisions themselves. The reform changed who bore the tax burden in two directions at once, lightening it at the bottom through base narrowing and shifting it at the top through the destruction of shelters, and the distributional consequences of those changes were among the reform’s most deliberate achievements.

At the bottom of the income scale, the changes were unambiguous. The earned income credit grew from 11 percent to 14 percent of the first 5,714 dollars of earnings, increasing both the credit’s value and the number of workers who qualified. The personal exemption climbed from 1,080 dollars in 1986 to 2,000 dollars by 1989, and the standard deduction rose to 5,000 dollars for joint filers in 1988. Together these provisions lifted millions of low-income households off the income tax rolls entirely. Secondary accounts drawing on Congressional Research Service figures put the number near six million, a figure that should be read as an order of magnitude rather than a precise count, since it depends on filing-behavior assumptions. The direction, however, is not in doubt. The reform made the income tax a tax that fewer poor households paid, and that change was part of the design, not an accident of it.

At the top of the income scale, the distributional story runs through the shelters. Before the reform, the combination of high statutory rates and generous preferences had produced a class of high-income filers whose effective tax rates bore little relation to the schedule. The passive loss rules, the investment credit repeal, and the lengthened depreciation schedules raised the effective rates of precisely those filers, not by raising the statutory rate they faced but by removing the mechanisms through which they had avoided it. This was horizontal equity in action, the principle that taxpayers in similar economic circumstances should face similar burdens. Two executives with the same salary had paid wildly different taxes before 1986 if one had bought into a shelter partnership and the other had not. After the reform, their burdens converged, because the shelter was gone.

The middle of the distribution experienced the reform mainly through the standard deduction. The increase to 5,000 dollars for joint filers removed several million middle-income homeowners from the ranks of itemizers, as James Poterba documented in his 1992 study. For those households, the change was a simplification of a modest kind: they no longer needed to track deductible expenses because the standard amount exceeded what they could itemize. But the change also concentrated the remaining itemized deductions, above all the mortgage interest deduction, among higher-income filers. Poterba calculated that in 1988, more than half of the tax losses from the mortgage interest deduction accrued to the top 8 percent of taxpayers. The deduction that had been defended as a middle-class benefit had become, in its incidence if not its design, a benefit for the affluent.

The distributional pattern that emerges is a barbell, and the barbell explains the politics of everything that followed. The largest gains went to the two ends of the distribution: high-income wage earners, who received the deepest rate cuts with the least exposure to the base broadeners, and low-income households, who were removed from the tax rolls entirely by the enlarged exemption, the doubled standard deduction, and the expanded credit. The middle bore the mixed effects: the standard deduction simplified their filing but the bubble raised their marginal rate, the exemption sheltered more of their income but the Pease limitation and the phaseout, arriving in 1990, clawed some of it back for the upper middle. The concentrated losses fell on the shelter investors and the real estate syndication industry, groups too narrow and too compromised to reverse their defeat. When the 1990 and 1993 acts raised the top rates, they were taxing the end of the barbell that had gained the most from the 1986 cuts, which is why the reversals were politically sustainable. A rate increase that falls on the winners of the last reform is the easiest tax increase to pass, and the 1986 act had helpfully identified the winners in advance.

The family-size dimension of the distributional story is often overlooked, and the personal exemption is the reason. Because the exemption applied per person, the near-doubling from 1,080 to 2,000 dollars delivered larger absolute benefits to larger families, and the effect compounded with the doubled standard deduction for households whose income had previously been fully absorbed by exemptions and the standard amount. A married couple with three children saw 10,000 dollars of income removed from the base by the exemption alone at the 1989 level, before the standard deduction applied. This was base narrowing with a demographic tilt, and it explains part of the six-million figure: the households removed from the rolls were disproportionately families with children whose earnings had been modest enough that the enlarged tax-free threshold covered them entirely. The provision’s design thus carried a family policy inside the rate reform, one that later Congresses preserved and extended through the child credit rather than through the exemption itself.

The political logic of the distributional bargain is worth stating plainly, because it explains why the coalition held. The rate cuts delivered their largest statutory benefits to high-bracket filers, who faced the steepest marginal rates. The base broadening imposed its largest costs on the same filers, who had been the principal users of the shelters and preferences. The two sides of the ledger offset each other within the same income groups, which meant that the reform could be presented, with some justification, as changing the composition of high-income tax burdens rather than their level. Low-income households, who got the exemption and credit expansions without losing preferences they had never had, were unambiguous winners. That structure, gains at the bottom financed by the elimination of avoidance at the top, was the distributional signature of the act.

The 1990 and 1993 reconciliation acts layered a different distributional logic on top of the 1986 base. The Pease limitation on itemized deductions, the phaseout of personal exemptions, and the new 36 and 39.6 percent brackets were all aimed explicitly at higher-income filers, and they moved the system away from the 1986 design in a progressive direction. Those changes belonged to the deficit politics of their own era, not to the reform’s design, but they interacted with the 1986 base in ways that amplified its progressivity. The broad base the reform had built made the higher rates of the 1990s more productive of revenue than they would otherwise have been, which is one of the underappreciated fiscal legacies of the base broadening. A narrow base with high rates invites avoidance. A broad base with high rates collects revenue. The reconciliation acts benefited from the base the reform had left them.

EFFICIENCY WITHOUT GROWTH: THE DISTINCTION THAT MATTERS

The finding that the reform did not measurably increase aggregate output is sometimes misread as a finding that it did nothing of economic value, and the misreading rests on a confusion between efficiency and growth. Growth means more output: more labor, more capital, more production. Efficiency means better allocation: the same labor and capital deployed where they earn the highest return. A tax reform can improve the second without increasing the first, and the evidence suggests that is what the 1986 act did.

The mechanism was the removal of tax-driven distortions in investment. Before the reform, capital flowed toward activities the code favored, shelter partnerships, tax-motivated real estate, equipment churned for depreciation, regardless of their economic merit. After the reform, with the preferences gone and the rates lower and flatter, investment decisions responded more to pre-tax returns and less to tax engineering. That reallocation was a real gain in economic welfare even though it left the growth rate unchanged. An economy that builds the apartments people want to live in, rather than the apartments that generate the best tax losses, is a better economy at any given level of output.

The distinction matters for how the reform is judged against its aims. The efficiency objective was about reducing distortions, and on that objective the act has a credible claim to success. The growth objective was never one the reformers stated; they promised a fairer, more efficient, simpler code, not a faster-growing economy. Measuring the act against a growth promise it never made, and then declaring it a failure for missing the target, is the same error as crediting it with revenues it never collected. The honest scorecard separates the aims the statute announced from the aims later commentators projected onto it, and on its own announced aims the efficiency record is the strongest part of the case.

THE SURVIVE-OR-REVERSE LEDGER

The findable artifact for this article is the ledger its argument requires: each major change the act made, whether it remains operative, the statute that modified it if one did, and the measured or estimated effect. The table below collects the provisions discussed above and several companion changes, so that the pattern, structural provisions surviving and headline provisions reversing, can be seen at a glance.

Major change made by the act	Operative status	Modifying statute, if any	Measured or estimated effect
Two-rate individual schedule, 15 and 28 percent, effective for tax years beginning 1988	Reversed	Omnibus Budget Reconciliation Act of 1990, Public Law 101-508, added 31 percent bracket; Omnibus Budget Reconciliation Act of 1993, Public Law 103-66, added 36 and 39.6 percent brackets	Top individual rate 39.6 percent from 1993, above the reform’s 28 percent; rate schedule dismantled within seven years
33 percent bubble rate from 5 percent surcharge	Reversed	Omnibus Budget Reconciliation Act of 1990, Public Law 101-508, eliminated the surcharge	Bubble mechanism removed; three-rate schedule of 15, 28, and 31 percent took its place for 1991
Transitional five-bracket schedule, 11 to 38.5 percent, for tax year 1987	Reversed	Superseded by the two-rate schedule for tax years beginning in 1988	One-year bridge; the top rate fell from 50 percent to 38.5 percent before reaching 28 percent
Passive activity loss rules, section 469	Operative, never repealed	None for the core rule; 1993 reconciliation act added the real estate professional exception, section 469(c)(7)	Collapse of the mass-market retail tax shelter industry; reported partnership and S corporation net income swung from negative 2.2 billion dollars average, 1981 to 1986, to positive 32 billion dollars in 1987, per Gordon and MacKie-Mason 1990
Repeal of the investment tax credit	Operative, never restored	None	Removed a central shelter subsidy; no Congress has reenacted the credit in its pre-1986 form
Lengthened depreciation schedules	Operative in substance	Subsequent acts adjusted specific asset lives	Reduced the tax motivation for equipment and real estate sheltering; specific lives modified without restoring pre-1986 acceleration
Repeal of the 60 percent capital gains exclusion, gains taxed as ordinary income with 28 percent maximum	Partially reversed	1990 reconciliation act capped gains at 28 percent while raising top ordinary rate to 31 percent; 1993 act kept 28 percent cap against 39.6 percent top ordinary rate	Preference the reform abolished was in large part restored within seven years
Corporate rate cut, 46 to 34 percent	Operative	None through the 1990 and 1993 reconciliation acts	Survived both rate reversals; corporate base broadening also survived
Individual rate below corporate rate, 28 versus 34 percent	Reversed	1990 and 1993 reconciliation acts raised top individual rate above corporate rate	Inversion lasted only until 1991; the pass-through organizational shift it triggered continued for decades
Phaseout of consumer interest deduction	Operative	None	Completed on schedule; personal interest no longer deductible
Repeal of state and local sales tax deduction, income tax deduction retained	Operative	None	Sales tax deduction never restored as an itemized deduction in its pre-1986 form
Expanded individual alternative minimum tax	Operative in expanded form	1990 reconciliation act raised individual AMT rate from 21 to 24 percent	Broader AMT base persisted; rate adjusted upward rather than repealed
Increased standard deduction and personal exemption	Operative, later adjusted	Subsequent inflation adjustments and legislation	Removed several million middle-income homeowners from itemizer ranks; low-income filer rolls reduced by near six million per CRS-derived secondary accounts
Earned income credit expansion, 11 to 14 percent	Operative, later expanded	Subsequent legislation expanded the credit further	Foundation for later EITC growth; original expansion never reversed
Pease limitation on itemized deductions and personal exemption phaseout	Added after the reform, not part of it	Created by 1990 reconciliation act; made permanent by 1993 reconciliation act	New base-broadening provisions layered onto the 1986 base, moving further from the reform’s rate design

WHY ECONOMISTS STILL TREAT IT AS THE REFERENCE CASE

A reform whose signature rate schedule lasted less than five years might seem an odd candidate for the status the 1986 act enjoys in public finance, where it remains the reference case against which every subsequent reform proposal is measured. The reasons for that status follow directly from the evidence reviewed above, and they have little to do with the rates.

The first reason is the quality of the natural experiment. The act changed marginal rates sharply, at a known date, differentially across taxpayers, and in combination with base-broadening provisions that altered shifting opportunities in observable ways. That combination generated the data that built the elasticity literature that grew up around the reform, and every subsequent estimate of taxpayer responsiveness is calibrated, explicitly or implicitly, against the 1986 evidence. The consensus range of 0.12 to 0.40 that Saez, Slemrod, and Giertz reported in 2012 is the starting point for revenue estimation across the profession, and it exists because the 1986 reform gave researchers something to measure. The series companion comparing the 1986 reform with the 2017 tax act examines how that reference status shaped the later debate.

The second reason is the demonstration that base broadening is politically possible. Before 1986, the conventional wisdom held that concentrated preference holders would always defeat diffuse rate-cut beneficiaries, and that no coalition could assemble to repeal enough preferences to finance meaningful rate reduction. The 1986 act proved otherwise, under divided government, with a Republican president and a Democratic House, through the regular legislative order. That proof has independent value regardless of what happened to the rates afterward. Every subsequent reform effort had to answer the question the 1986 act answered affirmatively: can the preferences be repealed. The answer the act supplied was yes, with the qualifier that the repeal must be purchased with something the preference holders value more, which in 1986 was lower rates.

The third reason is the negative demonstration, the warning the act’s aftermath carries. The rate schedule was dismantled within seven years because rates are the most visible and therefore the most politically exposed part of the tax system. The base broadening survived because it was structural and comparatively invisible. That asymmetry is a finding about political economy, not only about this statute, and it disciplines every subsequent reform design. A reform that buys its rate cuts with base broadening must reckon with the possibility that a later Congress will keep the base and raise the rates, which was exactly what the 1990 and 1993 acts did. The rate cuts of 2001 and 2003, examined in the series companion on the Bush-era tax legislation, were designed with that history in view, and their different fate reflected the different fiscal circumstances rather than a different lesson.

The fourth reason is the organizational transformation. The pass-through shift that the rate inversion triggered, documented by Gordon and MacKie-Mason and extended by Carroll and Joulfaian, permanently changed the structure of American business taxation. By the time later Congresses legislated on business income, they were legislating for a pass-through economy that the 1986 act had created, whether or not they acknowledged it. A reference case earns its status by continuing to explain the world after its own provisions have been superseded, and on that test the 1986 act qualifies. Its rates are gone. Its consequences are not.

The reference status also shapes how the reform functions in policy argument, and here a caution is in order. Because the 1986 act is the best-studied reform, it is also the most cited, and citation is not the same as applicability. Advocates of rate cuts have invoked the large early elasticity estimates long after the literature moved past them, presenting Feldstein’s 1.04 as the profession’s view when the survey consensus had settled near 0.12 to 0.40. Advocates of base broadening have invoked the shelter collapse as proof that preferences can always be repealed, without acknowledging the unusual coalition, the revenue-neutral constraint, and the divided government that made the 1986 bargain possible. The reference case is most useful when it disciplines argument, and least useful when it is mined for congenial numbers.

What keeps the reform central, despite the misuse, is that no subsequent episode has replaced it as a source of evidence. Later rate changes, the 1990 and 1993 increases, the 2001 and 2003 cuts, the 2017 act, each generated their own literatures, but none combined rate changes as large, base changes as sweeping, and data as accessible as the 1986 reform did. Researchers estimating the effects of a new proposal still begin by asking what the 1986 evidence implies, then adjust for the differences in design and context. That procedure, begin with 1986 and adjust, is what reference-case status means in practice. It does not mean the 1986 results transfer directly. It means they are the starting point from which the adjustments are made, and the adjustments, for shifting opportunities, for timing, for the rate structure’s interaction with organizational form, are where the real analytical work happens.

THE VERDICT: STRUCTURE AGAINST HEADLINES

The evidence assembled in this article supports a clear verdict, and it is the one stated in the brief that commissioned it. The 1986 act’s rate schedule was dismantled within seven years while its base broadening largely endured, which means the reform that is remembered for cutting rates is better described as the reform that permanently widened the base. The base outlived the rates.

That sentence is worth sitting with, because it inverts the popular memory of the statute. The popular memory holds a triumph of rate cutting: the top rate fell from 50 percent to 28 percent, the schedule collapsed from fourteen brackets to two, and the code was made simple and fair. The measured record holds something different and, in the end, more impressive. The passive loss rules ended the mass-market shelter industry and were never repealed. The investment credit repeal, the depreciation lengthening, the interest deduction phaseout, the sales tax deduction repeal, all survived. The elasticity literature the reform generated disciplined a generation of revenue estimates. The pass-through shift reorganized American business. And the rate schedule, the part everyone remembers, was gone within seven years, replaced first by a 31 percent bracket and then by 36 and 39.6 percent, rates higher than the transitional schedule of the reform’s own first year.

Assessing a statute against its own aims, the series thesis that frames this article, produces a mixed but clarifying scorecard. On fairness, the act succeeded structurally: the shelter industry that had let high-income filers escape the schedule was destroyed, and the base that remained was broader and more uniform. On efficiency, the act succeeded in the sense the research can measure: capital allocation improved as tax-motivated distortions fell, even though aggregate labor supply and saving showed little measurable response. On simplicity, the act failed by the best available measure: compliance costs did not fall, because complex anti-abuse rules replaced the simple preferences they displaced. On revenue neutrality, the act did what it promised within the budget window, and the later deficits belong to later legislation and to the economy, not to this statute.

For readers working through the provisions behind these verdicts, the companion on the complete 1986 statute sets out the text that was evaluated here, and a legislation study notebook is available for keeping the statutes, dates, and estimates in this article organized against the record. The habit the article has tried to model is the one the evidence demands: separate what survived from what was reversed, name the author and period behind every estimate, distinguish real behavioral response from reclassification, and never let the headline provisions stand in for the structural ones. Judged that way, the 1986 act was not the rate-cutting triumph of popular memory. It was something rarer: a base-broadening achievement that outlived the rates it was enacted to buy.

The series thesis that frames this article, assessing a statute against its own aims and distinguishing structural change from headline change, proves its value here precisely because the two kinds of change point in opposite directions. The headline change, the rate cuts, failed the durability test. The structural change, the base broadening, passed it. A reader who judges the reform by its headlines will conclude that it was tried and reversed. A reader who judges it by its structure will conclude that it permanently altered the American tax system. Both readers are looking at the same statute. The difference is what they have chosen to measure, and the One Test that opened this article was designed to force the second choice: to name what survived, to date what was reversed, to cite the research on the responses, and to explain why the reference status endures.

There is a final sense in which the reform’s achievement exceeds its reputation. The 1986 act is often described as a bipartisan miracle, a product of unusual comity that could not be repeated. The description misses the harder truth, which is that the act was not a miracle at all but a bargain, and bargains can be studied, understood, and adapted. The revenue-neutral constraint, the coalition that traded rate cuts for preference repeal, the phase-ins that defused transition resistance, the regular-order legislating through the tax-writing committees, these were techniques, not wonders. Later reform efforts have succeeded or failed in proportion to their command of those techniques, and the failures are usually traceable to the abandonment of one of them: the rate cuts without the base broadening, the base broadening without the coalition, the coalition without the constraint. The 1986 act remains the reference case not because it was perfect, but because it was complete. It showed what a tax reform contains, and in doing so it set the terms on which every subsequent reform would be judged.

The ledger also carries a practical lesson for anyone designing the next reform, and it is worth stating directly. Durability in tax legislation comes from structure, not from salience. The provisions of the 1986 act that survived were the ones that changed the architecture of the system: the loss limitations, the repealed credits, the lengthened depreciation lives, the narrowed preferences. The provisions that failed were the ones that changed only the parameters: the rates, the brackets, the surcharge that held the two-rate shape together. Parameters are easy to legislate and easy to reverse, because every future Congress faces the same fiscal pressures and the same temptation to adjust the most visible dials. Architecture is hard to legislate and hard to reverse, because it rewires the incentives of organized constituencies and creates new facts, like the pass-through sector, that later Congresses must take as given.

That lesson cuts against the instincts of reform politics, which reward the visible and discount the structural. Rate cuts can be announced, celebrated, and credited. Base broadening is a long list of repeals, each with its own losers and none with its own parade. The 1986 coalition held together because the two were bound in a single revenue-neutral package, so the visible gains and the invisible costs moved together. When later Congresses unbound them, keeping the base and raising the rates, they confirmed the lesson rather than refuting it. The structure endured because it was structural. Anyone who wants the next reform to last should design for the ledger, not for the headline, and should expect the headline to be the first thing a future Congress revises.

Frequently Asked Questions

Q: Did the Tax Reform Act simplify the tax code?

By the measure economists use, no. Joel Slemrod, writing in the Journal of Economic Perspectives in the winter of 1992, found that the compliance cost of the income tax system was probably higher in the early 1990s than it had been in the early 1980s, and concluded that the act achieved little, if any, simplification. The paradox is that the statute did eliminate dozens of preferences and collapse fourteen statutory rates into two. But the anti-abuse rules that replaced the preferences, the passive loss provisions, the expanded alternative minimum tax, the tightened at-risk and interest deduction limits, imposed their own intricate recordkeeping burdens. The code had fewer lines but harder lines. Slemrod allowed that the reform may have dampened what would otherwise have been a steeper rise in complexity, but that counterfactual cannot be measured. What can be measured is the compliance burden, and it did not fall.

Q: Did the Tax Reform Act end tax shelters?

It ended the mass-market retail shelter industry, which was the industry that mattered. Section 469, the passive activity loss provision, suspended passive losses against wage, salary, and portfolio income, removing the engine that had driven limited partnership offerings marketed to high-income professionals. Combined with the investment tax credit repeal and lengthened depreciation schedules, the new rules made tax-motivated investing uneconomic, and the retail shelter market collapsed within a few years. Section 469 has never been repealed, and no statute has restored the pre-1986 treatment of passive losses. Two qualifications apply. The original act included a 25,000 dollar rental real estate offset for lower-income taxpayers, and the 1993 reconciliation act added a real estate professional exception. Neither restored the shelter business.

Q: What happened to tax rates after the Tax Reform Act?

The two-rate schedule did not survive the decade. The Omnibus Budget Reconciliation Act of 1990, Public Law 101-508, eliminated the 33 percent bubble and added a 31 percent bracket, effective for tax years beginning in 1991. The Omnibus Budget Reconciliation Act of 1993, Public Law 103-66, added 36 percent and 39.6 percent brackets, retroactive to January 1, 1993, and made the Pease limitation and personal exemption phaseout permanent. The top individual rate therefore moved from 28 percent in the reform years to 31 percent in 1991 and 1992 and to 39.6 percent from 1993 onward, above even the 38.5 percent transitional top rate of 1987. The corporate rate cut from 46 to 34 percent survived both reconciliation acts. The capital gains preference the reform had abolished was partially restored within four years.

Q: Did the Tax Reform Act grow the economy?

The research record says no, at least not in any amount the data can distinguish from other forces. Alan Auerbach and Joel Slemrod, surveying the evidence in the Journal of Economic Literature in June 1997, found that the act’s primary measured impact was a shift in the composition of reported income rather than a change in its level, and that aggregate labor supply and saving showed little measurable response. Henry Aaron, reviewing the evidence for Brookings, found little apparent effect, and less than expected, on broad economic aggregates. The finding is consistent with the later, lower elasticity estimates: if rate cuts had generated large real responses, the aggregates should have shown them. The confounders are substantial, including the late-1980s expansion, the October 1987 crash, lingering 1981 tax cut effects, and the 1990 and 1993 reversals. The act was scored as revenue-neutral and behaved that way.

Q: How long did the Tax Reform Act rate structure last?

The two-rate 15 and 28 percent structure took effect for tax years beginning in 1988 and survived in its enacted form only through tax year 1990, less than five years. A transitional five-bracket schedule from 11 to 38.5 percent had covered 1987. The 1990 reconciliation act added a 31 percent bracket for 1991, and the 1993 reconciliation act added 36 and 39.6 percent brackets retroactive to the start of 1993. Within seven years of full implementation, the top individual rate stood at 39.6 percent, higher than the rate the reform had replaced in the public mind. The 33 percent bubble, the surcharge mechanism that had preserved the two-rate shape, was eliminated by the 1990 act. The base broadening that had paid for the low rates, by contrast, largely endured, which is why the reform is better described as a base-broadening achievement than as a rate-cutting one.

Q: What are passive activity loss rules from the Tax Reform Act?

Section 469 of the Internal Revenue Code, added by the 1986 act, provides that losses from passive activities, generally trades, businesses, or rental activities in which the taxpayer does not materially participate, cannot be deducted against nonpassive income such as wages, salaries, or portfolio income. Suspended losses carry forward and may offset future passive income or be allowed in full when the taxpayer disposes of the entire interest. The provision attacked the mechanism of the retail tax shelter industry, which had sold high-income professionals paper losses to shelter their salaries, rather than any single investment. Congress phased the rules in over five years for pre-enactment interests, reaching full application in 1991. The original act included a 25,000 dollar offset for rental real estate available below set income thresholds, and the 1993 reconciliation act added an exception for qualifying real estate professionals. The core rule has never been repealed.

Q: Did the Tax Reform Act hurt real estate?

It hurt tax-motivated real estate investment by design, and the rental construction sector contracted sharply. The reform simultaneously lengthened depreciation lives, repealed the investment tax credit, trapped paper losses under the passive activity rules, and reduced the value of remaining deductions through lower marginal rates. James Poterba, in a 1992 National Bureau of Economic Research study, connected the reduced rental housing incentives to the collapse in new multifamily starts from 500,000 units a year in 1985 to under 150,000 by 1991, and found that the mortgage interest deduction had become concentrated at the top, with over half its tax losses accruing to the top 8 percent of taxpayers in 1988. On the owner-occupied side, lower rates reduced the tax wedge between owning and renting, a genuine efficiency gain. Contemporaneous forecasts of large rent increases proved gloomier than the outcomes, and the savings-and-loan crisis confounds any clean attribution.

Q: Is the Tax Reform Act still considered a model?

Economists treat it as the reference case for tax reform, which is a narrower and more durable status than popular admiration. The act generated the natural experiment behind the elasticity of taxable income literature that grew up around the reform, and the consensus range that emerged has anchored revenue estimation across the profession since. It demonstrated that base broadening is politically possible under divided government through the regular legislative order, a proof every subsequent reform effort has had to reckon with. It also supplied a negative lesson: the rate schedule was dismantled within seven years while the base broadening endured, warning reform designers that visible rate cuts are politically exposed while structural base changes are comparatively protected. The pass-through shift the rate inversion triggered permanently reorganized American business taxation. Its rates are gone; the consequences the research documented kept shaping how tax legislation was designed and scored through the decades that followed.

Q: What was the 33 percent bubble rate in the 1986 tax reform?

The bubble was an effective 33 percent marginal rate created by a 5 percent surcharge layered over the advertised 15 and 28 percent brackets. The surcharge phased in across a middle income range and clawed back the benefit of the 15 percent bracket and the personal exemption for higher earners, so filers inside the phase-out band paid 33 percent on each additional dollar, a higher marginal rate than filers above the band paid. The mechanism existed to preserve the two-rate shape while limiting its benefits, which meant the celebrated two-bracket schedule was, for upper-middle-income filers, a three-rate schedule with its highest rate in the middle. The bubble damaged the act’s simplicity claim, since nothing on the face of the two rates explained it, and it did quiet distributional work by phasing out the low bracket’s advantage. The Omnibus Budget Reconciliation Act of 1990 eliminated the surcharge, replacing the bubble with a straightforward 31 percent third bracket.

Q: What was the 1987 transitional tax schedule under the Tax Reform Act?

For tax year 1987 only, the act imposed a five-bracket transitional schedule running from 11 percent at the bottom to 38.5 percent at the top, bridging the 50 percent top rate of 1986 and the 15 and 28 percent two-rate schedule that took effect for tax years beginning in 1988. The transitional year mattered for measurement: taxpayers filed under five brackets while planning under the two rates arriving the next year, which produced a wave of retiming that researchers spent a decade disentangling. Income was accelerated into 1986 ahead of the capital gains exclusion repeal, and deductions were deferred into the lower-rate years. Anyone describing the 28 percent top rate as the law of 1987 misdates the central provision by a full year. The transitional schedule also sets up the irony of the later reversals: by 1993 the top rate of 39.6 percent stood above the transitional 38.5 percent of 1987 itself.

Q: What did the Joint Committee on Taxation’s revenue-neutral score of the 1986 act actually mean?

It meant the package was scored as approximately revenue-neutral over the five-year budget window under the static conventions then standard: the Joint Committee compared projected receipts under the proposed law with projected receipts under then-current law, holding taxpayer behavior largely fixed for the mechanical calculation, without adding any feedback loop for growth the reform might induce. The verdict described the legislation’s arithmetic, not a forecast of the deficit, which moved on spending, monetary conditions, and the business cycle. The five-year window shaped the statute’s architecture, since phase-ins pushed some base-broadening revenue beyond the scoring horizon while rate cuts lost revenue immediately, giving drafters an incentive to back-load broadeners and front-load cuts. The neutrality claim should always be cited with its conventions attached. It was the political keystone that converted rate-cut advocates into preference repealers, but it bound only the enacting Congress. Later Congresses were free to keep the broadened base and raise the rates, which is what the 1990 and 1993 reconciliation acts did.

Q: How did suspended passive losses carry forward under section 469 of the Tax Reform Act?

Section 469 provides that passive losses disallowed in a given year are not lost but suspended and carried forward indefinitely. In each future year the suspended losses may offset passive income from any passive activity, and they are released in full when the taxpayer disposes of the entire interest in the activity in a fully taxable transaction. The quarantine was the point: a loss that can only offset income from the same losing activity has no market, which is why the shelter industry collapsed rather than adapted. For pre-enactment interests Congress supplied a five-year glide path instead of a cliff, disallowing 35 percent of passive losses in 1987, 60 percent in 1988, 80 percent in 1989, 90 percent in 1990, and the full amount from 1991 onward, giving every investor a visible countdown. The original act also preserved a 25,000 dollar rental real estate offset for taxpayers below set income thresholds, and the 1993 reconciliation act added the real estate professional exception. The carryforward machinery itself has never been repealed.

Q: What was the Navratil critique of the early elasticity estimates?

Jeffrey Navratil’s 1995 critique showed that Martin Feldstein’s 1995 estimates of the elasticity of taxable income from the 1986 reform were biased upward by a control-group problem. Feldstein compared high-income filers, who received the largest rate cuts, with other groups, but the taxpayers with the largest cuts were systematically different from everyone else in income level, income composition, and access to shifting opportunities, and top-bracket incomes were already rising through the 1980s for reasons unrelated to marginal rates, including widening skill premiums and changing compensation practices. A filer whose 1985 income was temporarily depressed would show rapid growth to 1988 through mean reversion alone, and that growth would be misattributed to the rate cut. No clean group of similar taxpayers who did not receive a cut existed against which to measure the treated group. The survey by Saez, Slemrod, and Giertz in 2012 treated the Navratil critique as decisive in concluding Feldstein’s 1986-based estimates were biased upward, and it is the mechanism behind the literature’s move from estimates near 1.0 to the consensus range of 0.12 to 0.40.

Q: What caused the swing from negative 2.2 billion to positive 32 billion dollars in reported business income after 1986?

Roger Gordon and Jeffrey MacKie-Mason documented in 1990 that net income from partnerships and S corporations reported on individual returns averaged negative 2.2 billion dollars a year from 1981 through 1986 and swung to positive 32 billion dollars in 1987. The negative sign was the signature of the shelter era: these entities existed largely to generate paper losses for high-bracket owners. The swing combined two effects of the reform pulling in the same direction. First, the passive loss rules shut down the manufacture of artificial losses, so previously netted losses stopped appearing. Second, the rate inversion, with the top individual rate at 28 percent below the 34 percent corporate rate, pulled genuine business income into individual form through S elections and partnership reorganizations. Approximately 375,000 firms filed S elections in the first half of 1987, against a semiannual average near 150,000 in the years 1983 through 1986. The figure measures reported income, not new economic activity: some of it was reclassification, and some was the end of loss manufacturing. But the direction is unambiguous evidence that the reform simultaneously destroyed the shelter economy and repriced the choice of business form.

Q: How did the Tax Reform Act of 1986 change the earned income tax credit?

The act raised the credit from 11 percent to 14 percent of the first 5,714 dollars of earnings, increasing both the credit’s value and the number of workers who qualified. The expansion was part of the reform’s deliberate base narrowing at the bottom of the income scale, financed the same way the rate cuts were, from the base broadening above. Together with the personal exemption’s rise from 1,080 dollars toward 2,000 dollars and the standard deduction’s increase to 5,000 dollars for joint filers, the larger credit lifted millions of low-income households off the income tax rolls entirely; secondary accounts drawing on Congressional Research Service figures put the number near six million. The credit kept its expanded form and became the foundation for later growth under subsequent legislation, which expanded it further rather than reversing the 1986 change. In the distributional barbell the reform produced, the credit expansion was the clearest gain at the bottom end: households that had never used the shelters the act destroyed received relief without surrendering preferences they had never held.

Q: What was the Pease limitation on itemized deductions?

The Pease limitation, enacted in the Omnibus Budget Reconciliation Act of 1990 and made permanent by the 1993 act, reduced itemized deductions for higher-income filers, applying above 100,000 dollars of adjusted gross income. It was a deficit-reduction instrument layered onto the 1986 base: rather than reopening the preference fights of 1986, Congress raised revenue by trimming deductions at the top. The provision moved the system away from the 1986 design in a progressive direction, and because it applied to the broadened base the reform had left behind, it raised revenue more efficiently than the same limitation would have on the old narrow base. Pease belonged to the deficit politics of its own era, not to the reform’s design, but its interaction with the 1986 base illustrates the ratchet the reform built: the base stayed broad, and later Congresses extracted revenue by adjusting parameters on top of it. Making the limitation permanent in 1993 locked in a new base-broadening overlay that the 1986 act itself had not contained.

Q: What was the personal exemption phaseout added after the Tax Reform Act?

The personal exemption phaseout, known as PEP, was enacted in the Omnibus Budget Reconciliation Act of 1990 and made permanent by the 1993 act. It phased out personal exemptions for higher-income filers, applying to joint filers with income between 150,000 and 275,000 dollars. Like the Pease limitation enacted in the same statute, PEP was a deficit-reduction measure that raised effective marginal rates at the top without changing the statutory brackets, moving the system further from the 1986 design. The irony is structural: the 1986 act had nearly doubled the exemption, from 1,080 dollars in 1986 to 2,000 dollars by 1989, as part of its base narrowing at the bottom, and within a few years later Congresses began withdrawing that enlarged exemption at the top. Both provisions illustrate the asymmetry this article’s ledger documents. The 1986 base changes were structural and survived; the parameters layered on afterward were negotiable in every budget cycle. PEP and Pease together formed the upper-income overlay that the 1990 and 1993 acts placed on the 1986 foundation.

Q: How did the Tax Reform Act change the personal exemption?

The act raised the personal exemption from 1,080 dollars in 1986 to 1,900 dollars for 1987, then to 1,950 dollars in 1988 and 2,000 dollars in 1989, a near-doubling in three years that removed income from the base before any rate applied. Because the exemption applied per person, the increase delivered larger absolute benefits to larger families: a married couple with three children saw 10,000 dollars of income removed from the base by the exemption alone at the 1989 level, before the standard deduction applied. The provision carried a family policy inside the rate reform, and it explains part of the roughly six million low-income taxpayers removed from the rolls in Congressional Research Service-derived accounts, since the households leaving the rolls were disproportionately families with children whose modest earnings the enlarged tax-free threshold covered entirely. Later Congresses preserved the enlarged exemption and extended its family-policy logic through the child credit rather than through the exemption itself, while the 1990 and 1993 acts added a phaseout withdrawing it at higher incomes.

Q: How did the Tax Reform Act expand the alternative minimum tax?

The 1986 act broadened the individual alternative minimum tax substantially, expanding the list of adjustments and preferences caught by the parallel system at the same time the regular tax was being cleaned of shelters. The logic was insurance: a code that had closed the most popular shelters needed a backstop for the ones it had missed and the ones promoters would invent next, a parallel computation that added back preferences and applied a flat rate so high-income filers could not reduce liability to zero. The individual AMT rate was set at 21 percent, and the 1990 reconciliation act later raised it to 24 percent rather than narrowing the base, widening the backstop instead of dismantling it. The minimum tax endured because its function was independent of the rate debate: brackets could be raised and lowered without affecting the need for a floor under preference-heavy filers. Over time the unindexed exemption pulled in filers below the high-income population it was designed to catch, a parameter controversy, but the structure survived. Like the passive loss rules, the broadened AMT belongs in the ledger’s durable column.

Q: How did the Tax Cuts and Jobs Act pass-through deduction respond to the 1986 organizational shift?

By 2017 the pass-through sector was so large a share of American business income, the product of the migration the 1986 rate inversion set in motion, that any major tax legislation had to address it, and the Tax Cuts and Jobs Act answered with a dedicated pass-through deduction. The connection is structural rather than incidental. The 1986 act cut the top individual rate to 28 percent while the corporate rate fell only to 34 percent, making individual form cheaper than corporate form, and owners responded with roughly 375,000 S elections in the first half of 1987 alone. The migration continued for decades after the inversion was reversed, because reorganizing has fixed costs, the professional infrastructure built around pass-through form made staying cheap, and later developments such as limited liability company statutes reinforced it. The 2017 provision therefore legislated for a pass-through economy the 1986 act had created, responding to the world the earlier reform left behind rather than reversing it. No finding here extends to any later proposal; the episode is presented as history.