Taxing Homeowners in China: Who Are the Winners and Losers? (Cho, Park & Zhang, 2026)

Summary

Yunho Cho, Jinseong Park and Sisi Zhang, "Taxing homeowners in China: Who are the winners and losers?," Real Estate Economics (published online 17 September 2026), asks two questions about a recurrent tax on housing in China. The first is empirical: what did Shanghai's 2011 pilot, which targeted a narrow set of homes, do to the housing market? The second is a general-equilibrium accounting exercise: in a calibrated overlapping-generations model of owning and renting, who gains and who loses if the tax is universal or targeted, and the extra revenue is handed back to households as equal lump sums?

The published abstract reports "little impact on the overall housing market, with suggestive evidence of a reduction in the unit prices of larger homes" in Shanghai, and a model in which a universal tax lowers house prices and raises rents, leaves housing services less affordable, and produces a long-run welfare gain driven by redistribution and lower prices. It adds that "along the transition, 55% of existing households experience a welfare loss," with poorer households gaining and richer households losing (abstract, published version).

The magnitudes in this entry come from the July 2025 working-paper version (KDI School Working Paper 25-08), because the published full text was not available for reading. The two versions differ in places (see Nuances and Limits), so every figure below is labelled as working-paper. Page numbers ("WP p.") are those of the working-paper PDF; its printed page numbers run two lower.

The tax studied taxes land and buildings together. It is not a land value tax, and the paper does not separate the two components in its policy experiments.

The Core Argument and Findings

The Shanghai pilot and the reduced-form result (working-paper version)

Shanghai began collecting a tax on 28 January 2011. In the Lincoln Institute's description, it applied to "newly purchased second homes of residents and first homes of nonresidents based on transaction value, after the exclusion from the tax base of 60 square meters per person" (Man 2012, section "Major Achievements"). The working paper describes the same Shanghai pilot more loosely, as "a tax on second homes at a rate of 0.6%" (WP p. 9). The working paper does not define the base further, so the rate cannot be turned into an effective burden from that source alone. It describes Chongqing's pilot separately, with a 0.5% rate (1% above RMB 15,000 per square meter) on investment houses and homes over 140 square meters (WP p. 9); the Lincoln article describes Chongqing's targets differently, and the paper's empirical work covers Shanghai only because Chongqing lacks pre-treatment data (WP p. 11).

The paper applies a synthetic difference-in-differences design to city-quarter averages of new-home transactions, comparing Shanghai with Beijing, Tianjin, Ningbo, Guangzhou, Hangzhou and Wuhan (161 observations; data from the China Index Academy's real-estate index system) (WP pp. 11, 14). The estimated effect on log price ranges from -0.003 to -0.021, that is, a fall of 0.3% to 2.1% across three specifications, and on area per transaction from +0.002 to +0.010. The authors describe all of these as "statistically insignificant and mostly economically small" (WP p. 14, Table 2). The event-study estimates turn modestly negative over the first 18 months of the pilot and then rise, and none is significant (WP p. 15).

The working paper places this result beside two earlier Shanghai studies. Du & Zhang (2015) also find no price effect. Zheng & Zhang (2013) estimate a fall of 2,127 yuan per square metre, or 13.4%, and the authors attribute the gap to differences in research design and estimation period (WP p. 15).

The general-equilibrium model (working-paper version)

The model has overlapping generations of households who choose between renting and owning, face income shocks, and can borrow against housing subject to down-payment rules that differ for first and investment homes. It is calibrated to the 2015 wave of the China Household Finance Survey, which covers 37,289 households (WP pp. 12, 16 to 20). Housing supply is fixed in the main version, on the grounds that the government tightly controls residential land (WP p. 19). Both experiments use a 0.5% annual tax, chosen to resemble the pilot rates. The extra revenue is returned as equal lump-sum transfers, and baseline revenue is simply spent by government (WP pp. 18, 24). In the targeted experiment only investment housing is taxed; in the universal one every homeowner is taxed (WP p. 24).

(WP pp. 26, 27, 31.)

Three features of the results matter most.

1. Housing becomes less affordable even as prices fall. The universal tax lowers house prices but raises rents and adds an ongoing tax bill for owners. The authors' affordability measure (the expenditure needed to hold utility at its baseline level) deteriorates, and more so for higher-income households, who own more housing. Lump-sum transfers are large enough to offset the higher housing cost for low-income households under the universal tax, but not under the targeted tax (WP pp. 5, 28 to 29).
2. Redistribution drives the welfare gain. Newborn welfare under the universal tax decomposes into a direct effect of -1.29%, a redistribution effect of +1.75% and a general-equilibrium (price and rent) effect of +0.26%. The authors conclude that "implementing a universal property tax without redistributing the additional tax revenue to the public will, despite lowering house prices, reduce welfare in both the short and long run" (WP p. 37; decomposition WP p. 31).
3. The transition creates a majority of losers. For households alive when an unexpected, permanent universal tax arrives, average welfare falls by 0.13% and 58% suffer a welfare loss (about 42% gain). The authors give two reasons: roughly 90% of households own homes and so face a higher tax burden, and existing owners are hurt by the sharp initial fall in house prices, of which the paper reports 2.6% in the first period (WP pp. 5, 34, 43).

Who wins and who loses (working-paper version)

Under the universal tax, welfare falls steadily with wealth. Poorer households with little housing gain from the transfers and small tax bills; richer households lose from higher tax bills and lower housing values. Renters gain significantly because they pay no tax and receive transfers, older households lose more than younger ones, and owner-occupiers and landlords both lose, with landlords cushioned somewhat by higher rents (WP pp. 34 to 36). Under the targeted tax the pattern reverses: poorer and younger households, more of whom rent, lose more, because rents rise by 5.5% and the transfer is small (WP pp. 32 to 36). The paper also reports that a universal tax raises the gap between desired and actual housing consumption for owners (its "misallocation wedge," WP pp. 29 to 30).

Relation to the Georgist Case

The paper concerns a tax on land and structures together, so its bearing on the Georgist case is indirect. It is relevant in three ways, and the second and third are interpretation rather than findings of the paper.

1. A narrow pilot moved prices little. The Shanghai result is a rare quasi-experimental test of a recurrent housing tax that covered only some purchases. Its null price finding is consistent with the model's result that a tax confined to investment housing leaves prices roughly unchanged (+0.3%) while pushing rents up. The pilot's tax base was small: the Lincoln article describes the pilot's tax base, rate and collections as "all very small" in the first year (Man 2012), and in the model the targeted tax raises revenue equal to 2% of income-tax revenue. A weak price response from a design this limited says little about a broad-based tax, and the paper's own model predicts a much larger effect for a universal tax.
2. A general-equilibrium accounting of a tax that replaces nothing and is rebated. The decomposition separates the burden of the tax, the value of returning its revenue, and the price and rent adjustments. That is the accounting a Georgist proposal for a rent-based citizen's dividend or for cutting other taxes would also need to face. In the paper, the rebate is what turns a net loss into a net gain.
3. The distributional shape. Renters and poorer owners gain, and richer, older owners lose, because housing wealth is concentrated among the wealthy while transfers are equal per household. A land-based tax would plausibly share that shape, since land value is likewise held unevenly, though this paper does not test it. The transition loss to existing owners from falling prices parallels the one-time wealth shock discussed for land taxes, and the capitalization of a recurrent tax into prices is the mechanism at work in both.

The land/structure distinction cuts in a specific place. In the main model both land and the housing stock are fixed, so the tax falls on an asset in fixed supply. The appendix relaxes this by letting construction respond to prices (a supply elasticity of 3): the universal tax then lowers house prices by about 2.0% rather than 3.7%, housing supply falls by about 5.5%, and the fall in under-35 homeownership deepens to 10.1% (WP pp. 42 to 43). The authors present this as a robustness check, not as a land-versus-improvements comparison. A Georgist reader might see in it the familiar contrast between taxing improvements, which can shrink what gets built, and taxing land, which cannot, but that reading is an interpretation and the paper does not attempt a split-rate or land-only experiment (compare split-rate taxation).

Nuances and Limits

• Two versions differ. The published abstract reports "suggestive evidence of a reduction in the unit prices of larger homes" and says 55% of existing households lose. The July 2025 working paper finds no significant price effect and puts the share of losers at 58%. The working paper's rate, magnitudes and design details may not match the published article, and only the abstract is attributed here to the published version.
• A tax on housing wealth, not on land. The model taxes the value of housing assets. It has no separate land-value component, and its fixed-supply assumption is a modelling choice about the aggregate housing stock, not a land-only tax.
• Imprecise estimates. The Shanghai point estimates come with standard errors of 0.16 to 0.38 log points (WP p. 14, Table 2). An interval that wide does not exclude large price effects in either direction, including the 13.4% fall found by Zheng & Zhang. "Little impact" therefore means "no detectable impact in this design," not a demonstrated zero.
• Confounding and coverage. Housing purchase restrictions began in Shanghai in October 2010, shortly before the tax (WP p. 14, note 6). The outcome is a city-quarter average over new-home projects, not only the taxed purchases, and the data cover new-home transactions only (WP p. 11). The authors also withhold the area-per-transaction event study because it showed pre-treatment trends (WP p. 15, note 7).
• Calibrated model, not an observed policy. The universal-tax results are predictions conditional on the model's structure. The authors list what it leaves out, including capital gains on housing and China's household-registration (hukou) migration restrictions (WP p. 20). Their own comparison notes that another quantitative model (Zhu & Dale-Johnson 2020) finds a 5.8% price fall and a 16% rent rise under a targeted 0.5% tax, against +0.3% and +5.5% here (WP p. 26, note 11).
• The rebate is an assumption. The welfare gain depends on returning revenue in equal lump sums. The paper says explicitly that without redistribution the tax lowers welfare (WP p. 37), and real revenue in China would not necessarily be spent that way. The paper also notes that pilot revenue in Shanghai has been directed largely toward affordable housing construction (WP p. 10).
• Tension with another Shanghai study. Liu (2026) uses synthetic control on the same 2011 pilots and finds lower housing-price-to-income ratios in both cities, at abstract level. Cho, Park & Zhang find no significant price effect in Shanghai. The outcome variables, city samples and controls differ, and neither finding should be treated as settled.
• Broad-based tax, no simple map to land. The wealth pattern that leaves richer owners worse off rests on how housing wealth is distributed in China, where homeownership is about 90%, and may not carry over to countries with different ownership patterns (WP pp. 3, 34).

Bears On

• Research: Liu: Shanghai and Chongqing's Real Estate Tax Pilots — a contrasting reading of the same 2011 pilots (improved price-to-income ratios there; no significant price effect here). Neither is settled.
• Research: Hou, Kumhof & Shao: Land Value Taxation Simulation for China — a second calibrated general-equilibrium model for China, of a land-based tax rather than a tax on land and buildings, with different assumptions and welfare accounting.
• Research: Xu: China's Land Finance as Active Mode of Land Development — the lease-sale revenue system that a recurrent housing tax would sit beside.
• Objection: LVT inflicts a one-time wealth shock on current owners — the 58% loss share among existing households under a universal tax is a model-based illustration of transition losses, for a tax on land and buildings.
• Claim: Land value taxation can improve housing affordability — a caution from a different tax base: in this model a universal tax on land and buildings lowers prices but worsens the authors' affordability measure before transfers, and lump-sum transfers offset the increase only for low-income households, so price, rent and rebate need to be read together.

See Also

• Liu: Shanghai and Chongqing's Real Estate Tax Pilots
• Hou, Kumhof & Shao: Land Value Taxation Simulation for China
• Xu: China's Land Finance as Active Mode of Land Development
• Chu: Property Tax Incidence and Housing Market in Taiwan
• Land Value Tax
• Tax Capitalization
• Citizen's Dividend
• Split-Rate Taxation

Sources

1. Yunho Cho, Jinseong Park & Sisi Zhang (2026), "Taxing homeowners in China: Who are the winners and losers?," Real Estate Economics, published online 17 September 2026, DOI 10.1111/1540-6229.70074. doi.org — the full text is behind the publisher's access wall and was not read; the published abstract (as deposited with Crossref) was read at last review (2026-09-29) — used for the published version's statements on Shanghai ("little impact," "suggestive evidence" of lower prices for larger homes), the 55% share of existing households losing, and the summary of the model's three results (B-claim; abstract only).
2. Yunho Cho, Jinseong Park & Sisi Zhang (2025), "Taxing Homeowners in China: Who are the Winners and Losers?," KDI School Working Paper 25-08, 1 July 2025, 46 pp. KDI School archive — full text read at last review (2026-09-29); this earlier version differs from the published article (the working paper finds no significant price effect and 58% of existing households losing; the published abstract reports suggestive price reductions for larger homes and 55%), and its figures may have changed in publication; page numbers cited are PDF pages (printed numbering runs two lower) — used for the Shanghai synthetic difference-in-differences estimates (Table 2), the model description, the steady-state comparisons (Tables 5 and 6), the welfare decomposition (Table 8), the transition and heterogeneity results, the elastic-supply appendix (Table A-1) and the stated model limitations (B-claim: peer-reviewed article, text read from the working-paper version).
3. Joyce Yanyun Man (2012), "China's Property Tax Reform: Progress and Challenges," Land Lines, April 2012, Lincoln Institute of Land Policy. lincolninst.edu — used for the 28 January 2011 start of the Shanghai pilot, its coverage of newly purchased second homes of residents and first homes of nonresidents, the transaction-value base with a 60 square metre per person exclusion, and the description of the pilot's tax base, rate and collections as very small (A-claim; article describes the pilot as of 2012).