Taxing Tech Rents — an Instrument Comparison

The Question

Dominant digital platforms earn large, durable returns from network-effect moats, accumulated data, and gatekeeping — positions that behave like land-like locations everyone in a market must pass through. If those returns are economic rent, the Geoist instinct is to capture them for public good rather than leave them as private windfalls (Geoism). This page compares the instruments that have been proposed or tried to do it, and grades each on one practical question: does it actually reach platform rent, and does the burden stay on the platform?

Five families are in play, and they split across the two poles the data-rents page identifies — capture the rent (tax it) or dissolve it (compete it away):

1. Rent-targeting corporate tax bases — ACE and the cash-flow tax / DBCFT: tax excess returns wherever they are, without deciding which are "tech" rents.
2. Digital services taxes (DSTs) — the tried-and-badly-aimed case.
3. Romer's progressive digital-advertising tax — aimed at the attention/surveillance business model.
4. Data dividends / data-as-labor — price the unpaid input rather than the output.
5. Rent dissolution via antitrust, the DMA and interoperability — destroy the moat instead of capturing what it throws off.

Read This First — the Gradient Is Steepest Here

The wiki's rent gradient rule (EDITORIAL §0) applies harder in this domain than almost anywhere else. Land is the clean case: fixed supply, no production to discourage, a century of incidence evidence. Platform returns are not. Whether big-tech profit is unearned rent (from moats, gatekeeping and data monopsony) or the quasi-rent that temporarily rewards genuinely superior products and intangible capital is genuinely disputed, and this page carries both readings at strength:

• The rent reading. Aggregate markups rose from about 21% above marginal cost in 1980 to about 61% by 2016, concentrated in the upper tail of firms (De Loecker, Eeckhout & Unger 2020, via corporate profits increasingly rents); two-sided-market theory explains why platforms tip toward a single winner (Rochet & Tirole); and an official UK panel found network effects and returns to scale on data keep digital markets tipped (Furman Review).[6]
• The efficiency counter. Much of the profit rise traces to real but mismeasured intangible capital — software, brands, processes — which would make a large slice quasi-rent, not rent (Crouzet & Eberly 2019); and the "superstar firm" account reads concentration partly as efficient firms winning share (Autor et al. 2020).[7][8]

No one has a clean rent/quasi-rent decomposition for platforms of the kind the wiki has for land. Every instrument below therefore also runs into the innovation-incentive objection the wiki steelmans at Taxing quasi-rents kills innovation — the return that motivated an invention is not free to tax the way a location windfall is. Grades here are grades on aim and incidence, not verdicts that the rents are large or that capture is warranted.

The Grades at a Glance

The grades are the wiki's analysis, not a source's ranking. They read the instruments against the Geoist ideal — reach the rent, leave the incentive — and none clears it the way an LVT does on land.

1. Rent-Targeting Corporate Tax Bases (ACE / DBCFT) — B

The mainstream public-finance answer is not a tech tax at all: it is to reshape the ordinary corporate tax so its base is only above-normal returns. An allowance for corporate equity (ACE) deducts a notional "normal" return on equity alongside interest, so a marginal investment earning exactly the required return pays nothing and the base collapses, by construction, to economic rent.[9] A cash-flow tax reaches the same rent-only base by immediate expensing of investment — E. Cary Brown's (1948) "silent partner" result — and its destination-based variant (DBCFT) taxes location-specific rents tied to where sales occur.[10]

Why it grades well on aim. It needs no regulator to decide which profits are tech rents; the base itself excludes the normal return, so whatever excess a platform earns is caught alongside every other firm's. The Mirrlees Review (2011) recommended an ACE for UK corporation tax "to ensure that only 'excess' returns to investment are taxed."[9] US Treasury's own distributional method treats the supernormal-return share of the corporate base — roughly 63% (Cronin et al. 2013), rising toward ~75% (Power & Frerick 2016) — as borne by shareholders, which is the analytic basis for calling a rent-only corporate tax progressive.[11]

Why it is only a B. Three honest deductions:

• Source-blind. It taxes rent wherever it sits; it makes no attempt to separate a network-effect moat from a favourable location or a patent. That is a virtue for administrability and a limit if the goal is specifically platform rents.
• It reaches quasi-rents ex post. An above-normal return realized on a risky, successful innovation is exactly the quasi-rent the innovation objection protects; a true rent tax spares it only if loss treatment is symmetric (full refundability — Bond & Devereux 1995).[10]
• Political fragility and an untested incidence hole. Every full European ACE has been repealed — Croatia, Austria, Latvia, Belgium (2006–2023), Italy — and Belgium's notional deduction was worth about €4.5bn in 2014, ~35% of that year's corporate tax revenue, with ~40% of the deductions flowing to tax-planning special-purpose vehicles.[9][12] The DBCFT came closest to law in the US in 2017 and died to concentrated import-lobby opposition.[10] And Fuest, Peichl & Siegloch (2018) show about half the German corporate-tax burden reaches workers through rent-sharing in wage bargaining — so even a tax that falls only on surpluses can be partly shifted, and no direct wage-incidence study of an actual ACE exists.[13] Verdict: rent-targeting corporate taxes reduce debt bias is graded on the wiki as real but mixed.

2. Digital Services Taxes — D

DSTs are the one family with real-world incidence evidence, and it is discouraging. A DST is a levy on the gross revenue a large platform earns from a jurisdiction's users; since 2019 the UK (2%), France (3%), Austria (5%), Turkey (7.5%), Spain, Italy and India have enacted them, largely as a stop-gap pending the OECD/G20 "Pillar One" deal, motivated by the European Commission's 2018 finding of a 9.5% effective tax rate for digital businesses versus 23.2% for traditional firms.[1] The most careful theoretical defence — Cui & Hashimzade (2019) — rationalises the DST as a tax on location-specific rent, "just as many countries already levy royalties on rent from extracting natural resources," mapping it onto the resource-royalty template.[2]

Why it fails on incidence. A DST is levied on revenue in a two-sided market, not on the rent or a fixed factor, so it behaves like an excise: the platform re-optimises its fees and pushes the burden onto whichever side is less elastic. The record is explicit —

• Amazon announced UK seller fees would "increase by 2%" from 1 September 2020 to cover the UK DST, having already raised French seller fees 3% for France's DST a year earlier; Google added a 2% "DST Fee" to UK advertisers from November 2020, with parallel surcharges for Austria and Turkey.[5]
• Muddasani & Langenmayr (2025) put numbers on it using Amazon marketplace data: "on average, Amazon increased its fees by roughly half the amount of the digital service tax," and sellers "largely passed these increased fees on to consumers," so "large digital firms thus bear only a small part of the tax burden."[3]

This is the wiki's cleanest badly-aimed rent tax: a tax intended to fall on rent, designed to fall on revenue, that does not behave like a rent tax. It is why the DST grades D — not because platform rents are uncapturable, but because taxing revenue is not taxing rent, and the LVT contrast is exact: a fixed factor's owner cannot pass the tax forward because supply does not change, whereas a platform simply re-prices.[3] One partial silver lining the authors note: the DST may still shift competitive balance toward brick-and-mortar rivals — an antitrust-adjacent effect, not rent capture.

3. Romer's Progressive Digital-Advertising Tax — C

Nobel laureate Paul Romer's proposal (2019, developed 2021) is a progressive tax on the revenue platforms earn from targeted digital advertising — the Google/Facebook business model.[R1] Its design turns on one incidence insight: "Because revenue has a location, a tax on revenue will not be subject to" the profit-shifting that lets multinationals move reported profit offshore.[R2] Two features distinguish it from a plain DST: a steeply progressive schedule ("If big is bad, tax big") calibrated so a dominant firm could cut its bill by splitting in two; and a deliberate escape hatch — a platform that abandons targeted ads for subscriptions escapes the tax entirely. Romer frames the harm as political (he reports Facebook controlled 59% and Google 18% of US digital political-ad spending), and Acemoglu & Johnson (2024) recast a flat-rate-above-a-threshold variant as a public-health "sin tax."[R3]

Why it grades C as a rent instrument. On the wiki's own reading, Romer's tax is only incidentally a rent-capture device. Romer does not especially mind if it is passed through, because the point is behavioural — to make the ad model unattractive relative to subscriptions. That reframes it as closer to Pigouvian correction and antitrust than to clean rent capture: its success metric is a smaller, less concentrated, less surveillance-dependent industry, not a captured rent stream. It is a dissolve-leaning device reached by a fiscal route — a cousin of the Furman Review remedies. And the ad-revenue base is not pure rent: it funds the zero-price services users actually value, so a tax calibrated to force firms out of the model is, by construction, not a light touch on the return to building the service — squarely the quasi-rent objection. It is untested, and the DST record cautions against assuming the intended incidence will materialise.[R2]

4. Data Dividends / Data-as-Labor — C-minus

A distinct strand targets the input side. Arrieta-Ibarra, Goff, Jiménez-Hernández, Lanier & Weyl's "Should We Treat Data as Labor?" (2018) argues users create the data that trains and powers digital services, yet it is treated as free capital "harvested" by a handful of platforms that hold monopsony power over data provision, capturing and under-rewarding the value users create.[14] The remedies: pay users for their data, or levy the platforms and distribute the proceeds as a citizen's dividend — turning an unpriced input into a priced one. Posner & Weyl's Radical Markets develops the fuller programme, and the Harberger tax / COST is its self-assessment cousin for asset holding.

Why C-minus. The diagnosis — data monopsony extracting an uncompensated contribution — is a genuine rent story on the input side, and it points to a specific remedy rather than a blunt one. But every instrument here is untested, and the binding problem is valuation: no one has a working method to price an individual's non-rival data contribution the way mass appraisal prices a site. Korinek & Stiglitz's warning generalises — "non-distortionary" taxation of a fixed factor is exact for pure land rent but inherits the whole valuation problem when applied to data, where separating the rent from the return to genuine intangible capital is unsolved.[4] Right diagnosis, no design ready to grade higher.

5. Rent Dissolution — Antitrust, the DMA, Interoperability — B (for the dissolve pole)

The last family refuses the capture framing entirely: instead of taxing the rent, destroy the moat that creates it. This is the pole where the digital frontier offers a move the clean land case never can — land cannot be un-scarced, but a data moat can be engineered open.

• The diagnosis is mainstream and non-Georgist. The Furman Review (2019) found digital markets "tip" to a single winner made durable by "network effects and returns to scale of data," documenting Google at over 90% of UK search since 2008 — the empirical precondition for calling the returns a rent rather than a transient lead. Its prescription, though, is to dissolve: a digital-markets unit, strategic-market-status conduct rules, data mobility and interoperability, and tougher merger policy — deliberately not a rent tax.[6]
• The DMA legislates it. The EU's Digital Markets Act (Regulation 2022/1925, in force since November 2022) binds designated "gatekeepers" — defined by an "entrenched and durable position," the competition-law definition of a durable rent — to open their app stores (Art. 6(4)), hardware/OS features (Art. 6(7)), and, for the first time in law, their messaging services to interoperability (Art. 7, phasing in one-to-one, group, then voice/video over four years).[15] Kades & Scott Morton (2020) supply the analytical case: mandatory interoperability "could overcome the network effects that protect the incumbent from entry," at low incremental cost — though it is "likely a necessary, but not necessarily a sufficient, condition."[16]

Why it grades B on its own terms. Interoperability finesses the quasi-rent objection rather than answering it: it leaves the platform's revenue alone and removes the barrier to competition, so a firm winning on genuine quality keeps its customers while one coasting on lock-in loses them. Its two structural limits are real: it raises no revenue — if it works, the rent is competed into lower prices rather than a citizen's dividend, so capture and dissolve genuinely diverge on who ends up with the surplus; and its effectiveness is unproven — the DMA is early (first gatekeepers designated September 2023), and whether opening the moat erodes the rent or merely reveals that the advantage was real quality is the same is-it-rent question, relocated from the tax base to the remedy.[15][16]

The AI Coda — Where the Whole File Is Heading

Korinek & Stiglitz (2017) supply the strongest mainstream authority that the AI economy generates capturable "innovator rents," and that in the long run the gains from worker-replacing technology accrue to whatever irreproducible factor stays scarce — "the owners of non-reproducible factors absorb all the rents" — which, "since it is in fixed supply, they can be taxed and redistributed without creating distortions."[4] That is the law of rent restated for the machine age, and it is Georgist in structure though the authors intend no Georgist argument. But they are equally explicit that the innovator surplus "rewards innovators for what they accomplish," and their own second-best model trades patent length against a distortionary capital tax — i.e. they take the innovation-incentive cost seriously. The Georgist case is strongest exactly where their logic is strongest — on the fixed factor — and should claim no more certainty than that on the AI frontier.

Counter-Evidence and Honest Limits

• The base may not be rent at all. The single largest limit is the one the gradient names: Crouzet & Eberly and Autor et al. show a substantial share of high platform profit may be the return to real (if mismeasured) intangible capital and to genuine productivity divergence — quasi-rent, not rent. If so, every capture instrument above is partly taxing the incentive to build good products, and the honest reading is that the pure-rent component's size is contested and not precisely known (corporate profits increasingly rents).
• The one hard incidence datum points the wrong way. The DST is the one instrument here with a real-world incidence estimate, and it found the dominant platform bore only a minority of a tax nominally levied on it.[3] That is a genuine caution against assuming any revenue-based instrument (including Romer's) will land where intended.
• Capture and dissolve are not free lunches either way. Rent-targeting corporate taxes are politically fragile (universal ACE repeal) and their platform incidence is untested; interoperability raises no revenue and strains security (end-to-end encryption); data dividends have no working valuation method.[9][15]
• No instrument here approaches the land case. For land the wiki can point to fixed supply, non-shiftable incidence, and a century of evidence. For platforms it can point to a durable position (well evidenced), a disputed rent share (unresolved), and instruments that are either badly aimed (DST), untested (ACE-on-platforms, ad tax, data dividends), or unproven (DMA). The standing rule holds: never let the airtight land case lend its certainty to this frontier.

See Also

• Platform and Data Rents — the diagnosis these instruments act on · Geoism — the rent-domain program and its gradient
• Allowance for Corporate Equity · Cash-Flow Tax — the rent-only corporate bases
• Digital Services Taxes and Their Incidence · Romer's digital advertising tax — the two capture-side revenue taxes
• Furman Review · The Digital Markets Act — the dissolve pole, diagnosis and legislation
• Korinek & Stiglitz — AI and income distribution — the AI-rents theory
• Superstar Firms (Autor et al.) · Corporate profits increasingly rents — the efficiency counter-reading
• Harberger Tax (COST) · Radical Markets — the data-as-labor / self-assessment program
• Objection: Taxing quasi-rents kills innovation — the mandatory counterweight to every instrument above
• Quasi-Rent · Economic Rent · Pigouvian Taxation

Sources

(Empirical and design claims are cited to the external sources below; the wiki pages linked inline are navigation to the wiki's own treatment, which carries the same citations. Several sources were fetched and read in the sessions that built the research pages this synthesis draws on; DST and DMA figures come from working papers and primary legislation as noted.)

1. Tax Foundation (2020), "Who Will Ultimately Pay the UK Digital Services Tax? Amazon Passes the Cost Along to Sellers." Tax Foundation — used for the EC 2018 9.5%/23.2% effective-rate contrast and the statutory-vs-economic incidence framing. Via DST incidence.
2. Wei Cui & Nigar Hashimzade (2019), "The Digital Services Tax as a Tax on Location-Specific Rent," working paper. Allard PDF — used for the location-specific-rent rationale and the resource-royalty analogy that make the DST a deliberate rent-capture attempt.
3. Rohit Reddy Muddasani & Dominika Langenmayr (2025), "Navigating the Amazon: The Incidence of Digital Service Taxes," WU Vienna / KU Eichstätt / CESifo. WU PDF — used for the core incidence finding (Amazon raised fees ~half the DST; sellers largely passed the increase to consumers; large firms bear a small share). Working paper — final figures to be re-verified.
4. Anton Korinek & Joseph E. Stiglitz (2017), "Artificial Intelligence and Its Implications for Income Distribution and Unemployment," NBER Working Paper 24174. PDF — used for innovator rents, the "absorb all the rents" / non-distortionary-taxation-of-fixed-factors result, and the AI-coda framing (unrefereed working paper).
5. Amazon (2020) UK Seller Central fee-change notice; Andersen LLP / Accountancy Daily (2020) on the France precedent; Simon Sharwood, The Register (2 Sept 2020) on Google's DST Fee. Seller Central · The Register — used for the primary pass-through announcements (UK 2%, France 3%, Google 2% UK, Austria 5%, Turkey 7.5%).
6. Nicolas Crouzet & Janice Eberly (2019), "Understanding Weak Capital Investment: the Role of Market Concentration and Intangibles," NBER WP 25869. NBER — used for the intangibles counter-reading (a large share of high profit may be quasi-rent). Via crouzet-eberly-intangibles.
7. David Autor, David Dorn, Lawrence Katz, Christina Patterson & John Van Reenen (2020), "The Fall of the Labor Share and the Rise of Superstar Firms," QJE 135(2); NBER WP 23396. NBER PDF — used for the efficiency/productivity-divergence reading of concentration; and for the De Loecker–Eeckhout–Unger markup magnitudes (21%→61%, 1980–2016) carried at corporate profits increasingly rents.
8. Mirrlees et al. (2011), Tax by Design (ACE recommendation); IFS Capital Taxes Group (1991) for the ACE's origin; European ACE adoption-and-repeal record. IFS — used for the ACE design, the "only 'excess' returns" recommendation, and the universal-repeal political-economy record. Via ACE.
9. Meade Committee / IFS (1978), The Structure and Reform of Direct Taxation; Auerbach, Devereux, Keen & Vella (2017) on the DBCFT; Bond & Devereux (1995) on loss-refundability. Meade PDF — used for the cash-flow rent-only base, the DBCFT location-specific-rent argument, the 2017 US episode, and the symmetric-loss-treatment condition. Via cash-flow tax.
10. Julie-Anne Cronin, Emily Lin, Laura Power & Michael Cooper (2013), "Distributing the Corporate Income Tax," National Tax Journal 66(1); Laura Power & Austin Frerick (2016), OTA WP 111. Treasury — used for the supernormal-return share (~63%, rising toward ~75%) borne by shareholders.
11. IMF (2017), "Belgium: Selected Issues," Country Report 17/70 — used for the ~€4.5bn 2014 NID cost (~35% of corporate tax revenue) and ~40%-of-deductions-to-SPVs figures. Via ACE.
12. Clemens Fuest, Andreas Peichl & Sebastian Siegloch (2018), "Do Higher Corporate Taxes Reduce Wages? Micro Evidence from Germany," American Economic Review 108(2). AEA — used for the rent-sharing wage-incidence caveat (about half the burden reaches workers).
13. Imanol Arrieta-Ibarra, Leonard Goff, Diego Jiménez-Hernández, Jaron Lanier & E. Glen Weyl (2018), "Should We Treat Data as Labor? Moving Beyond 'Free'," AEA Papers and Proceedings 108, 38–42. AEA — used for the data-as-labor / data-monopsony diagnosis and the data-dividend remedy.
14. European Parliament and Council (2022), Regulation (EU) 2022/1925 (Digital Markets Act), CELEX 32022R1925. EUR-Lex — used for the "entrenched and durable position" gatekeeper test and the interoperability obligations (Arts. 6(4), 6(7), 7). Via DMA interoperability.
15. Michael Kades & Fiona Scott Morton (2020), "Interoperability as a competition remedy for digital networks," Washington Center for Equitable Growth WP. PDF — used for the "overcome the network effects" case, low incremental cost, and the "necessary but not sufficient" qualifier (Scott Morton discloses antitrust consulting).
16. Paul Romer (2019), "A Tax That Could Fix Big Tech," NYT [R1]; Romer (2021), "Taxing Digital Advertising," adtax.paulromer.net [R2]; Daron Acemoglu & Simon Johnson (2024), "The Urgent Need to Tax Digital Advertising," MIT [R3]. MIT PDF — used for the revenue-has-a-location argument, the progressive/split-up and subscription-escape design, the 59%/18% political-ad figures, and the Acemoglu–Johnson "sin tax" variant. Via Romer's digital advertising tax.

Digital Competition Expert Panel (Furman, chair; Coyle, Fletcher, Marsden, McAuley), Unlocking Digital Competition, HM Treasury, March 2019. PDF — used for the tipping / "network effects and returns to scale of data" diagnosis, the

90% UK-search figure, and the dissolve-side (interoperability, data mobility, merger) remedy set.