# 6. The consistency objection: housing and equity

;; 6.01
Municipal bonds are not special, and this section takes up the objection that follows from that fact. Owner-occupied housing, retirement accounts, pass-through structures, life insurance — every preference that is capitalized into an asset's price drives the same wedge between remitted and borne. Correcting one asset held at the top of the distribution, and stopping there, biases the overall picture toward more measured progressivity. The objection is correct, and the paper's answer is not to dispute it.

;; 6.02
The answer is a division by how much the market tells us, and the division is one of degree. Municipal bonds are the asset class where market prices come closest to identifying the implicit tax, because a tax-exempt bond and a taxable bond trade side by side and their yield ratio is observable every day. That observability is necessary and not sufficient: Section 3 showed the raw ratio is contaminated, the clean design is confined to a historical short-end market, and carrying its estimate to 2021 and across maturities is an assumption the Table 4 scenarios expose rather than resolve. Owner-occupied housing has no taxable twin trading beside it at all; corporate equity has a marginal investor who may not be taxable at all. So the paper corrects the asset the market speaks to, sizes the two large neighbors with explicit arithmetic, and states the direction each pushes the headline. The arithmetic below is quantitative rather than rhetorical, but it is a scale illustration under stated assumptions — not a bound, and not a correction: neither exercise brackets a true value, and this section does not produce a percentile-level schedule for either asset.

## 6.1 Housing

;; 6.03
The housing correction has a known direction, which is less than a known size. The preference bundle — untaxed imputed rent, the mortgage interest deduction, deductible property taxes — is capitalized into house prices: Poterba (1984) is the asset-market treatment, the user-cost framework of Himmelberg, Mayer, and Sinai (2005) carries it, and Sommer and Sullivan (2018) find in general equilibrium that repealing the mortgage interest deduction lowers house prices by about 4.2 percent. Capitalization means owner-occupants earn a depressed implicit return on housing as bondholders earn a depressed yield on munis. The analogy stops short of an equivalence, and the gap matters for what can be computed: capitalization into the purchase price supports a lower housing-service yield at current market prices, but it does not by itself deliver an annual household-level rate correction, and it leaves the transitional gains and losses — which accrued to whoever owned when the preference was enacted or changed — unallocated across cohorts. Housing is held far further down the distribution: {{N:own_rate_p50_90_pct}} percent of households between the 50th and 90th percentiles own homes worth ${{N:housing_p50_90_tn}} trillion, against ${{N:muni_p50_90_tn}} trillion of municipal holdings in the same group.

;; 6.04
The magnitudes are what make the question live. Treasury scores the three housing preferences at ${{N:te_housing_total_bn:.0f}} billion a year for fiscal 2024 — the imputed-rent exclusion alone is among the largest individual income tax expenditures — against ${{N:te_muni_bn:.0f}} billion for public-purpose municipal interest: roughly eight to one. If a quarter of the housing preference were capitalized into depressed implicit returns, the implied middle-class implicit tax would be of the same order as the entire municipal correction, and it would land on the 50th to 90th percentiles rather than the top. State what that calculation is and is not. The one-quarter conversion is assumed, not derived; an annual tax-expenditure score is a revenue estimate rather than a measure of depressed annual returns, and the step from one to the other is exactly the capitalization model this paper does not build. So the arithmetic establishes that the housing margin is large enough to matter and that it runs the other way — not that it would offset most of the municipal correction, which is a claim this paper cannot support and does not make. One nuance disciplines it further: imputed rent is absent from the income base of the CBO, Treasury, and JCT frameworks altogether, so for them both the income and its implicit tax are missing and the bias runs through the denominator rather than the recorded rate; the symmetry argument as stated applies to the national-accounts-based frameworks that do include imputed rent. The direction — a correction landing in the middle of the distribution — survives under either treatment. A second nuance cuts the other way: the municipal wedge is collected by a government as borrowing-cost revenue, which is Section 2's reason an avoidable wedge counts as a tax; the housing wedge is capitalized into prices and accrued to whoever owned at enactment, with no government collecting it. The two corrections are symmetric as measurement of depressed pre-tax returns, not as taxation — which weakens the consistency objection's force even as the paper concedes its direction.{{FN: Poterba and Sinai (2008) give the distributional texture within owner-occupancy: user costs of 5.0 percent for the highest-income owners against 6.9 percent below $40,000, and mortgage-interest tax savings that rise steeply with income. The explicit expenditures favor the top; the capitalized implicit tax reaches the middle.}}

## 6.2 Equity

;; 6.05
A benchmark arithmetic poses the equity question. Suppose dividend and capital-gains preferences depress pre-tax equity returns by twenty basis points. On ${{N:z1_domestic_equity_tn}} trillion of domestic corporate equity that is ${{N:equity_20bp_all_bn}} billion a year, several times the municipal numbers. The arithmetic is right; the two premises it needs are not established. The base matters by a factor of nearly four: only 27 to 28 percent of United States corporate equity sits in taxable accounts, with roughly 32 percent of publicly traded equity held by foreigners and 34 percent in retirement accounts (Rosenthal and Mucciolo 2024). And the existence of any aggregate wedge at all depends on who the marginal investor is.

;; 6.06
That dependence is the marginal-investor problem, and it is fatal to a point estimate rather than merely attenuating. If the price-setting investor in the aggregate equity market is a pension fund or a foreign institution, individual-level dividend and capital-gains preferences are not capitalized into prices at all: the implicit tax on equity is zero, and the preferences are pure windfall to taxable holders. Whether that is the case is exactly what the empirical literature splits on.

;; 6.07
The split is real and the paper preserves it. Sialm (2009) estimates an aggregate effective tax yield of about 0.4 percent of equity value in 2006 — roughly forty basis points, and itself the low end of a series that has declined from over 3.5 percent in 1950 as rates fell and ownership migrated to untaxed accounts. Amromin, Harrison, and Sharpe (2008) find no aggregate imprint of the 2003 dividend tax cut on stock prices. Auerbach and Hassett (2005) find cross-sectional price responses to the same reform; Chetty and Saez (2005) find payout responses. The reference points stand at zero and roughly forty basis points, and we refuse to average them. The scale check a referee will run: forty basis points on the ${{N:taxable_equity_tn}} trillion of taxable-account equity is ${{N:equity_40bp_taxable_bn}} billion a year — larger than the entire municipal correction. Note what the upper figure is. An effective tax yield measures tax paid relative to equity value; it does not identify how much of that tax is capitalized into equilibrium pre-tax returns, and the relative-price evidence below does not identify that level either. Forty basis points is therefore a conditional scenario, not a demonstrated ceiling — the true aggregate wedge could sit above it as easily as between it and zero. An equity margin of this scale, with its sign unsettled and its magnitude unbounded in either direction, can swallow the headline, which is precisely why the paper's quantitative claims are munis-only. Two questions sharpen the range, and the next paragraphs answer them in turn: does modern ex-dividend-day evidence show dividend taxes reaching prices at all, and do partnership structures show the capital-gains preference being priced?

;; 6.08
The ex-dividend-day literature is the sharpest lens on whether dividend taxes reach prices, and its modern arc answers the first question. Elton and Gruber (1970) made ex-day pricing a tax thermometer: on 1966-67 data the ex-day price drop implied a marginal stockholder bracket of 36.4 percent, matching tax-return-based estimates, with clientele patterns by yield. The modern reforms supplied the experiments, and they produced both results at once. Zhang, Farrell, and Brown (2008) find the mean price-drop ratio rose from 0.65 before the cut to 0.86 after, significant at the 1 percent level, with tax-motivated ex-day volume receding. Bell and Jenkinson (2002) find the United Kingdom's 1997 removal of pension funds' dividend credits moved the valuation of high-yield companies — identifying pension funds, not top-bracket individuals, as the marginal investor in that market.

;; 6.09
The same literature explains why those results cannot size an aggregate wedge. Chetty, Rosenberg, and Saez (2005) show the 2003 comparison fails placebo tests: the relation between tax rates and ex-day premia over 1962-2004 is so noisy that responses cannot be credibly detected even around large tax reforms. Hartzmark and Solomon (2019) document non-tax demand for dividends that contaminates ex-day pricing. The answer to the first question, then: modern ex-dividend-day studies show dividend taxes in relative prices where taxable traders are pivotal, and they neither reject the zero reference point nor the forty-basis-point one for the aggregate. The two stay where they are, unranked.

;; 6.10
Partnership structures carry the capital-gains preference, and they answer the second question in the affirmative, with a caveat that leaves the aggregate as unsettled as before. Individual capital-gains taxes demonstrably move equity prices: in the week of the May 1997 budget accord that cut the top rate, non-dividend-paying stocks returned about 12 percent against 6 percent for payers (Lang and Shackelford 2000), and Dai, Maydew, Shackelford, and Zhang (2008) identify both capitalization and lock-in effects around the same reform, sorted by taxable ownership. Where a partnership structure packages a tax advantage for an identifiable taxable clientele, the advantage is priced at close to its arithmetic value. Sinai and Gyourko (2004) find the 1997 rate cut shrank the UPREIT partnership deferral subsidy: acquisitive UPREITs fell roughly 14 percent relative to non-acquisitive UPREITs, in a design benchmarked against regular REITs. In 2021, Magellan Midstream's board publicly priced the present value of corporate taxes from converting its master limited partnership at about $2.3 billion and found no offsetting valuation premium from corporate form: the pass-through shield is in the price. Congress restricted publicly traded partnerships in 1987 precisely because the form eroded the corporate tax base (Congressional Research Service 2011).

;; 6.11
The caveat: every clean partnership result is a relative price — UPREIT against REIT, MLP against C-corporation, high-yield against low-yield stock. Relative-price capitalization confirms the mechanism this paper applies to municipal bonds: where the taxable investor is pivotal, her taxes are in the price. It does not identify the level of aggregate pre-tax C-corporation returns against a no-tax counterfactual, which is what an economy-wide equity correction needs. The margin is quantitatively first-order — pass-throughs now generate over half of United States business income, far more concentrated at the top than C-corporation income (Cooper et al. 2016) — so the question matters; it is simply not answerable from relative prices. The answer to the second question: partnership evidence moves the plausibility of equity capitalization up, and the aggregate wedge remains unidentified between the corners.

;; 6.12
What a full treatment would require, stated once: for housing, a capitalization model joining imputed rent, deductions, and prices; for equity, a marginal-investor identification that does not currently exist; for retirement and insurance vehicles, deferral accounting. Section 8 assigns them to the research design. This paper does one asset from market prices, and for the other two states the direction, the scale, and the assumption each figure rests on.
