Feldstein, "The Surprising Incidence of a Tax on Pure Rent" — the Portfolio-Shifting Challenge to Full Capitalization

Summary

This page carries the standard mainstream challenge to the Georgist premise that a tax on land rent is fully unshifted and fully capitalized into a lower land price — the premise underlying ATCOR, EBCOR, and the landlords-cannot-pass-LVT-to-tenants outcome, presented per the wiki's steelman rule.

Martin Feldstein (1977), "The Surprising Incidence of a Tax on Pure Rent: A New Answer to an Old Question" (Journal of Political Economy 85(2), pp. 349–360), argues that the classical Ricardian conclusion — net rent falls by the full tax, land price falls by the capitalized tax — is false in a general-equilibrium growth model. Because land and produced capital are competing vehicles for life-cycle (retirement) saving, a tax that lowers the value of land pushes household saving into produced capital. The resulting capital deepening raises land's marginal product and lowers the interest rate at which rent is capitalized, so part of the tax is shifted off land — onto capital (a lower net yield) and labor (a higher wage) — and the land price can end up higher, not lower, than before the tax.[1]

This is genuine counter-evidence to full capitalization. Its force is bounded on two sides, both flagged below: (a) Feldstein's own model assumes it away under alternative specifications he names explicitly, and (b) two direct replies — Calvo, Kotlikoff & Rodriguez (1979) and Fane (1984) — show the classical result returns once bequests/altruism or a properly compensated experiment are admitted. Note also what the result does not say: the shifting is onto capital owners, via a savings channel — it is not a mechanism by which a landlord raises rents on tenants.

Key Findings — Feldstein's Mechanism at Strength

The claim, stated flatly (abstract). "The classic example of an unshiftable tax is the general tax on pure rental income. Since Ricardo, economists have believed that the annual net rental income of unimproved land falls by the amount of the annual tax and its price by the capitalized value of this tax. This paper shows that these conclusions are false, that the tax on pure land rents is at least partly shifted, and that the price of land may be increased by the imposition of a tax."[1]

Why — land and capital are competing stores of retirement wealth. "The essential oversight of the classical analysis is to ignore the fact that land and produced capital are alternative components of individual life-cycle wealth."[1] Working through the overlapping-generations savings model, Feldstein continues: "If the tax on pure land rent reduces the value of land, a larger amount of the desired wealth must be accumulated in the form of produced capital. The tax on rental income thus induces an increase in the equilibrium capital stock and therefore in the equilibrium ratio of capital to land. This raises the marginal productivity of land and reduces the rate of interest at which net land rents are capitalized. Part of the tax on pure rent is thus shifted in the form of a lower net yield on capital and a higher wage rate. Moreover, the price of land does not fall by as much as the traditional theory predicts."[1]

Two distinct channels, both missed by the classical account. "even if income effects are excluded, the traditional analysis is wrong in two important ways. First, it overlooks the fact that, because land is an asset, a tax on rent can change the supply of other factors even if there are no income effects … Second, the traditional theory ignores the effect of the portfolio balance requirements."[1] The first channel (induced capital accumulation) is developed in Sections I–II; the portfolio-balance channel in Section III.

Independent restatement. Alberto Petrucci summarizes the result cleanly: "The imposition of a tax on land rental income, by reducing the value of unimproved land, diverts saving from the fixed asset toward reproducible capital … spurring capital formation and raising output. The rate of interest falls, while the wage rate and the marginal productivity of land increase. The price of land, after the initial drop, may increase in the final equilibrium because of the interest rate decline. These results, discovered by Feldstein (1977), are very robust as they are independent of alternative uses of the land tax revenues."[5]

The Load-Bearing Assumptions — Feldstein's Own Hedges

Feldstein is candid that the result depends on a specific setup. Two assumptions do the work, and he flags both in the paper:

• A one-good economy, revenue not returned as capital or transfers. In footnote 2 he writes: "The assumption of an economy with a single good avoids any effect of the tax on the composition of a demand. It is important to recognize that this is a crucial assumption. The conclusions of both the traditional analysis and of my own study would be different if the revenue were used to accumulate capital, to finance transfer payments to the aged, etc."[1]
• Non-altruistic, life-cycle demographics. The savings-shift mechanism requires that households save only for their own retirement, so that a fall in land value forces them to hold more produced capital. As Petrucci puts it, "The Feldstein results, however, hinge on the crucial assumption of non altruistic overlapping-generations demographics."[5] This is exactly the hinge the replies push on.

The Replies — the Classical Result Returns

Calvo, Kotlikoff & Rodriguez (1979), "A New (?) Reason for an Old Answer." In a model with Ricardian demographics — dynamic life-cycle saving with bequests and intergenerational transfers (à la Barro 1974) — the Feldstein shifting vanishes and the classical incidence is restored. Petrucci: "Calvo, Kotlikoff and Rodriguez (1979) (henceforth CKR) demonstrate that in an intertemporal optimizing model of saving and capital formation with Ricardian demographics, like a dynamic life-cycle model with bequests and intergenerational transfers (as in Barro, 1974), the Ricardian effects of a compensated rent tax are confirmed."[5] Intuitively, altruistic households linked across generations already choose the capital stock at the "modified golden rule," so taxing land no longer forces extra capital accumulation — the tax stays on land. The title's parenthetical "(?)" telegraphs the point: the old answer (full incidence on land) holds.[2]

Fane (1984), "The Old Reason for the Old Answer." Fane shows the classical result also returns in Feldstein's own non-altruistic setting once the tax experiment is properly compensated. As Petrucci (2006) describes it: "Fane (1984) argued that, once a fully compensated land tax is considered in a model with finite-lived disconnected generations, the unique effect of taxation is to cause a fall in the land value with no shifting; a land tax is fully compensated when the land tax shift is accompanied by the issuance of perpetual government bonds, whose sale proceeds are used to make lump-sum transfers to the landlords."[6] Che, Kumar & Stauvermann (2021) concur that Fane "applied a fully compensated land tax in the context of Feldstein's approach whereby the landowners will be fully compensated for the tax burden … Later, Buiter (1989) validated the outcome of Fane in a neoclassical growth model."[7] The shifting in Feldstein, on this reading, is an artifact of an uncompensated wealth effect, not of the land tax as such.

What It Cuts Against — and What It Doesn't

Cuts against: the strong form of full capitalization / full incidence on the landowner — the premise that a tax on rent (and, symmetrically for EBCOR, the excess burden of other taxes) is borne entirely by land rent as a pure residual. If a tax on rent is only partly borne by rent, that residual-surplus premise is not airtight in general equilibrium. This is why the result is wired as a theory-side challenge to landlords cannot pass LVT to tenants and cited on the ATCOR and EBCOR pages.

Does not cut against the anti-tenant-pass-through claim directly. Feldstein's shifting runs to capital (lower interest) and labor (higher wages) through an economy-wide savings channel — the pre-tax land rent actually rises in his model. It is emphatically not a mechanism by which a landlord raises the rent charged to a sitting tenant. A careful reader should not cite Feldstein for the proposition that LVT reaches renters; it does the opposite on that specific margin.

Distinct from the timing critique. Feldstein's challenge (a savings/portfolio incidence effect) is a different line of attack from the Bentick–Mills development-timing critique (a non-neutrality arising from assessing land at market value that capitalizes development options). Both qualify land-tax neutrality/incidence, but through unrelated mechanisms and with different rebuttals — they should not be conflated.

Scan Depth (honest note)

The paper's body was read verbatim through pp. 349–351 — the abstract, introduction, the exclusion of income effects, and the setup of the savings-equilibrium model, including the two footnoted caveats — from the free full-text PDF hosted by the School of Cooperative Individualism (an image scan; text recovered via extraction). The formal portfolio-balance derivation (Section III) and the concluding section (pp. 352–360) were not read verbatim — the scanned pages did not OCR cleanly and the JSTOR/JPE full pages are paywalled — so those are represented via the verified secondary restatement in Petrucci (2004/2006).[5][6] The CKR (1979) and Fane (1984) replies are paywalled and are characterized here from verified secondary descriptions (Petrucci; Stauvermann & Kumar), not from primary quotation; a future editor with JSTOR access should upgrade those to direct quotes.

Bears On

• Outcome (challenges): Landlords cannot pass a land value tax to tenants — the full-capitalization premise, on the incidence (not the rent-pass-through) margin
• Concept (challenges premise): ATCOR · EBCOR — land-as-pure-residual
• Concept: Tax Capitalization · Deadweight Loss
• Adjacent theoretical critique (distinct mechanism): Bentick–Mills timing neutrality

See Also

• Nicolaus Tideman — neutrality-side rebuttals in the assessment/timing branch of the debate
• Economic Rent

Sources

1. Martin Feldstein (1977), "The Surprising Incidence of a Tax on Pure Rent: A New Answer to an Old Question," Journal of Political Economy 85(2), pp. 349–360. DOI · JSTOR 1830795 · free full-text PDF — primary source for the shifting result, the savings/portfolio mechanism, and the two load-bearing assumptions (A-claim; read verbatim pp. 349–351, remainder via secondary restatement — see Scan Depth).
2. Guillermo A. Calvo, Laurence J. Kotlikoff & Carlos A. Rodriguez (1979), "The Incidence of a Tax on Pure Rent: A New (?) Reason for an Old Answer," Journal of Political Economy 87(4), pp. 869–874. DOI — used for the bequest/altruism reply restoring classical incidence (C-claim; via secondary description, paywalled). A new source for this wiki.
3. George Fane (1984), "The Incidence of a Tax on Pure Rent: The Old Reason for the Old Answer," Journal of Political Economy 92(2), pp. 329–333. DOI — used for the compensated-tax reply restoring classical incidence (C-claim; via secondary description, paywalled). A new source for this wiki.
4. EBCOR · ATCOR · Landlords cannot pass a land value tax to tenants — wiki pages whose full-capitalization premise this result challenges.
5. Alberto Petrucci (2004), "On the Incidence of a Tax on Pure Rent with Infinite Horizons," FEEM Nota di Lavoro 160.2004 (published as Journal of Public Economics 90(4–5), 2006, pp. 921–933). Working-paper PDF · journal version DOI — used for the verified restatement of Feldstein's mechanism, its dependence on non-altruistic demographics, and the CKR and Fane replies (B-claim). A new source for this wiki.
6. Alberto Petrucci (2006), "The Incidence of a Tax on Pure Rent in a Small Open Economy," Journal of Public Economics 90(4–5), pp. 921–933 — publisher; the abstract text used here to characterize Fane (1984)'s compensated-tax argument.
7. S. Che, Ronald Ravinesh Kumar & Peter Josef Stauvermann (2021), "Taxation of Land and Economic Growth," Economies 9(2): 61, discussing Feldstein (1977), Calvo et al. (1979), Fane (1984), and Buiter (1989). MDPI — used as a second secondary source corroborating the Fane/Buiter characterization.