Financial Explanations of Art

Introduction

In a previous post, we explored some functional theories of art. However, we left open the question of specific financializations. In this post I’ll try to understand institutions of art, like auctions, museums, etc. to understand how the formal institutions actually impose a specific financial value on an object.

How to Commit Tax Fraud with Art

Imagine the following scenario:

1. There are a group of \(n\) “capitalists”. Each capitalist has possession of a large pot of money.
2. Together, they buy an auction house, splitting the price equally among themselves.
3. Together, they also buy a museum (or otherwise take control of it by possessing the board seats), splitting the price equally among themselves¹.
4. Each capitalist in turn takes some small amount of money ($\(x\)) out of their pot and commissions a piece of art.
5. The auction house auctions off the art to the other capitalists. One purchases the art for some price ($\(y\)), with \(y >> x\).
6. This repeats until each capitalist has commissioned, sold, and bought a piece of art. We can assume that they do this in a round-robin way, so that each capitalist ends up with a piece of art that they commissioned, and a piece of art that they bought from another capitalist, and \(y - x\) in their bank account.
7. Finally, the capitalists all donate their art to the museum (or sell it to the market), each receiving a tax deduction for the value of the art they donated, now valued at \(y\).

What happened? The capitalists spent \(x\) each, but got \(y\) in tax deductions! The entire process is a washing machine where each participant gets \(y*t-x\) profits for free (\(t\) is some tax factor). Obviously, due to the collusion among the parties, the above scenario is illegal and would be considered a form of tax fraud. But the example illustrates that the institutions around art seem to create value ex nihilo, assuming the right coordination.

Decentralized Version

Do we actually need the parties to collude? Or can we get a similar effect in a more decentralized way?

Instead of a cabal of capitalists, we can imagine a more decentralized market, with multiple artists, multiple auction houses, and multiple museums. The same effect can be achieved if the artists are able to coordinate with the auction houses and museums to create a similar cycle of commissioning, selling, and donating art.

This ties back to the central claim in the previous essay: art value emerges from the ability of agents to predict and coordinate on future coordination. The cabal example makes this explicit by brute force. But the deeper point is that the same structure can arise without explicit agreement, as long as agents are sufficiently good at anticipating one another. Instead, we have a system of mutually observing agents (artists, collectors, auction houses, museums) who are all trying to predict what others will do and coordinate on that. The value of art emerges not from nothing, but from the coordination and prediction of coordination, even without explicit agreements.

One interesting question here is whether there are bagholders. In the cabal example, there are no bagholders, since each capitalist ends up with a piece of art that they commissioned (which they value at \(x\)) and a piece of art that they bought (which they value at \(y\)), and they also get \(y*t-x\) in profits. In the decentralized version, however, there could be bagholders if some artists or collectors end up with pieces of art that they overpaid for or that don’t appreciate in value as much as they expected. This could happen if the coordination among agents is not perfect, or if some agents are less skilled at predicting future coordination than others.

Based on this analysis, we can think of the tax evasion scheme as a degenerate case of the ideas from functional theories of art, where the cabal uses pure coordination to create value in a way that is completely detached from any properties of the art correlating with its persistence.

Why Auctions and Museums?

An auction turns dispersed predictions into a single public signal. Collectors may privately believe that a work will matter in the future, but the auction forces them to express that belief through capital. A bid is a costly prediction. The final price becomes a public, timestamped record of how strongly agents were willing to coordinate on that object. What’s more, in order to make a bid, an agent must have access to a certain amount of capital, which creates a financial barrier to entry.

A museum, on the other hand, serves as a public institution that can confer legitimacy and prestige on certain works of art. By donating a piece of art to a museum, an artist or collector can signal that they believe the work has lasting value. The museum’s endorsement can increase the perceived value of the art, which in turn can affect its market price. Furthermore, the museum actively stores, curates,and promotes certain works, which can shape public perception and drive demand.

NFTs

We can also think about NFTs in this framework. An NFT is essentially a digital certificate of ownership for a piece of digital art. The common theory of NFTs is that they are a way to create “scarcity” and ownership in the digital realm. However, this makes no sense as low supply cannot create value on its own; only the combination of supply and demand can create value. The demand for NFTs is driven by the same kind of coordination and prediction as traditional art markets, but in a more decentralized and digital environment.

The NFT market operates on a blockchain, which serves as a public ledger that records ownership and transactions, similar to an auction. This creates a similar dynamic of predictions and coordination among collectors, artists, and platforms. However, the NFT market is more volatile and less established than traditional art markets, which can lead to more speculative behavior and price fluctuations. This is because it is still in the early stages of development, and there is a lot of uncertainty about its long-term value and legitimacy. In fact, this uncertainty about the future coordination could actually reflect less underlying value, since in this framework the value comes from expected future coordination (though the high variance may also be a feature in some portfolios). And the paranoid among us might speculate that the “pump-and-dump” dynamics of the NFT boom were driven also in part by the fact that the market was being manipulated by a small group of insiders (as in our original art scheme) who were able to coordinate on pumping the value of certain NFTs before dumping into the broader market.

NFTs had a transaction ledger (the blockchain) and auction-like marketplaces, but lacked the rest of the institutional stack that stabilizes traditional art coordination. There were no credible curators staking long-term reputations on specific works, no institutions whose function was to remove works from circulation and commit to their lasting significance, and no critical infrastructure operating on decadal timescales. Even worse, the community was openly financialized, with participants discussing “flipping” assets rather than maintaining even the pretense of aesthetic evaluation. This transparency may have actively undermined coordination, as it made the market resemble the cabal scheme above rather than a decentralized beauty contest. The NFT ecosystem had mechanisms for registering bids but almost no mechanisms for producing costly, long-horizon commitments, which are what stabilize coordination equilibria over time.

Conclusion and Open Questions

The auction-museum stack is just one instance of a general pattern. Auctions perform public price discovery, forcing agents to express predictions through capital. Museums perform long-horizon commitment by removing objects from circulation and anchoring their significance across time. Any asset whose value exceeds direct use value requires institutions performing these two functions. In theory, this should equally describe any speculative/Keynesian asset, including fiat currency, gold, blue-chip stocks above book value, Bitcoin, baseball cards, and real estate (modulo use value). The specific institutions may differ, but the underlying functions of price discovery and commitment are necessary for any asset to have value beyond its direct use.

For example, fiat currency has central banks (commitment) and foreign exchange markets (price discovery). Equities have exchanges (price discovery) and auditors, regulators, and index funds (commitment and legitimation). Real estate has appraisals and comparable sales (price discovery) and title registries, zoning, and mortgage markets (commitment). Gold has spot markets (price discovery) and central bank reserves (commitment). Bitcoin has exchanges (price discovery) but, much like NFTs, weak commitment institutions, which may partly explain its volatility.

The tax evasion scheme from the beginning of this essay is the degenerate case, where there is pure coordination with no institutional friction and no connection to underlying properties of the asset. Real markets sit on a spectrum between that scheme and pure use-value pricing. The thickness of the institutional stack determines where on the spectrum the market lies. NFTs are a useful example because they show what happens when the price-discovery machinery appears before the symbol-survival machinery is mature.

If value-above-use is sustained by coordination, and coordination is bounded by available capital, then we should expect art market “inflation” (more objects valued highly, and higher valuations for existing objects) to correlate with growth in the collector/institution base and available liquidity, not surges in artistic quality (however defined). This seems empirically true, since art market booms track wealth concentration and expansion of the gallery/auction/museum ecosystem. Also, asset classes with thinner institutional stacks should exhibit higher price volatility, controlling for fundamentals. As examples, we could investigate NFTs vs. traditional art, different national art markets with different institutional densities (number of auction houses, museums, MFA programs, critics per capita), etc.

Beyond the financial value questions, there is an interesting question here related to symbol production. In the above scenario, a group of people create a symbol (the art) that is valuable because they coordinate on it. How many symbols can they create? Is there a limit to how many valuable symbols can be created through this process?

One plausible candidate for the binding constraint is attention rather than capital. Capital is fungible and expandable (you can print money, take on leverage). Attention isn’t. A collector can only track so many artists, a critic can only endorse so many works, a museum can only mount so many exhibitions. If the carrying capacity of a coordination network is bounded by attention rather than capital, then we’d expect the number of “valuable” symbols to scale sublinearly with wealth but roughly linearly with the number of active participants (curators, critics, collectors). This might explain why art market inflation tends to concentrate value in fewer works rather than spreading it across more: when capital grows faster than attention, the excess capital competes for the same attention-limited set of coordination points.

There are other questions as well. How exactly does the value of the art relate to the amount of coordination, the amount of capital liquidity, the attention, or the long-term properties of the art’s persistence? Can we tie this to information-theoretic principles? Additionally, if instead of a “symbol” we view the art as a vector reprensenting an object in the space of possible objects, how does the structure of that space affect the dynamics of value creation through coordination? Can we connect the space of objects to the institutions used to explore it? These are all interesting questions for future exploration.

Footnotes

1. This can also be a gallery or any other institution that can confer legitimacy and prestige on art, but for simplicity we’ll just talk about museums. Galleries are more complicated because they perform both the auction and museum functions, which can create some interesting dynamics that we won’t get into here.