Distributional Consequences of Converting the Property Tax to a Land Value Tax

Summary

"Distributional Consequences of Converting the Property Tax to a Land Value Tax: Replication and Extension of England and Zhao" is a peer-reviewed article by John H. Bowman (Department of Economics, Virginia Commonwealth University, Emeritus — VCU's economics faculty since 1981, specializing in subnational taxation) and Michael E. Bell (Institute for Public Policy, George Washington University, and a veteran state/local public-finance economist), published in the National Tax Journal, Vol. 61, No. 4, Part 1 (December 2008), pp. 593–607 — the flagship peer-reviewed journal of the National Tax Association / International Institute of Public Finance tradition in US public finance. Bowman and Bell were also long-standing Lincoln Institute of Land Policy-affiliated researchers on land and property tax valuation, giving the paper institutional grounding beyond a single university.

The paper is explicitly a replication and extension of an earlier study by Richard England and Min Qiang Zhao, "Assessing the Distributive Impact of a Revenue-Neutral Shift from a Uniform Property Tax to a Two-Rate Property Tax with a Uniform Credit," National Tax Journal 58(2) (2005), pp. 247–260, which examined a hypothetical revenue-neutral shift from a uniform property tax to a pure land value tax in Dover, New Hampshire, and found the change would be regressive: it would shift tax burden toward lower-value residential properties (and thus, the paper argued, toward lower-income and asset-poor households) and away from higher-value properties. Bowman and Bell apply the same method to a different city — Roanoke, Virginia — using assessment microdata for 28,680 single-family residential parcels for the 2003 tax year, to test whether England and Zhao's regressivity finding was a general property of land value taxation or an artifact of Dover's specific housing stock and land-value distribution.

The Core Findings

• Opposite direction of result. Where England and Zhao found a revenue-neutral property-tax-to-LVT shift to be regressive in Dover, Bowman and Bell find the opposite result for Roanoke: "We replicate their analysis for Roanoke, Virginia, with results opposite those for Dover. … Thus, both approaches for Roanoke show initial tax burden changes to be progressive" (p. 593). Concretely, all three home-value groups in Roanoke see tax cuts under the simulated switch (2003 rates: $1.21 per $100 of total assessed value replaced by a revenue-neutral $5.6303 per $100 of land value), and the median percentage reduction is largest at the bottom — 22 percent for the highest-value 30 percent of homes, 23 percent for the middle 40 percent, and 29 percent for the lowest-value 30 percent (citywide median −24.2 percent, mean −20.7 percent; Table 4, pp. 601–602). For Dover, by contrast, the paper reports England and Zhao's finding that "[f]or the lowest, middle, and highest home value groups the respective average tax changes are +$638, +$311, and –$138. Moreover, their respective percentages of homes with tax increases are 99 percent, 80 percent, and 26 percent" (p. 601).
• Two measures of incidence. As in England and Zhao, Bowman and Bell measure incidence primarily by linking parcel-level property tax changes to assessed property value (which, following England and Zhao, they treat as a "good prox[y] for permanent income," p. 595). They extend the analysis by also linking the tax changes to U.S. Census Bureau income and poverty data at the census-tract level for Roanoke's 23 tracts, testing incidence against income and poverty rates directly rather than only against property value.
• Progressive on both measures. By both measures, the Roanoke results point the same way: "the resulting tax change would benefit most those areas with lowest incomes and highest poverty rates" (p. 593). The median tax change is negative in 21 of the 23 census tracts (tract medians range from −56.4 percent to +6.9 percent), and the correlation between a tract's median percentage tax change and its median family income is +0.85 (+0.80 for per-capita income), with strongly negative correlations for poverty (−0.79 family poverty rate, −0.75 individual poverty rate): lower-income, higher-poverty tracts get the largest reductions, while higher-income tracts see smaller reductions or increases (Table 5, pp. 603–604) — a distributionally progressive result, contrary to Dover.
• Attributed to local housing-stock and land-value structure. The authors attribute the reversal to differences in property-tax-base composition, not to a flaw in England and Zhao's method: "These different results reflect the different characters of the two cities. … in Dover, improvements account for less than twice as much value as land per residential parcel, while in Roanoke improvements account for six times as much as land. In addition, the within-class variation in the land share of value for single-family residential properties in Roanoke is much higher than it is in Dover" (pp. 604–605). The driver of Dover's regressivity is that improvements there make up a rising share of value as home values rise: the correlation between a single-family parcel's total assessed value and its building-to-land value ratio is +0.48 in Dover versus −0.006 (essentially zero) in Roanoke (Table 3, mean value ratios 1.88 vs. 5.99, pp. 598–599). As the authors put it for Dover: "For households with higher incomes, improvements are a higher percentage of total property value, which means larger tax cuts if the property tax becomes a land value tax" (p. 599).

Relation to the Georgist Case

This paper is genuinely a mixed contribution to the wiki's progressivity case, and it should not be read as a simple confirmation. Its central empirical message is that the distributional effect of a property-tax-to-LVT shift is jurisdiction-dependent, not a fixed property of land value taxation in general. For Roanoke, Bowman and Bell's own results support land value tax can be progressive; but the very existence of this paper is a replication exercise built on a prior finding — England and Zhao (2005) on Dover, NH — that a similarly designed, revenue-neutral shift can instead be regressive. Read together, the two studies establish that whether a shift to LVT helps or hurts lower-income/lower-value-property households by this local-incidence method depends on the specific relationship between land value, improvement value, and income in the jurisdiction being taxed — a genuine complication for any unqualified claim that "LVT is progressive."

This sits alongside Common Wealth Canada's distributional modelling, which likewise found a national LVT alone could be regressive by current income absent a rebate, and against which a design fix (a flat credit) was needed to restore progressivity. Bowman and Bell's finding is different in kind — it is not about needing a rebate, but about the underlying incidence pattern varying by place — but the upshot for the wiki is the same: the progressivity outcome page's own caveat that "progressivity depends on design" should also be read as "and on local land-value/income structure," not only on policy design choices like exemptions or dividends.

Nuances and Limits

• Two data points, not a general law. The paper's own contribution is precisely that a single city's result (Dover) does not generalize; Roanoke's result does not generalize either. Bowman and Bell had earlier run the same kind of equal-yield simulation on two further Virginia localities — Chesterfield County (the largest jurisdiction in the Richmond metropolitan area, population then over 250,000) and Highland County (rural, Virginia's least-populous county at under 2,500) — in a 2004 Lincoln Institute of Land Policy working paper, which found that "[i]n all three localities, the move to an equal-yield split-rate tax would reduce the residential share of the real property tax while increasing the business tax share" (Bowman & Bell 2004, abstract) — a different but related distributional dimension (owner class rather than income). The 2008 NTJ article summarizes that earlier three-locality work (p. 595, citing Bowman and Bell 2004, pp. 45–6): simulated aggregate residential tax reductions of 20 percent in Roanoke, 11 percent for urban single-family homes in Chesterfield, and 72 percent in Highland's county seat of Monterey; the full Chesterfield/Highland analysis is in the 2004 working paper, not the 2008 article.
• Incidence proxy, not a full tax-incidence model. Like England and Zhao, the method assumes the statutory/assessed burden change falls on the current owner (no explicit modelling of capitalization, tenant pass-through, or long-run supply response) and uses census-tract-level income/poverty as a proxy for household-level income, which can mask within-tract heterogeneity (e.g., asset-rich, cash-poor retirees within an otherwise low-income tract).
• Revenue-neutral, residential-only framing. The comparison is a hypothetical revenue-neutral swap confined to the residential property tax base in each city; it does not model a broader national LVT, does not net out effects on renters versus owners explicitly, and does not address commercial/industrial or agricultural land separately from the residential parcels studied.
• Single-year, single-city snapshots. Each city's result is based on one assessment year's cross-section (Roanoke: 2003) rather than a panel across years or a broader sample of cities, so neither this paper nor England and Zhao's supports a claim about the average or typical American city.

Bears On

• Outcome: A land value tax can be progressive — Bowman and Bell's Roanoke result is direct affirmative evidence for this outcome, but the paper's own reason for existing (replicating a regressive finding elsewhere) means it should be cited as jurisdiction-dependent support, not as a general proof.
• Research: Common Wealth Canada — distributional impacts of LVT — a parallel case where LVT-alone incidence also failed to be straightforwardly progressive without an added policy design element; read together they show two independent routes (place-dependence here; rebate-design there) by which "LVT is progressive" needs qualification.
• Research: Dye & England — Assessing the Theory and Practice of LVT — the standard Lincoln Institute policy-focus report surveying LVT theory and practice more broadly; this paper is a focused empirical complement on the specific distributional question.
• Objection: LVT hurts the "asset-rich, cash-poor" — the census-tract income/poverty proxy used here is directly relevant to, but does not resolve, concerns about household-level (rather than tract-level) hardship cases.
• Concept: Land Value Tax — a concrete empirical test of how a shift to this tax base plays out for real assessment rolls.

See Also

• Plummer (2010) — the closest companion distributional study
• England & Zhao (2005), the Dover NH study
• Land Value Tax
• A land value tax can be progressive
• Assessing the Distributional Impacts of a Land Value Tax (Common Wealth Canada)
• Assessing the Theory and Practice of LVT (Dye & England)
• Objection: LVT hurts the "asset-rich, cash-poor"

Sources

1. John H. Bowman & Michael E. Bell (2008), "Distributional Consequences of Converting the Property Tax to a Land Value Tax: Replication and Extension of England and Zhao," National Tax Journal 61(4), Part 1, pp. 593–607. Full text (archived NTA copy) · IDEAS/RePEc record · DOI 10.17310/ntj.2008.4.02 — full text fetched and verified; used for the exact citation, methodology (28,680 single-family Roanoke, VA parcels, tax year 2003, Table 3), the Table 4/Table 5 magnitudes, the Dover-mechanism quotes, and the headline progressive finding contrasted with Dover, NH.
2. Richard W. England & Min Qiang Zhao (2005), "Assessing the Distributive Impact of a Revenue-Neutral Shift from a Uniform Property Tax to a Two-Rate Property Tax with a Uniform Credit," National Tax Journal 58(2), pp. 247–260. DOI 10.17310/ntj.2005.2.05 — the original Dover, New Hampshire study that Bowman & Bell replicate; citation details verified against the reference list of the 2008 article (which cites it as National Tax Journal 58 No. 2, June 2005, pp. 247–60) and the Crossref DOI record.
3. John H. Bowman & Michael E. Bell (2004), "Implications of a Split-Rate Real Property Tax: An Initial Look at Three Virginia Local Government Areas," Lincoln Institute of Land Policy Working Paper WP04JB1 (reprinted in State Tax Notes 32, April 26, 2004, pp. 261–91). PDF — the earlier three-locality (Roanoke, Chesterfield, Highland) analysis summarized in the 2008 article; used for the residential-versus-business share finding.
4. Common Wealth Canada, "Assessing the Distributional Impacts of a Land Value Tax," 2024 — wiki summary — used for the parallel point that LVT-alone incidence can fail to be progressive absent specific design choices.