A Harberger tax is a property arrangement defined by two rules: owners publicly self-assess the value of what they hold and pay a recurring tax on that declared value, and anyone may buy the asset at any time by paying the declared price. Understating the value invites a forced sale; overstating it inflates the tax bill. The combination pushes owners toward honest valuations and keeps every taxed asset perpetually for sale.

The mechanism was formalized by Eric Posner and E. Glen Weyl as the common ownership self-assessed tax (COST), and it has since been studied as a design for public resource licensing and adopted by blockchain developers as a template for always-for-sale digital assets.

Mechanism

Under conventional private property, an owner who does not wish to sell can hold out for a price far above their true valuation. A widely read 2018 explainer by Simon de la Rouviere argues that the Harberger forced-sale rule eliminates this seller holdout: because anyone can take the asset at its self-assessed price, owners can no longer demand inflated prices, removing a deadweight loss that otherwise blocks assets from flowing to more productive users — a central concern in the economics of transaction costs and property rights.

The gain is not free. As the same explainer summarizes:

Harberger tax reduces investment efficiency at a gain in allocative efficiency of ownership.

An owner who may be bought out at any moment has weaker incentives to improve the asset. Posner and Weyl's analysis of COST quantifies the trade: with tax rates of roughly 2.5–7% per year, the scheme captures an estimated 70–90% of the maximum possible allocative welfare gains, while investment distortions consume only about 10–20% of those gains.

Investment versus allocative efficiency

The formal treatment of this trade-off is given by E. Glen Weyl and Anthony Lee Zhang in Depreciating Licenses (2022). They show that no existing license design achieves the first-best on both margins: perpetual tradable licenses give owners strong incentives to invest but let them hold out for high prices in secondary markets, distorting reallocation, while short-term licenses allocate efficiently but destroy the incentive to invest.

Their proposed instrument, the depreciating license, is a perpetual license that decays at a chosen rate τ. Each period the administrator issues a fraction τ of new equity in the license and sells it in a second-price auction; the incumbent, holding the remaining 1−τ, either buys out the administrator's share at τ times the second-highest bid or sells to a higher bidder. The design is an explicit partial property right interpolating between short-term rental (τ = 1: fully allocatively efficient, no investment incentive) and perpetual ownership (τ = 0: fully investment efficient, distorted allocation).

The paper's main theorem is that the optimal depreciation rate lies strictly between these extremes: adding a small amount of depreciation to a perpetual license yields a first-order allocative welfare gain, through reduced markups, at only a second-order investment cost — and symmetrically at the rental end. Absent investment and private values, τ leaves total government revenue unchanged; when they are present, depreciation reduces revenue relative to perpetual sale, but the loss is second-order near τ = 0, so an administrator weighing revenue against welfare still optimally sets τ above zero. Extensions covering transaction costs, private-valued investments, additional competing buyers, and multi-period investment payoffs lower the optimal rate but never push it to a corner. Weyl and Zhang target real-world resource licensing — radio spectrum, fishing rights, pollution permits, mineral extraction rights, and land — and note that pricing the reclaimed share by competitive auction rather than owner self-assessment addresses a key critique of pure Harberger-style self-assessed taxes.

Blockchain experiments

De la Rouviere's explainer also argued that blockchains are natural infrastructure for Harberger property: property registries (cadastres) map well onto immutable ledgers, and smart contracts can enforce the tax and forced-sale rules automatically. It identified NFTs and virtual worlds such as Decentraland as closed-loop economies suited to low-stakes experimentation, sketched concrete designs — Harberger pixel maps, attention and advertising markets, and community access slots — and observed that tax revenue becomes programmable, flowing into community pools, bonding curves, or basic-income schemes rather than a traditional state treasury.

Art and patronage

In Radical Markets in the Arts (2018), published with the RadicalxChange movement, de la Rouviere proposed applying COST to artistic intellectual property: creators self-assess a work's value, pay tax on the declaration, and anyone can buy the work at the declared price. Always-for-sale intellectual property, he argued, circulates more efficiently than fixed copyright terms, whose expiration dates amount to arbitrary guesses about when a work should be exploited. The essay explored blockchain-based digital collectibles under COST — tradeable music rights and badges, concert-ticket and digital-art memorabilia, and CryptoKitties-style always-for-sale assets — and described Harberger-taxed patronage seats, in which holding a status position requires ongoing tax payments that flow to the artist. Tax revenue could recycle into the commons so that "the arts, more directly, would fund more of the arts," with quadratic funding (Liberal Radicalism) as a complementary mechanism amplifying many small contributions to niche work.

Choosing a tax rate

A follow-up 2019 post framed the tax rate as the key open design question for virtual collectibles and patronage markets. The academic literature recommends low rates, near an asset's expected turnover, because it targets physical assets where turnover is expensive and investment in the asset is desirable — assumptions that do not hold for virtual items. De la Rouviere argued that rates should instead vary by asset type: lower for items with use value, higher for pure collectibles and status assets. Extreme rates produce a pricing paradox: at 0.01% per annum, a patron wishing to give $100 a year must self-assess a prohibitive $1 million price, while at 10,000% per annum the price collapses to around $1, destroying any incentive to trade. For patronage markets he proposed roughly 100% per annum, so the annual donation equals the purchase price — an intuitive figure for patrons. The aim of such a rate is to direct capital to beneficiaries while retaining enough speculative interest to keep the market alive, though the post acknowledged unresolved tensions between altruism, status incentives, and market mechanics, and suggested simulation tooling such as cadCAD to model outcomes.

Relevance to Caper

Harberger-style mechanisms are candidate tools for DAOs that must allocate scarce rights — featured placements, named roles, or other exclusive positions — without running repeated manual auctions: self-assessed pricing keeps such rights permanently contestable and converts their occupancy into a continuous revenue stream for the community. For communities on Caper, adopting such a mechanism would be a question of governance design and treasury management. Caper does not implement Harberger taxes; they are described here as part of the design space available to on-chain communities.

References

1. Eric A. Posner and E. Glen Weyl (2017). Property Is Only Another Name for Monopoly. Journal of Legal Analysis 9(1): 51–123.
2. Simon de la Rouviere (2018). What is Harberger Tax & Where Does The Blockchain Fit In?. Medium.
3. E. Glen Weyl and Anthony Lee Zhang (2022). Depreciating Licenses. American Economic Journal: Economic Policy 14(3): 422–448.
4. Simon de la Rouviere (2018). Radical Markets in the Arts. RadicalxChange (Medium).
5. Simon de la Rouviere (2019). Exploring Harberger Tax Rates in Virtual Collectibles & Patronage Markets. simondlr.com.