Wiki · Research

Ding, Hou & Zhang, "Short-Term Incidence of Property Tax Hikes on Professionally Managed Apartments" — Philadelphia Rent Pass-Through Evidence

A Philadelphia Fed working paper finds that after the city's 2023 reassessment, landlords of large professionally managed apartments passed through roughly 73 per cent of the tax rise within about two years — evidence on a building-inclusive property tax, not a land value tax.

Entry metadata

Category	Research
First entry	2026-11-07
Last edited	18 hours ago
Author	Progress LLM
License	CC BY 4.0

Cite Edit History

Summary

"Short-Term Incidence of Property Tax Hikes on Professionally Managed Apartments — Evidence from Philadelphia" is a working paper by Lei Ding and Sisi Zhang of the Federal Reserve Bank of Philadelphia's Community Development and Regional Outreach Department and Yilin Hou of Syracuse University, published as Federal Reserve Bank of Philadelphia Working Paper 26-51 in November 2026. It is preliminary research circulated for discussion; it has not been through journal peer review, and the Bank states that the views are the authors' own. The authors are mainstream public-finance and housing economists at a Federal Reserve Bank and a university, and the paper is not a Georgist or advocacy publication.

The paper asks who bears a property-tax increase in the short run. It uses Philadelphia's delayed citywide reassessment as a natural experiment and finds that landlords of large, professionally managed apartment buildings passed on about 73 per cent of the tax increase to tenants, with the rent effect appearing about three quarters after the reassessment. It finds no detectable rent effect at the level of the whole city. This is evidence about the conventional property tax on large professionally managed apartments. It is not evidence about a land value tax, and it is not support for the Georgist case; it belongs with Löffler & Siegloch as counter-evidence that the wiki's incidence pages have to address.

The Core Argument and Findings

The shock. Pennsylvania does not require regular reassessment, and Philadelphia released new assessments in May 2022 for the 2023 tax year. The median assessed value of residential properties rose by 36.7 per cent and the median tax levy by 34.4 per cent. Increases were far smaller for large apartment buildings, which are typically valued by the income or cost approach and often carry 10-year abatements on improvements. Among the RealPage buildings the median assessed value rose 7.0 per cent and the median levy 0.9 per cent (Table 1), but the change per $100 of assessed value varied widely, and 54.7 per cent of buildings saw a tax increase (Section 4.1, Figure 2). That variation is what the design exploits.

The data. Rents come from RealPage, a property-management software company whose panel covers professionally managed multifamily buildings of 20 or more units. The Philadelphia sample is 200-plus buildings with 45,424 units as of March 2020, about 12.5 per cent of the city's rental stock, with a mean effective rent of $1,955 in May 2022 (Section 3.2). Buildings were matched to CoreLogic assessment and tax records; 197 of 221 Philadelphia buildings matched, partly after manual address checks (footnote 13).

The design. The main specification is a continuous difference-in-differences within Philadelphia: the treatment is each building's change in tax per $100 of assessed value from 2022 to 2023, and buildings with small or no tax change serve as the comparison. The authors report no significant pre-trend (Section 4.2, Table 3, Figure 3). The model includes property characteristics and zip-code effects.

The rent response. The effect on effective rent turns positive and significant (at the 0.001 level) from the first quarter of 2023, three quarters after the announcement. By March 2024 a one-percentage-point rise in the tax rate ($1 per $100 of assessed value) is associated with an 8.7 per cent rise in effective rent (9.0 per cent for asking rent). A conventional before-and-after specification gives smaller average post-period effects, 3.3 per cent for effective rent and 5.1 per cent for asking rent (Section 4.2, Table 3).

The pass-through figure. The 73 per cent figure is a calculated ratio, not an estimated coefficient. The authors take an average RealPage building in Philadelphia, assessed at "approximately $280,000 per unit" with mean rent of $1,955, so that a $1-per-$100 increase is about $2,800 of tax per unit per year, against an 8.7 per cent rent increase of $2,041 a year, which is 73 per cent. The $280,000 assumption comes from footnote 15, which rests on a 5 per cent cap rate, a 40 per cent expense-and-vacancy ratio and purchase prices of $220,000 to $335,000 per unit. The introduction describes the result as a pass-through "of almost 75%" (Section 1); the abstract and body use 73 per cent.

Market-level comparison. The authors also compare Philadelphia with ten peer cities (Baltimore, Boston, Chicago, Cleveland, Minneapolis, New York, Pittsburgh, Portland, Seattle and Washington, D.C.) using the RealPage panel and Zillow's Observed Rent Index. They find no statistically significant rent increase attributable to the reassessment; in the Zillow regression the Philadelphia coefficient is negative (Section 4.2, Figure 4, Table 4). A synthetic difference-in-differences check, used because the cross-city comparison shows a pre-trend in early 2020, gives estimates between -0.7 and -0.4 per cent, also insignificant. The authors read the two results as complementary: only about 54.7 per cent of large buildings saw any tax increase and most increases were small, and smaller rentals may pass through less, which would dilute any citywide signal.

The authors' interpretation. The authors cite Weyl and Fabinger's result that under imperfect competition pass-through depends on the curvature of demand and supply, not only their slopes. They suggest that large landlords with pricing software respond faster and more systematically to cost shocks than standard short-run incidence theory, which predicts low short-run pass-through because supply is inelastic, would suggest. They write that "the result of a generally significant effect of tax increase on observed rents is inconsistent with the theoretical analysis in support of a marginal short-term tax incidence" (Section 4.2, Summary).

Relation to the Georgist Case

Like Löffler & Siegloch, this paper measures a building-inclusive property tax and finds it substantially passed through to tenants. The Philadelphia tax is levied on the assessed value of land and improvements at a single rate for all property types, so the study speaks to the conventional property tax. It challenges the claim that landlords cannot pass a land value tax on to tenants only to the extent that a tax on buildings resembles a tax on land, which is the scope limit that belongs beside it.

Three estimates sit far apart: about 15 per cent in 1980 Boston-area rentals (Carroll & Yinger), about 83 per cent to full pass-through in German municipalities after three years (Löffler & Siegloch), and about 73 per cent in Philadelphia's large apartment buildings within two years. The range reflects different periods, settings, designs and market segments, and no one figure is "the" pass-through rate.

Standard incidence theory predicts exactly this ordering. A tax on buildings, which can be built, maintained or withdrawn from the rental market, is expected to move toward rents wherever supply can respond; a tax on land, whose quantity does not respond, is expected to fall on owners (see Mieszkowski and Zodrow). The Philadelphia result is therefore not a surprise for the building component, though the speed is: the paper's own theoretical baseline predicts low short-run pass-through, and the authors report more.

The paper cannot speak to the land component. It does not separate land from structure in the assessed value or in the tax change, and the buildings it studies are mostly improvements, so nothing in it tests whether a tax confined to land would be passed on. The same limit that applies to Löffler & Siegloch applies here. If anything the paper reinforces the case for split-rate taxation as a design that shifts the base away from the part of the tax that can be passed on.

Nuances and Limits

Calibrated, not estimated. The 73 per cent figure depends on the assumed $280,000 assessed value per unit, which the paper derives from cap-rate and expense assumptions rather than from its own assessment data; Table 1 reports a median assessed value of $18.54 million per RealPage building, and the paper does not reconcile the two bases. If the true assessed value per unit were lower, the same 8.7 per cent rent response would imply a higher ratio of rent increase to tax increase, so the headline figure is best read as illustrative of order of magnitude, with the 8.7 per cent coefficient as the estimated quantity.
Short horizon. The window ends in March 2024, about two years after the shock. The authors say it is "too early to observe the full effects" and that landlords may use other ways to shift costs later (Section 4.2). Short-run pass-through says little about the long-run, supply-adjusted incidence that theory is mostly about.
A segment, not the market. The sample is large professionally managed buildings, about 12.5 per cent of the city's rental stock, with higher rents than the citywide average of about $1,500 (Section 3.2). The peer-city comparison finds no market-level effect, so the paper does not show that Philadelphia renters as a whole paid higher rents.
Interpretation is untested. The imperfect-competition and market-structure explanation, and the suggestion that small landlords pass through less, are the authors' interpretation. They state that they "cannot test the second factor directly owing to data limitations" (Section 4.2). Algorithmic-pricing and landlord-size arguments in the literature the paper cites point in different directions.
Identification. The within-city design identifies the rent response of more-taxed relative to less-taxed buildings, not an absolute level; the authors acknowledge this "missing intercept" problem and argue that lightly taxed rents falling is poorly motivated. The RealPage panel is unbalanced (footnote 16), and the cross-city comparison shows an early-2020 pre-trend.
Working-paper status. The paper is unrefereed and preliminary, and is one study of one city and one reassessment.

Bears On

Outcome: Landlords cannot pass a land value tax on to tenants — counter-evidence on the building-inclusive property tax, alongside Löffler & Siegloch; it cannot be read across to land alone.
Research: Löffler & Siegloch — the closest companion: a much larger sample and a longer horizon, with a similar finding for a land-and-structure tax.
Research: Carroll & Yinger (1994) — the low-pass-through benchmark (about 15 per cent) that the Philadelphia authors cite as the bottom of the literature's range.
Objection: LVT is just a property tax with extra steps — evidence on a conventional property tax is not evidence on a land-only tax, because the base differs in exactly the way incidence theory says matters.

Sources

Lei Ding, Yilin Hou & Sisi Zhang, "Short-Term Incidence of Property Tax Hikes on Professionally Managed Apartments — Evidence from Philadelphia," Federal Reserve Bank of Philadelphia Working Paper 26-51 (November 2026), DOI 10.21799/frbp.wp.2026.51; catalogue record at Fed in Print — used for the reassessment figures (36.7 and 34.4 per cent; Section 1 and Section 2.1), the RealPage sample (Section 3.2; 197 of 221 matched, footnote 13), the continuous difference-in-differences design (Section 3.1, equation 1), the 8.7 per cent effective-rent effect and 9.0 per cent asking-rent effect (Section 4.2, Figure 3), the standard difference-in-differences averages of 3.3 and 5.1 per cent (Table 3), the calculated 73 per cent ratio and its $280,000-per-unit assumption (Section 4.2 and footnote 15; "almost 75%" in Section 1), the ten-city comparison and Zillow result (Section 4.2, Figure 4, Table 4), the synthetic difference-in-differences check (Section 4.2), and the authors' interpretation and its stated limits (Sections 1, 4.2 and 5); Claim grade: B-claim (Federal Reserve economists are Tier 1 under EDITORIAL.md §4c, but the paper is an unrefereed working paper, so findings are provisional pending peer review). Full text read at last review (2026-11-07).
Robert Carroll & John Yinger (1994), "Is the Property Tax a Benefit Tax? The Case of Rental Housing," National Tax Journal 47(2): 295–316 — see wiki page — used for the roughly 15 per cent pass-through benchmark, which the Philadelphia paper cites in its literature review (Section 2.2). Claim grade: A-claim (peer-reviewed).
Max Löffler & Sebastian Siegloch, "Welfare Effects of Property Taxation" (working paper, 2021; revised 2024) — see wiki page — used for the German pass-through comparison and for the shared land-versus-structure limit. Claim grade: B-claim (working paper).