A Philosophical Look at Harberger Taxation
Eric Posner and Glen Weyl’s recent book Radical Markets has stimulated a wide-ranging conversation about alternative property rights schemes. Inspired by economic thinkers such as Bill Vickrey and Henry George, the authors propose fundamental changes to property that would, in theory, increase economic efficiency while also sharing wealth more equally.
Their ideas have captured the imaginations of blockchain technologists in particular. That is because while it is hard to change real-world property law, it is easy to rewrite the rules governing ownership of crypto-assets. Thus, there is a fascinating possibility of architecting crypto-asset ownership schemes that are simultaneously more efficient and more egalitarian than the Common Law scheme governing the real economy.
This piece is short but somewhat dense. In the hopes of making it as accessible as possible, I will race through the basic ideas from Radical Markets. I will then take a sharp philosophical turn, looking at the ideas in Radical Markets through the lens of the theory of just ownership. Finally, I will outline a thought experiment designed to stimulate further conversation about Harberger taxation schemes.
1. A Quick and Dirty Introduction to Harberger Taxation
Harberger taxation (“HT”) is a property taxation scheme designed to weaken monopoly power and make markets more efficient. It has not been tried before. At its core:
- It lets property holders self-assess their assets for taxation purposes, reporting those values to a public ledger; and also
- Requires property holders to sell (within a reasonable time) their assets (or tranches of their assets that they have defined) at the prices they report.
Notice that these two features create opposite incentives. The “self-assessment” feature incentivizes under-assessment. And the “requirement-to-sell” feature incentivizes over-assessment.
One of the most important insights in Posner and Weyl’s book is this: If a Harberger tax on an asset were set precisely at the probability of that asset finding a buyer in a given period (the “turnover rate”), then the two opposite incentives would fall into perfect equipoise. Thus, property holders would be best served by truthfully reporting the exact prices for which they would be willing to sell their assets.
The clearest consequence of this scheme would be making it expensive to possess assets. Because people wouldn’t have as much of an incentive to hoard them, assets would flow more efficiently into the possession of those willing to pay the most for them at any point in time.
This might sound like it would favor the rich, but that is a misunderstanding, for two reasons. First, HT would function as a wealth tax, making it costly to own a lot of stuff, and reducing the returns to speculation. Second — and this is key — the huge revenue streams generated by HT could (and for reasons that will become clear, should) be distributed equally to everyone.
For those with little net worth, their distributions would cover the Harberger taxes on their assets, with plenty of room to spare. Thus, increasing asset values would be cause for common celebration, not division.
But most interestingly of all, HT would undermine monopolies. Owners of scarce, desirable assets would have to pay taxes on the true value of their monopolies, thus giving back to society the value of the inefficiency generated by their market power. In this way one of the key hindrances to economic growth might vanish.
Of course, it isn’t quite a free lunch. A more formal way of understanding HT’s appeal is that it would increase “allocative efficiency”, by causing assets to end up in the hands of people who need them or are putting them to profitable use, rather than people who are just sitting on them. But this would come at a cost to “investment efficiency”. In other words, HT would decrease incentives for people to improve their property. Very high Harberger taxes would harm investment efficiency so much that gains in allocative efficiency would be erased.
Thus, moderation is important. But at the margin, Harberger taxes would generate allocative efficiencies that would exceed harms to investment efficiency. This principle is illustrated in the following graph (where “tau” represents the Harberger tax rate and the red, blue, and purple lines represent allocative, investment, and overall market efficiency):
This is pretty promising. How many taxes figure to increase market efficiency?
Of course, theory is theory. There are obvious practical obstacles to implementing HT. Empirical studies are necessary.
People are therefore beginning to think about how to use blockchains to prototype experimental HT systems. These need not involve actual government taxes, obviously. They will simply be “sandboxes”, in which assets are held as non-fungible tokens representing Harberger “licenses”. Instead of paying a government to retain property, license holders would simply pay a license issuer to retain their license. And instead of distributing the HT returns to all the members of a society, the license issuer might distribute them to all the participants in the ecosystem. Still, such projects might generate valuable data concerning the workability of HT-like schemes.
Turnover Rates are a “Backstop” for HT
A Harberger tax at the turnover rate would capture a substantial part of many assets’ market value. It is helpful to see the magnitude of this effect. The following formula roughly indicates the impact on prices:
(Asset Value Post-HT) / (Asset Value Pre-HT) =
(Discount or Interest Rate) / (Turnover or HT Rate + Discount or Interest Rate)
Thus, for example, with a discount/interest rate of .045, and a HT rate of .07, a taxed asset would be worth about 39% of its previous value ((.045) / (.045 + .07) = 0.391).
The diminution in asset values would be smaller where interest rates were high, and for depreciating assets. Nonetheless, turnover-rate Harberger taxes would take a large bite out of the prices of many durable assets.
Recall that turnover rates constitute an upper limit on HT. Above that rate, HT would be self-defeating: it would cease to improve allocative efficiency, because the incentive to under-assess property for tax reasons would predominate over the incentive to over-assess it to keep it or fetch a high price. Assets would become hot-potatoes.
2. A Philosophical Digression
Property is unavoidably political (and therefore philosophical). If we want to prototype HT in blockchain ecosystems, we have to think about values, and understand the normative implications of our decisions. So let’s explore some deeper aspects of HT.
What Kinds of Things Should be Owned, and Why?
It is easy to get lost in efficiency analyses, forgetting that taxation schemes like HT must also be considered from the perspective of justice. Efficiency arguments, after all, did not suffice to justify trickle-down economics, and they do not suffice to justify HT.
As we saw, high Harberger taxes would (a) ensure the efficient of allocation of capital, but also (b) siphon away much of its financial value. It is sloppy to suggest that this would always increase justice, and equally sloppy to suggest it would never increase justice. The right answer is going to be a messy in-between: HT is just, but only at the proper rate, and the proper rate depends on the asset in question.
In a word, since turnover-rate HT amounts to a kind of maximum efficient appropriation, it is most justly applied to assets that — ignoring efficiency — ought to be owned in common. Conversely, barring efficiency justifications, it should not apply to assets that are justly privately held. We therefore find ourselves shunted to a deeper question: what should be privately held on grounds of justice (as opposed to mere efficiency), and what shouldn’t be? We have arrived at the murky shores of just ownership theory.
The Key Distinction: Natural Versus Artificial Capital
A century ago, economists inspired by Henry George drew a distinction between “natural” and “artificial” capital — that is, the products of nature and those of labor. They suggested that rights of private ownership attached more properly to artificial capital than to natural capital. The distinction between these two species of capital is often fuzzy in practice, but it is clear in theory, and important to understand. Let’s explore why.
The case against private ownership of natural capital is rooted in the observation that nobody’s labor created the spoils of nature. Gold, oil, land, and other natural resources exist independently of humanity. So it is illogical for the law to reward private parties for their existence. To be sure, people must procure natural resources. That labor must be compensated, and should be incentivized insofar as it is socially efficient. But, when the market value of a natural thing grossly exceeds the value of the labor needed to procure it, does justice entitle anyone to capture that surplus? What if we’re talking about the last water hole in a drought-stricken country? A massive oil field discovered by chance underneath a backyard?
Contrast natural capital with objects in which artificial capital predominates, like paintings. A painting derives most of its value from its labor, and very little from its constituent natural materials like oil, cotton, and wood. Therefore a painter, far more than a mine owner, enhances the value of naturally-existing materials, and deserves a reward for the sheer existence of her wares. Her right to sell her painting for as high a price as she likes seems much more justified. Likewise with other objects of artificial capital such as sailboats, software, and skyscrapers.
Thus, Georgist economists proposed that common ownership schemes should be applied to natural capital, but not artificial capital. This distinction has efficiency grounds: granting the privilege of private property helps incentivize people to make art, but does not result in the production of additional gold or land. But it also has justice grounds: the privilege of private property may be a just reward for bringing artificial capital into existence, but it is not clear why those who gain possession of naturally-existing things deserve quite the same reward.
Even Confirmed Libertarians Get a Little Squirrely About the Justifications of Private Property in Natural Capital
Henry George is not everyone’s cup of tea. So let us examine the views of two thinkers with stronger laissez-faire bona fides: John Locke and Robert Nozick.
It turns out that there is a clear recognition at the very core of the liberal/libertarian tradition that (a) natural and artificial capital are different, and that (b) property rights in natural capital are much harder to justify. This recognition flows from what Nozick called the Lockean proviso.
Nozick based his theory of just ownership on the insight that property justly acquired, and thereafter justly transferred, is justly held. In defining “just acquisition”, he looked to John Locke. Locke famously contemplated how ownership interests could arise in unowned things, like unclaimed land. He suggested that such interests arose when a person appropriated unowned land and “mixed his labor” with it (i.e., improved it). But he added the “proviso” that such an appropriation must always leave “enough and as good” in the unowned domain. Otherwise, the appropriation would prejudice others, and not be just or valid.
Locke’s proviso is a serious problem for Nozick and anyone else who wants to endorse private property interests in natural capital. Because the reserves of natural capital — land, gold, and the like — are inherently finite. Even back in Locke’s time, when the wilderness must have seemed almost endless, there surely wasn’t always “enough and as good” left over for just anyone, especially after the choice farmland, just outside of town, had been claimed by powerful local lords. And that problem has worsened as the world has grown more crowded.
Nozick argued that “each owner’s title to his holding includes the historical shadow of the Lockean proviso on appropriation.” He took the limit quite seriously:
“Once it is known that a person’s ownership runs afoul of the Lockean proviso, there are stringent limits on what he may do with (what is difficult any longer unreservedly to call) his property. A person may not appropriate the only water hole in the desert and charge what he will. Nor may he charge what he will if he possesses one, and unfortunately it happens that all the water holes in the desert dry up, except for his. This unfortunate circumstance, admittedly no fault of his, brings into operation the Lockean proviso and limits his property rights.” 
Thus, even from a libertarian perspective, doubts arise about the justness of property interests in natural capital whose scarcity results in market power. Friedrich Hayek himself called Georgism “the most seductive and plausible of all socialist schemes,” which has to count as a compliment, coming from him.  Hayek argued that common ownership of natural capital would make sense, if only it were possible to separate the “natural” and “artificial” components of assets’ values.
I think Hayek was a little too pessimistic. But he correctly identified the problem.
3. The Problem and a Thought
The Problem: Artificial and Natural Capital are Commingled, and Teasing Them Apart is Hard
Most capital, of course, is not entirely artificial or natural. It’s both, annoyingly mixed up together.
The commingled-ness of natural and artificial capital is a vexing problem. It vexed Henry George and his followers. It also vexes our present economy, which suffers waste from monopolistic control of natural capital. And it vexes Posner and Weyl. But they propose an interesting workaround.
Recall that in a world where the natural and artificial components of capital were magically unmixed, we might impose a Harberger tax near the turnover rate on natural capital, and a Harberger tax near zero on artificial capital. But, recognizing that we do not live in such an ideal world, Posner and Weyl propose to set HT rates at varying percentages of the turnover rate for different assets, depending on those assets’ investment elasticities. That is, assets whose value increases more readily with investment should generally enjoy lower HT relative to their turnover rate, to facilitate investment.
Determining investment elasticities for different assets or asset classes would be a data-intensive affair. But with good data, it could be done with acceptable accuracy. Implicitly, this methodology suggests a provocative definition of artificial capital. Namely, that artificial capital is value that emerges in response to incentives. There is much to say about this. But for now let us turn our attention to what is at stake in the determination of the HT rate.
Beyond the Turnover Rate: The Meaning of the HT Rate
Where assets with differing natural and artificial components have the same HT rates, unfairness may emerge. To illustrate, picture two adjacent parcels of real estate. Let us stipulate that they fall within the same putative asset class, and therefore pay the same HT rate — a certain percentage of their turnover rates.
Yet the two parcels are not quite the same. Parcel A derives most of its value from a fine structure on it. Parcel B has a decrepit structure on it, and derives most of its value from the land beneath. Their identical HT rate will result in prejudice: A high rate would unfairly penalize the improved property, while a low rate would harm society by granting an inefficient reward to the possessor of the unimproved property.
In other words, even though these two parcels look similar in many respects, parcel A is much more like artificial artificial capital than parcel B, which is essentially natural capital.
Yet, another problem emerges if we try to assign a bespoke HT rate to each property based on their prospective investment elasticities. Because we might notice that the value of the less-improved property (when viewed as a unit with its improvements) will be more responsive to investment than the more-improved one. Thus, our bespoke HT rate might turn out slightly lower for the dilapidated property than for its neighbor. This difference would exacerbate the injustice described above, even as it slightly increased investment efficiency.
The lesson is simple but subtle: we might have to resist the perfectionist urge. No matter how tightly we zoom our microscope in to economic reality, we will not discover a perfect alignment of efficiency and justice. Worse, HT rates that are sensitive to the particular histories of each asset become gameable. We can and should accept that HT rates set in the vicinity of the efficient rate, and applied to asset classes rather than individual assets, are a vast improvement over HT rates of zero, in terms of both efficiency and justice.
For this reason and others, the quest for a formula yielding perfectly just HT rates tailored to individual assets is somewhat Platonic. But I want to explore it anyway, because it throws into relief the justice concerns underlying all HT schemes. Thus, if we muted concerns about gameability and tried to derive just HT rates tailored to individual assets, what factors might we look at?
To this end, a thought.
A Thought: Artificial Capital Decays Into Natural Capital
As time passes, artificial capital starts to resemble natural capital.
Think of a new boat, built yesterday. Now think of the Parthenon. The labor that made the boat can and should be rewarded. It makes sense for the spoils of boat ownership to accrue to its builder. But the labor that made the Parthenon has dissolved into the mists of time. There is no sense rewarding it. We simply find the building in our environment, like an ocean, a mountain, or a nickel deposit. Whoever possesses it deserves an incentive for its upkeep, but not a reward for its existence. Any profits from Parthenon ownership ought to be distributed broadly, and not end up in any particular pocket. Thus, unlike the new boat, the Parthenon ought to be treated like natural capital. Yet it is the product of human labor; when erected, it was the very epitome of artificial capital.
The idea that artificial capital “decays” into natural capital offers some purchase on the problem of defining labor’s contribution to a complex asset’s value.
Consider an asset less ancient than the Parthenon: the Empire State Building. Since its completion in 1931, it has changed hands many times. Each time it is transacted, it grows more attenuated from the labor of its construction, and its resemblance to natural capital therefore increases. Indeed, from the perspective of its current owners, it may as well be natural capital, yet they enjoy precisely the same rights as its 1931 builders. Perhaps the private property right should have suffered a partial “decay”, in the form of a Harberger tax gradually approaching the turnover rate.
This idea of private property decaying from artificial to natural capital (that is, from private to public property) is not without precedent. In the context of real estate, historical preservation ordinances reflect the public’s gradual acquisition of a rightful interest. Intellectual property law provides an even clearer example: Everyone recognizes that works of information must eventually cease to be things with exclusive owners. Imagine the harm to society if, for example, Tesla’s patent for alternating current never expired. Some investor would be forever collecting useless rents from society.
So why does the right of private property in so many kinds of assets survive perpetually, no matter the attenuation between owner and creator? Might there be some way of charting a “decay function” in the private property right?
A Decay Function For Artificial Capital
In principle, such a decay function should reflect three ideas:
- artificial capital should have a Harberger tax rate near zero
- natural capital should have a Harberger tax near the turnover rate
- artificial capital becomes more like natural capital as more time passes and/or it changes hands more times
Tentatively, it might look something like the following:
Harberger Tax =
Asset Value * (transfers since creation / years since creation) * (1 - (1 / (transfers since creation + years since creation)))
The first two terms (asset value * transfers/years) simply give us the turnover-rate Harberger tax.
The third term introduces a kind of decaying “creator’s incentive.” Thus, when a piece of capital is created, it is initially treated as pure artificial capital, with a Harberger tax rate of zero. As time and transactions accumulate, the Harberger tax approaches the turnover rate.
To illustrate: suppose an asset is created. Within a year of its creation, it is purchased twice. A nonzero Harberger tax takes effect beginning with this third owner. For this owner, the Harberger tax would be half the turnover rate (two sales since creation + zero years since creation = 2 → (1 - ½) = 50%). The first anniversary of the asset’s creation is reached with no further transactions. Now the Harberger tax increases to two thirds the turnover rate (two sales since creation + one year since creation = 3 → (1 - ⅓) = 67%).
By the time the asset has changed hands many times, and much time has passed, the creator and her most proximate patrons have already had their rewards, and the asset may be regarded essentially as natural capital.
Do not forget that the “transfers” and “years” variables in the third term of the equation could easily be adjusted with multipliers, the proper size of which is a political question.
This is just a back-of-the envelope idea. It is intended as an intuition pump with respect to the justice (rather than efficiency) concerns addressed by HT, and it has complex implications that I cannot fully explore here. But I hope it provokes conversations and ideas about refinements to HT schemes.
How Does this Address the Problem of Commingled Capital?
By articulating a process by which artificial capital decays into natural capital, we gain a tool for determining the ratio of natural to artificial capital in mixed assets such as improved real estate.
An illustration. Suppose I buy a parcel of raw land (natural capital) for $1m, and provably invest $500k in it by erecting a building (artificial capital) upon it. Going forward, the property could be taxed in two “buckets”: two thirds of its value are natural capital, and one third is artificial capital. As the first owner, I will not pay any Harberger taxes on the artificial capital bucket (until some time has passed). Thus, if I increase my self assessment of the property to $1.8m, I will pay a turnover-rate Harberger tax on $1.2m, and no Harberger tax on 600k. As time passes and/or successive owners take possession, a Harberger tax on the artificial component of the property will kick in, as explained above.
Questions remain. After time has passed and the property has changed hands, how can we know whether the value of the parcel is still composed two-thirds of natural capital and one-third of artificial? (After all, if the property is purchased as a tear-down, then clearly no part of its value is artificial capital.) Should gifts, or devises to heirs count as transactions? How does this approach to determining capital’s “artificiality” dovetail with the concept of investment elasticity? These issues, and many others, are beyond the scope of this short piece.
The decay formula above is intended mainly as a thought experiment. The HT schemes described by Posner and Weyl strike an excellent balance between the values of justice, efficiency, and non-gameability. In other words, linking HT to the investment elasticities of asset classes has important practical advantages over the approach I explore here. I nonetheless find it useful and stimulating to formally consider the factors that attenuate capital from labor on an asset-by-asset basis. This inquiry, nowhere near exhausted by these brief thoughts, might sharpen our view of HT’s moral foundations.
Hopefully, this piece will inspire further conversations about the design of just HT schemes. I look forward to experimenting with such schemes in blockchain-based sandboxes soon.
— — —
 Robert Nozick, Anarchy, State, and Utopia (1974), p. 180.
 Robert V. Andelson, On Separating the Landowner’s Earned and Unearned Increment: A Georgist Rejoinder to F. A. Hayek. The American Journal of Economics and Sociology, Vol. 59, №1 (Jan., 2000), pp. 109–117.
*Special thanks to Glen Weyl for invaluable comments on an earlier draft of this piece.
Blockchains are Governance
A Closer Look at the Innate Governance Models of Blockchain-based Technologies