Radical Markets by E. Glen Weyl and Eric Posner is bursting with ideas. Like many readers, I admire the book’s ambition and continue to think about it. But this post lays out some concerns I have about one of their biggest ideas: Common Ownership Self-assessed Taxes (COST). In particular, I focus on it’s impact on the direction of investment and innovation.
(h/t: in many ways, this post is an extension of Tyler Cowen’s post on the problem of complementary assets under a COST system. Thanks to Nintil for putting together a list of criticisms so that I could see if others have already made these points.)
First a brief introduction to the COST idea. According to Weyl and Posner, a problem with our current system of private property is that assets (think land and capital) do not flow to their most productive uses. Weyl and Posner give the example of Elon Musk’s hyperloop and the struggles it would have to acquire necessary land rights to build a route along the West coast. A hold-out sitting on 50 acres in the proposed path of the hyperloop would be able to extract huge concessions from the hyperloop’s builder. Indeed, it could potentially extract nearly the entire value of the investment. Worse, the owner may be a curmudgeon who doesn’t like the project and cannot be bought off with any price. Because of the hold-out, the hyperloop may be delayed for decades or never built at all, even if it would enhance the welfare of millions of riders.
So, Weyl and Posner argue, for all private property (to some degree). Almost all assets are unique in some dimensions, rendering the owner a mini monopolist. Textbook monopolists restrict access/output in exchange for higher rents (just as in the above example), and these mini monopolies distort allocation across the entire economy.
To fix this distortion, Weyl and Posner propose a radical rethinking of private property. Ownership disappears, essentially to be replaced by continuous rental. For any asset, the current user assigns a valuation and pays an annual percentage of this valuation in the form of a tax. For example, in the example above, the current user of those 50 acres of prime real estate might assess the value of those acres at $500,000 and then pay a 7% tax per year ($35,000/year) for usage rights. The key idea is that valuations must be chosen carefully because another party may purchase usage rights from you at any time by paying your valuation. Thus, Hyperloop inc. could secure its required property rights for $500,000 without any costly litigation or bargaining. It might then raise it’s own valuation of the land to $10,000,000, paying an annual tax of $700,000.
The combination of the tax and purchase rights forces users to carefully assess the value they put on usage. Choose a low valuation to reduce your tax burden, and you risk losing usage rights to someone willing to pay your low valuation. Choose a high valuation to secure usage in perpetuity and you pay for it in the form of very high taxes.
This scheme neatly allocates assets to the user with the highest current valuation. This is also presumably the user who can put it to the most productive use. COST has the additional benefit of generating substantial tax revenues that can be returned to the public via social dividends or investment in public goods. Weyl and Posner lay out a host of other problems the scheme solves (they call it “killing a flock of birds with a stone”). And one of the strengths of the book is the extent to which it tries to head off the concerns you probably already have from this brief description. Indeed, I wouldn’t be surprised if the concerns brought up in this post have already been addressed in a research paper somewhere (although I have looked).
Dynamic vs. Static Welfare
A useful framework for thinking about these schemes is as a continuous auction. Usage rights are continuously sold to the highest bidder. Economists know auctions usually allocate goods to the person with the highest valuation (the Winner’s Curse suggests this may not be an ideal state of affairs, but that is not the topic of this post). Economists also know that in most auctions the price paid in expectation is not the highest valuation, but rather the expected value of the second-highest valuation.
The intuition is simple. The user of any asset simultaneously cares about minimizing their tax burden and not losing usage rights prematurely. The best way to do this is to choose a valuation equal to or slightly above the valuation of the user with the next highest valuation. Any less and you lose usage rights to the next highest bidder. Any more and you pay higher taxes than strictly necessary to secure usage.
Let’s be a bit more specific. Suppose V1 is the true valuation of the highest bidder and V2 the true valuation of the second-highest bidder. A valuation slightly above V2 secures usage rights at annual tax tV2, where t is the tax rate. The high bidder obtains consumer surplus of V1 less annual taxes of tV2.
This works great in a static setting where the supply of assets are fixed, but creates some strange incentives in a dynamic setting. In our current system of private property, an investment is attractive if it generate value to the investor at low cost. The value to the investor comes both from the private use value the investor obtains, but also from the future resale value of the asset to other potential users. Under COST, however, resale value is taxed. This incentivizes the creation of assets with high V1 relative to V2. In other words, it incentives the creation of assets that are highly valuable to the user, but have low value to others.
A simple example will illustrate. Suppose I plan to live in my house until I die (so I don’t worry about the value of resale). I am considering making upgrades to my kitchen. I have two possible upgrades in mind. Under upgrade #1, I install granite countertops. Under upgrade #2 I keep the linoleum countertops but raise the height of all the counters by 5 inches to better accommodate my 6'5" height. Suppose both upgrades raise my private valuation of the house from V1 to V1 + x and cost the same.
Because other people also like granite countertops, upgrade #1 raises the valuation of my house by the second-highest bidder to V2 + x. To keep the house I need to match this increase, which leads to an increase in my tax payments by tx. However, because most people are not as tall as me, upgrade #2 actually reduces the valuation of my house by the second-highest bidder to V2 - z. I can therefore lower my valuation by z without fearing a forced sale and this leads to a reduction in my tax burden by tz. Since my personal valuation is V1 + x under either upgrade, COST nudges me towards upgrade #2 which lowers my tax bill. When I eventually die though, the next owner of the house is more likely to have to endure extra expense to undo my “improvements”, compared to upgrade #1. The net effect of long-run welfare is not obviously positive.
Three More Examples
Countertops are a kind of silly example. So here I sketch out some more realistic and costly examples of the distortionary impact of maximizing the difference between V1 and V2.
(As an aside: Weyl and Posner think in terms of the probability that another bidder will come along with a higher valuation than me, rather than the framework of V1 and V2 sketched above. Their preferred tax rate is correlated with this probability, so that the tax rate is lower for goods with low turnover. This has the same incentive effect as my presentation; reducing the resale value of an asset to other lowers your tax burden)
First, COST would seem to nudge agents towards excessive customization of assets. Consider a firm that needs to make a large capital investment: say in a new headquarters, a new fleet of robots, a new software system, etc. The expected life of the asset is long, and so the resale value is not a major driver of the purchasing decision. There is a continuum of specialization and customization of the capital investment though. In general, buying standardized non-custom capital is cheapest: think boring commercial real estate, the “standard” line of robots, the Microsoft office suite, etc. Customized capital is more costly, but can be tailored to the needs of the firm. Maybe the building is given an unusual configuration to accommodate a specific production process. Maybe the robots have a specific tool permanently affixed instead of a modular one that can be easily removed. Maybe the company could benefit from some bespoke features added to Microsoft excel, but doesn’t really need powerpoint at all.
Just as in the countertop example, COST raises the value of the customized/non-standard capital investment. The resale value of customized capital is lower, and therefore so is the associated tax burden. This would push firms away from offering a small number of standardized products and towards offering many niche, specialty, and artisanal products.
But this would seem likely to impose costs on dynamic efficiency. Standardization is a way automation, returns to scale, and other productivity enhancing innovations are realized. Standardization also creates large markets for complementary products and services, and thereby incentivizes innovation. A shift to specific niche capital would seem likely to lower economy-wide productivity growth.
Other forms of “capital” cannot be transferred and therefore have zero resale value. Human capital is one such example considered by Weyl and Posner. This is often thought of as education or workplace training, but we can think of it as any form of knowledge that resides primarily in human brains and bodies and that cannot be easily codifiable. This includes new discoveries and innovations, especially if they are difficult to communicate, transport, or reverse engineer.
Because it cannot be sold and is therefore untaxed, COST would favor companies whose primary assets are in the form of human capital, and assets whose value only exists in conjunction with specific human capital. For example, suppose google’s supercomputer assets are only useful in conjunction with algorithmic know-how. Currently, google is quite open in how it shares algorithmic innovation — it publishes papers and releases open source code. In a world where its assets are taxed according to the value of the next highest bidder, it would seem to be in google’s interests to radically restrict this flow of information. It would not want to tell other people how to make good use of supercomputers. To a greater extent than present, it would want that information to remain locked away in the brains and social interactions of it’s own programmers.
I worry that this would be a general phenomenon. New ideas can be protected in a number of ways, but two important ones are intellectual property rights and opacity. By opacity, I mean the new idea is not immediately understood upon purchase of the associated good or service. The product might be difficult to reverse engineer, or the production process might require extensive finetuning and adjustment in ways not easily communicable. Secrecy (either proactive or merely a passive laziness about explaining the idea) also plays a role. Any business built on opaque knowledge can report a low valuation and pay low taxes, because rivals will be unable to make good use of its assets without the complementary knowledge, stored in non-transferable human capital.
In contrast, intellectual property rights have the nice feature of disclosing how an innovation works, allowing others to build on the knowledge in a non-infringing way. But under COST, disclosure raises the value of ideas and complementary assets that are easily communicable to other bidders, which in turn raises the tax burden of disclosing firms. It creates a wedge between the value of communicable and opaque knowledge.
It’s probably not a good idea to incentivize relatively more opaque knowledge production. I suspect opaque knowledge generates fewer spillovers and less cumulative innovation than communicable and readily revealed knowledge. Again, the risk is that COST would achieve static efficiency at the cost of slower productivity growth.
Last and most speculatively, I wonder if COST would induce firms and consumers to deliberately engineer wasteful frictions and barriers to resale into assets. Suppose some investment reduces your own valuation from V1 to V1- x, but reduces the valuation of the next highest bidder from V2 to V2- x- z. A deliberately wasteful investment like this would still reduce your annual tax burden by tz and therefore might be worth undertaking, even though it reduces your private valuation too.
Some speculative examples:
- An in house software package written in a firm specific programming language.
- A line of consumer electronics with custom, non-standard electrical plug configurations present only in the owner’s house.
- A set of industrial robots made out of a more expensive but more delicate material. They will break if they lift more weight than is strictly necessary for the current production process.
- A car with driver seat setting fixed permanently in one position.
The purpose of these modification would be to make resale difficult so that valuations and taxes can be kept low. It’s tempting to say we would just outlaw investments whose only purpose is to reduce resale value. But in each of these cases, it could be claimed the friction had inherent value of some kind (perhaps aesthetic).
Such a system would be wasteful. To the extent these frictions have higher fixed costs to introduce, even if they are cheaper (under COST) in the long-run, they might be preferentially adopted by the wealthy, entrenching inequality. And this kind of problem isn’t unthinkable. In the world of intellectual property, firms have often experimented with engineering costly and wasteful barriers to reuse: think DRM or terminator seed technology.
Like most readers, I was impressed with Radical Markets. After reading two books about the tenacity of inequality, (The Son Also Rises and The Great Leveler) it was refreshing to read a book with proposals that might rise to the challenge posed by inequality in capitalist societies. And I agree with Weyl and Posner’s that these ideas warrant further study and small scale testing. Indeed, it may be that the magnitude of the effects highlighted here are very small, and easily outweighed by the benefits. That’s an empirical question.
On the other hand, this isn’t my only concern with COST. I worry the wealthy will hire special firms to arbitrage price opportunities from individuals too mentally taxed to efficiently update their valuations. I worry peace of mind and security is itself a consumption good that COST largely destroys. I worry about the instability it would engender in our lives, especially during periods of rapid price inflation, such as housing booms. But as the authors frequently point out, our current way of organizing society was once a radical suggestion, likely subject to just as many hypothetical “what-abouts.” So I don’t mind giving it a try on a small scale. If we do, just make sure to look and see if the it starts to have funny effects on investment decisions.