New Evidence on Property Tax Capitalization

Summary

"New Evidence on Property Tax Capitalization" (running head "Confirmations and Contradictions") is a 1998 article by Oded Palmon (Rutgers University) and Barton A. Smith (University of Houston), published in the Journal of Political Economy, vol. 106, no. 5 (October 1998), pp. 1099–1128 (DOI: 10.1086/250041). It is a direct empirical descendant of Oates (1969): both estimate how much of a property-tax differential shows up as a discount in house prices, and both appear in the same journal that anchors the capitalization literature. Palmon and Smith's contribution is a data set — 501 home sales in 1989 across 50 subdivisions in unincorporated northwest Houston, Texas, each served by a distinct municipal utility district (MUD) — designed specifically to solve the measurement problem that had left three decades of prior capitalization studies stuck at a wide range of "partial capitalization" estimates (commonly 15–30%). Because it is a peer-reviewed identification-focused study in the field's flagship general-interest journal, not an advocacy paper, its result is a credible mainstream data point on how fully a tax differential on a fixed asset gets priced into that asset's value.

The Core Argument and Findings

The identification strategy. Palmon and Smith's central methodological point is that most prior capitalization studies suffer from a spurious-correlation problem: property tax rates and public-service levels are usually positively correlated across jurisdictions (higher taxes fund better schools and services), and since researchers can never perfectly measure "public services," the unmeasured residual gets absorbed into the tax-rate coefficient, biasing estimated capitalization downward. Their Houston MUD sample was chosen to break this correlation. All 50 subdivisions sit in the same unincorporated area, receive identical county services and near-identical school-district services (three suburban districts with similar tax rates, test scores, and demographics), and all MUD infrastructure (water, sewer) was built by private developers and financed through municipal bonds serviced by MUD property taxes. Because bond-financing terms varied enormously — MUDs formed in the mid-1970s borrowed at 5–7%, those formed in the early 1980s at over 12%, and build-out rates (homes actually built versus platted) varied from full build-out to under 25% of planned homes — the resulting MUD tax rates varied from $2.30 to $19.00 per $1,000 of assessed value (mean $7.90) even though the underlying services funded were essentially the same everywhere. Texas law also required annual reassessment to market value, so stated and effective tax rates were the same, eliminating another common source of measurement error.

The findings. Estimating a capitalization model (property value as a function of housing characteristics and a tax-capitalization-adjusted discount rate), the authors report a baseline estimated capitalization coefficient of .617 (specification 2a), rising slightly to .639 in a specification controlling for build-out, commercial share, and foreclosure rate (specification 2c) — i.e., roughly 62–64% capitalization under the 3% net discount rate assumed by the prior benchmark study (Yinger et al. 1988), a rate the authors note is "substantially larger than the capitalization rates of 15–30 percent found by Yinger et al. under similar assumptions." The authors are explicit that at this discount-rate assumption, "this rate is significantly below full capitalization" and the null hypothesis of full capitalization can still be rejected.

The paper's headline result comes from a further step: using a higher, and by the authors' account more empirically grounded, discount rate (approximately 6.5%, following Linneman & Voith's 1991 estimate of the net user cost of owner-occupied housing for comparable demographics), re-estimation "indicates a slight overcapitalization, but the estimates are not significantly different from those that would indicate full capitalization." On this basis the authors conclude: "using the discount rate estimated by Linneman and Voith (1991) along with our unique data set, we obtain empirical results from which the full capitalization hypothesis implied by both Tiebout (1956) and Ricardian equivalence cannot be rejected" (Palmon & Smith 1998, p. 1110).

Relation to the Georgist Case

Palmon and Smith's result strengthens the empirical case underlying Landlords cannot pass a land value tax on to tenants, via the same capitalization logic documented on Oates (1969) and the tax capitalization concept page. The precise chain is:

1. The empirical finding. In a setting purpose-built to remove the confound (a fixed-services, tax-rate-varying sample), the measured degree to which a property-tax differential is priced into the value of the taxed asset is substantially higher than earlier, more confounded studies suggested — high enough that, under a realistic discount-rate assumption, the hypothesis of full capitalization cannot be statistically rejected.
2. What capitalization implies about incidence. Full (or near-full) capitalization means the entire discounted stream of the tax differential is reflected in the sale price at the moment the tax is imposed or becomes anticipated — i.e., the burden lands on the owner who holds the asset when the tax (or its expectation) changes, not on subsequent buyers or occupants. The authors state this explicitly: full capitalization "implies that housing market participants do, in fact, rationally discount properties burdened by higher taxes, implying that only unexpected tax changes can be passed on to new buyers" (p. 1110) — i.e., future owners do not bear a fresh burden; only the owner at the time of the tax change does.
3. The connection to non-pass-through. This is the same mechanism, one level removed, as the tenant-incidence claim: if a tax differential is capitalized into the purchase price an owner pays (so a new buyer is fully compensated via a lower price for the tax they will owe), the parallel logic is that a landlord cannot recover an already-capitalized tax from a tenant through higher rent, because the market-clearing rent is set by tenants' willingness to pay for the use of the property, not by the landlord's costs on a fixed-supply asset. Capitalization evidence of this kind is indirect for the landlord-tenant case specifically — Palmon and Smith study owner-occupied home sales, not rental incidence — but it is a direct empirical confirmation of the general principle the tenant-incidence claim depends on: a tax on a fixed asset is absorbed into that asset's price/value at the time of imposition rather than shifted forward.

Two scope limits should be kept precise. First, like Oates (1969) and most of this literature, Palmon and Smith's tax variable is the general property tax (largely funding MUD infrastructure bonds, not general revenue) applied to land plus structures together, not a land-value-only tax; the paper is evidence for the capitalization mechanism a Georgist LVT argument relies on, not a direct test of an LVT. Second, MUD taxes here fund debt service on infrastructure that benefits the specific taxed properties (water and sewer systems built to serve those subdivisions) — a benefit-tax element in the spirit of Hamilton's benefit-tax view — so the "burden" being capitalized is not purely a rent-extraction tax unconnected to any service received, though the paper's finding concerns the tax's price effect, not whether the tax is a fair or efficient charge for services rendered.

Nuances and Limits

• The full-capitalization result is discount-rate-dependent, not a single clean estimate. The authors' own baseline specifications (assuming a 3% discount rate, to match the prior literature) yield ~62–64% capitalization, explicitly "significantly below full capitalization." Only when a higher discount rate (~6.5%, from a different study) is substituted does the estimate become statistically indistinguishable from full capitalization. A reader should describe this paper's finding as "capitalization substantially higher than prior estimates, with the full-capitalization hypothesis not rejected under a plausible higher discount rate" rather than as an unqualified "near-full capitalization" finding — the paper itself flags the discount-rate sensitivity as a caveat, not a footnote.
• Small, geographically narrow, cross-sectional sample. 501 home sales in one year (1989) in one submarket (unincorporated northwest Houston) during a period of regional real-estate distress (following the mid-1980s Houston oil-bust housing slump, which the authors note directly — some MUDs had far fewer homes built than bonds anticipated, inflating effective tax rates). Generalizing the specific ~62–64% (or higher) capitalization rate to other cities, tax instruments, or time periods requires caution; the authors present the result as a methodological corrective to a biased literature, not as a universal capitalization constant.
• Residual confounding is reduced, not eliminated. The authors themselves note that MUD tax rates correlate with build-out ratio, commercial-property share, and foreclosure rate, and control for these in specifications 2c–2d; they acknowledge "there are no strong priors... regarding the extent or sign of the relationship between the value of housing services and these neighborhood characteristics," i.e., some residual neighborhood-quality confounding cannot be fully ruled out even in this improved design.
• Studies owner-occupied sales, not rental markets. The paper's capitalization evidence concerns the price a buyer pays for a home, not the rent a tenant pays a landlord. Its relevance to landlord-to-tenant incidence is by analogy to the general capitalization mechanism, not a direct rental-market test.
• Tests general property tax capitalization, not LVT incidence specifically. As with Oates (1969), Mieszkowski (1972), and Zodrow (2001), this is evidence about how property taxes (land plus improvements, here financing local infrastructure debt) get priced, feeding the broader capitalization literature that Georgist incidence arguments draw on — it is not a test of a land-value-only tax.

Bears On

• Outcome: Landlords cannot pass a land value tax on to tenants — provides updated, methodologically improved empirical support for the capitalization mechanism (tax borne by the owner at the time of imposition, not shifted to future buyers/occupants) that underlies the non-pass-through claim, though the paper studies home sales rather than rental incidence directly.
• Research: Oates (1969), the founding capitalization study — Palmon & Smith directly build on and revise Oates's approach, addressing the same spurious-correlation problem Oates's critics raised.
• Concept: Tax Capitalization — a methodologically strengthened empirical demonstration of this concept, with an unusually clean identification strategy (fixed services, varying tax rates).
• Research: Mieszkowski (1972), the "new view" of property tax incidence — Mieszkowski's theoretical claim that a fixed-supply asset's tax burden is capitalized onto the owner is the theoretical counterpart this paper empirically tests.
• Research: Hamilton's benefit-tax view — the MUD taxes studied here fund infrastructure benefiting the taxed properties directly, a benefit-tax-like setting worth reading against Hamilton's framework.

See Also

• Oates (1969), property tax capitalization
• Tax Capitalization
• Mieszkowski (1972), the "new view" of property tax incidence
• Zodrow (2001), "A Room with Three Views"
• Hamilton's benefit-tax view
• Outcome: Landlords cannot pass a land value tax on to tenants

Sources

1. Oded Palmon & Barton A. Smith (1998), "New Evidence on Property Tax Capitalization," Journal of Political Economy 106(5): 1099–1128. DOI: 10.1086/250041 (paywalled via University of Chicago Press/JSTOR; free full-text mirror consulted at matthewturner.org) — primary source for all findings, data description, regression estimates, and direct quotations on this page; the full text was successfully retrieved and read in this session.
2. Wallace E. Oates (1969), "The Effects of Property Taxes and Local Public Spending on Property Values," Journal of Political Economy 77(6): 957–971 — wiki summary — used for the founding-study context and the spurious-correlation critique this paper addresses.
3. John Yinger, Howard S. Bloom, Axel Borsch-Supan & Helen F. Ladd (1988), Property Taxes and House Values: The Theory and Estimation of Intrajurisdictional Property Tax Capitalization, Academic Press — used as the benchmark prior study (15–30% capitalization estimates, 3% discount-rate assumption) that Palmon & Smith's results are directly compared against.
4. Peter Linneman & Richard Voith (1991), "Housing Price Functions and Ownership Capitalization Rates," Journal of Urban Economics 30(1): 100–111 — used as the source of the higher (~6.5%) discount-rate estimate that produces Palmon & Smith's full-capitalization result.
5. Peter Mieszkowski (1972), "The Property Tax: An Excise Tax or a Profits Tax?", Journal of Public Economics 1(1): 73–96 — wiki summary — used for the theoretical framing of why a fixed-supply asset's tax burden falls on the owner.