Framing complex financial instruments more clearly as products may help to spark rational patterns of consumer behavior

Introduction

It is well known in crypto that the term “stablecoin” is a misnomer. For one, these assets can vary widely in their stability. Secondly, the term “stablecoin” refers to many distinct techniques for maintaining some value or “peg” inside a crypto ecosystem. Some stablecoins are sustained by huge vaults of cash, others by elaborate computational and financial engineering. Retail investors who are interested in stablecoins face the challenge of navigating this complex landscape. They must assess risk with their hard-earned savings.

Managing this risk amidst so much complexity is not just about information. Information isn’t helpful if we are not inclined to engage with and critically assess that information. To spark this kind of engagement, it may help to emphasize that crypto schemes are financial products. When consumers make a product purchase, they can engage in a number of thought processes. Researchers have parsed these thought processes into distinct stages: defining the purchasing problem, product category and information search, evaluation of alternatives, and more.

When we also frame stablecoins as products, it may help to spark similar patterns of consumer behavior. Products compete, have more or less favorable traits, so thinking in these terms could encourage careful weighing of pros and cons. This careful product assessment can happen when we purchase everything from undergarments to a new laptop. Yet the term “stablecoin” may have led some to tragically miss out on this assessment, and dangerously weigh only one feature (such as interest rate on a lending platform).

The goal of this post is to discuss various properties of stablecoins as products. I write not as an expert in this macroeconomic landscape, but as a data scientist, consumer and market participant myself (see disclosures below). So I’m fascinated by this complexity. First, I share some history, and the post ends on exploring implications of the product concept.

Definition + Some History

The most common stablecoin is one that bears a 1-to-1 relationship to the US dollar. 1 unit of the crypto asset is equal to $1 US dollar. The benefits of this setup are obvious—stability of the US dollar for trading, but the convenience of having it live in a specific crypto ecosystem. The first crypto stablecoins emerged around 2014, including the famous Tether stablecoin (known as “Realcoin” back then).

The “Second Bitcoin White Paper” (MasterCoin) by Willett, 2012

The stablecoin concept emerged well before all the recent technical developments, illustrated in the quote above in the MasterCoin white paper. In a 2012 article, George Selgin described a potential asset class that would “bear a perfectly-elastic supply schedule, so as to preserve a stable purchasing power.” I learned about Selgin’s paper in this fantastic interview with Nic Carter. Nic himself has an informative and detailed “crypto-dollarization miniseries” as part of the “On the Brink” podcast. The miniseries explores how dollar-denominated stablecoins might be the first true “killer app” of public blockchains. The miniseries covers extensive examples, technical details and applications (such as cash disbursements to healthcare workers in Venezuela).

As presaged by such discussions, stablecoins are now a massive component of the overall crypto space. Alex Svanevik recently observed that stablecoins pegged to the US dollar are among the top in terms of market capitalization. Alex comically quipped, “3 out of top 7 cryptocurrencies are now the US dollar.” At the time of this writing, 3 of the top 5 in terms of 24-hour volume are the US dollar:

Source: CoinGecko

This volume and the general attractiveness of stablecoins are a reflection of the strength of the US dollar. Traders can anchor to the US dollar without settling in actual US dollars.

But what goes into a stablecoin? Their internal details matter greatly. The recent collapse of Terra’s LUNA and UST is a reminder that these various species of stablecoin have distinct features and it is essential to assess and critique each thoroughly — and inform retail investors about them. As an asset class, stablecoins are unfortunately lumped together in terms of their label and best-known feature: “stability.” But underneath each stablecoin is considerable technical detail. In the next section, I draw from a superb new resource that aims to do this kind of educational work: stablecoins.wtf.

Main Stablecoin Categories

An excellent resource for tracking and learning about stablecoins is stablecoins.wtf, by @dennis_zoma and @mike1third. The website features concise, clear explanations of the major categories of stablecoin. In this scheme, stablecoins are classed under three categories: fiat-backed, crypto-backed, and algorithmic (and possibly a fourth, a hybrid of these approaches). Let’s briefly consider each category.

Screenshot of stablecoins.wtf interface

Fiat-backed: A fiat-backed stablecoin maintains some correspondence between the stablecoin issued and assets backing it, conventionally USD holdings. Tether famously uses a mixture of assets to back its $1 peg, including other digital tokens. An ideal fiat-backed stablecoin has at least a 1:1 ratio between stablecoin issued and USD-equivalent in reserve. Stablecoin holders can be confident that 1 unit of their stablecoin can always be redeemed for $1 in the service’s reserves. Fiat-backed examples: USDC (Circle), USDT (Tether), USDP (Paxos), BUSD (Binance)

From Tether’s assessment of reserves (spring 2022)

From USDC’s website (spring 2022)

Crypto-backed: Crypto-backed coins maintain their $1 USD peg by holding other crypto assets, and often by over-collateralizing stablecoins with these assets. Consider the most successful crypto-backed stablecoin DAI. DAI’s peg is protected by a 1:1.5 collateral using ETH and other coins, including other stablecoins like the fiat-backed USDC. In other words, for every 1 unit of DAI, there is $1.50 (or more) in crypto assets backing it in reserve. DAI accomplishes this through a contract vault-based commitment of collateral. Crypto-backed examples: DAI (MakerDAO), MIM, LUSD (Liquity), RAI (Reflexer).

Algorithmic (or “seignorage”): An algorithmic stablecoin can be defined as a coin that is engineered to remain pegged to some external value (like the US dollar) without requiring collateral by using an algorithm internal to a crypto ecosystem. For example, an on-chain algorithm can pair the stablecoin with another asset. This second asset, a native asset of an ecosystem of which the stablecoin is a part (like UST and LUNA of Terra), is in a mint-burn relationship with the stablecoin. When owners want to redeem 1 unit of the stablecoin (UST), they receive $1 equivalent of the native asset (LUNA). Conversely, if LUNA holders want some UST, they can burn $1 of LUNA to get 1 UST. This can be maintained by an on-chain arbitrage, entirely internal to that blockchain (such as Terra). Examples: IRON (Iron Finance), UST (Terra), FEI (Fei Protocol).

Fluctuation in price? An important observation is that prices of all stablecoins can fluctuate. The market price of a stablecoin is partly determined by the behavior of traders on exchanges. USDC for example is backed 1:1 by USD holdings, but trading for USDC can make its exchange price fluctuate within about 0.1% day by day. This is because there is a process of arbitrage buying and selling to take advantage of these little swings. Arbitrage with balanced incentivization participates in maintaining the peg. But stablecoin projects sometimes have other ways to adapt, too. For example, MakerDAO once voted to adjust its collateralization ratio in order to bring down DAI’s price, which had risen rapidly on exchanges during the crypto crash of 2020 (more on this below).

Trilemma: Efficiency, Stability, Decentralization

Adapted from stablecoins.wtf

These three categories of stablecoin are often contrasted using a trilemma (the “impossible trinity”). A stablecoin can optimize only two of three criteria: capital efficiency (how easily the asset can be created), price stability, and decentralization. For example, fiat-backed stablecoins are highly centralized, as they depend on holdings of one or a few particular organizations (like USDC), but this allows them to have high efficiency. Crypto-backed stablecoins are more decentralized, but cannot have high capital efficiency because they often rely on over-collateralization for their peg (like DAI). Algorithmic coins are both decentralized and efficient, because they rely on that automatic on-chain algorithm — but this makes them considerably riskier and more volatile.

These tradeoffs are easily visible in data. I extracted the history of a major fiat- (USDC) and crypto-backed (DAI) stablecoin indexed by Coin Metrics, and some data on the algorithmic UST from CoinMarketCap. In the plot below, the red dots reflect days when the coin experienced a 0.5% or greater (absolute) departure from the $1 peg. DAI is the most prominent crypto-backed stablecoin, and you can see some instability during the crypto crash of 2020 as the pandemic set in. This is explained in the Coin Metrics State of the Network of April 2021.

Diagram from the author; data from Coin Metrics, CoinMarketCap (UST)

The trilemma is illustrated in the extreme by the so-called “death spiral” of stablecoins. Recently, UST and LUNA both collapsed to approximately $0 USD value. At the time of collapse, this represented over $40 billion of capitalization—it would be like 2 or more entire Ethereum NFT markets just vanished within a day or so in their entirety.

UST and LUNA, as described above, were in a mint-burn relationship on-chain in the Terra ecosystem. Arbitrage buying and selling maintained this peg, but there are situations in which the peg cannot be sustained and the death spiral leads to an unstable hyperinflation of LUNA supply and a collapse of the UST peg. Put simply, as the spiral gets started, owners of UST panic withdraw (burn), causing LUNA to be minted, but this causes the price of LUNA to drop too quickly (and panic selling on exchanges, putting more pressure on LUNA’s market price). The result is a feedback loop, and LUNA cannot maintain the peg. As LUNA supply increases, its value drops, causing yet more LUNA to be minted as owners flee UST. Nansen has written a careful analysis of this depegging event, and really the collapse of the entire Terra ecosystem, perhaps the largest such collapse in crypto history.

Explosion of LUNA supply, area of dot ~ supply; diagram by the author

In a recent post about stablecoins, Vitalik Buterin shares an important point about understanding the limits of these assets. Developers should take greater care in subjecting their designs to thought experiments — carefully thinking through how the coins would respond to extreme situations, such as having 0 users or volume. He describes how the LUNA/UST pair would obviously crash under this scenario, but how a new stablecoin RAI stands up well. A main detail he considers is how a stablecoin’s ecosystem can be drawn down gracefully, and RAI is more nimble in this respect because it has on-chain collateral (ETH)¹. More on this below.

Many Possible Products

The stablecoins that peg to the US dollar are the best known. But there are many other examples of stablecoins, including ones that peg to other crypto assets. The largest example may be wrapped Bitcoin (WBTC; BitGo and others). Owners of Bitcoin can participate in the Ethereum ecosystem by locking Bitcoin into the WBTC service. When they do, a corresponding amount of WBTC is then minted on its ERC-20 contract. Intriguingly, the total locked value of WBTC may represent one of the largest collective sums of Bitcoin held. Now owners of BTC can participate in the DeFi ecosystem on Ethereum using their BTC as an asset. As another example, the project pTokens allows cross-chain asset pegging with BTC and many other assets.

Over 270,000 BTC locked into WBTC; source: Etherscan

RAI is sometimes considered a stablecoin, but it has dynamics that are more complex, and not pegged to a specific external value or asset. RAI is backed by ETH, which is deposited to create RAI. RAI uses a contract-based adaptive mechanism to keep its market and redemption prices close to each other. RAI’s creators playfully referred to it as the “first true stablecoin,” and the technical details are fascinating. A control-theoretic mechanism implemented in a smart contract helps stabilize the relative market and redemption prices. So although the stablecoin fluctuates in its value, you can always redeem it for about its going rate on the market. RAI was initialized at a starting value of $3.14, and still fluctuates around $3 or so across any given day.

Future innovative schemes may be devised for stablecoins that represent more complex combinations of assets. Such a mechanism is well known in traditional finance, of course. The very popular exchange-traded funds (ETFs) have strong similarities to this idea. It is fun to imagine how crypto systems and innovative financial engineering may create new kinds of stablecoins in the future. In our imagination, we could envision an asset called “$BASKET” that uses an oracle to fluctuate in accordance with the average price of a load of groceries for one family, permitting commodity-denominated donations to charity services or as assistance programs. Or “$TUITION,” a coin that fluctuates with 1 year’s worth of schooling at a college or other institution. Parents or relatives could invest fractional $TUITION values incrementally over time until 4 $TUITION tokens are secured for the future.

This is not a game of ICO design, not exactly at least. The fundamental distinction here from a hollow ICO is the external commodity, service or other asset, the peg, the linkage to something else as the primary value mechanism. The stablecoin would index it.

Stablecoins as Financial Product

Tascha from Tascha Labs has a compelling thread on Terra and UST that frames stablecoins in very general terms. A helpful tweet down the thread is the following, describing stablecoins as a “product”:

Her argument is that the asset backing a stablecoin should accrue true network effects in a manner that is uncorrelated with the stablecoin. Terra and UST were too tightly entwined, exacerbated by the fact that Anchor (a lending platform) once had over 70% of all UST created. Terra (the ecosystem) was insufficiently diverse in on-chain activity. RAI, as noted by Vitalik, is backed by ETH, a cryptocurrency with perhaps the largest network effect and diverse applications. For this reason the demand for ETH will be less correlated with demand for stablecoins that get collateralized by it (like RAI). This means RAI is less likely to fall into that death spiral, less likely to get snatched up in a nasty feedback loop because there is enough other economic activity to buttress it.

This creates yet more complexity: Not only should we concern ourselves with the mechanics of stability, but we have to assess features of a stablecoin’s wider ecosystem. In fact, some have contemplated that stablecoins may grow so much that they could impact wider macroeconomic systems in traditional finance. In a recent podcast discussion between David Beckworth and Manmohan Singh, Manmohan describes a number of potential touch points between the wider financial system and stablecoins, such as how it might impact availability of collateral broadly. Depending on how stablecoins are collateralized, and how quickly they grow, it may cause shortages in some sectors.

A discussion of various DeFi limitations is found in a just-released report intriguingly entitled “On Impossible Things Before Breakfast” by Ross Stevens, Nic Carter, and Allen Farrington. The report covers Terra and algorithmic stablecoins in detail, describing a purely internal seignorage system as being like “alchemy”: alluring in its promised magic, but simply impossible to achieve. The report is also quite sharp in its critique of DeFi on Ethereum and other smart-contract chains more broadly. As we’ve discussed here, they also warn of misleading terminology which may “maliciously discourage” important research. They end their report on the idea of “LiFi,” apparently a new model for DeFi-like offerings on the Lightning Network. Regardless of one’s feeling about this new framing, one should subject all EVM-based DeFi and this newfangled LiFi to the same probing questions of any product purchase.

1. Soundness of product benefits. If I get something from this (like interest) where does it come from? Is it from genuine, sustainable growth in a product offering of an ecosystem? Or is it an unsustainable feedback loop of recursive rehypothecation?
2. System adaptivity. How unstable is the peg and/or ecosystem, and how many tools does this ecosystem have at its disposal to adapt to shocks? Does it have substantial reserves on- or off-chain to maintain the peg?
3. Refund policy. If the market value of the stablecoin collapses, can I still redeem in units of the peg?
4. Vitalik’s conundrums. What if I were suddenly the only person left in this ecosystem? Or, what if everyone in crypto clamored in? What if everyone suddenly decided to leave?
5. Centralization and control. After all these considerations, how drastically could the operators of this stablecoin simply alter policies without my knowing? Could a sudden change in policy or implementation alter product assessment above?

There are far more details than all this, more than can be considered in a blog post. We’ve not discussed borrowing/lending platforms (like Anchor, mentioned above). There are also impressive potential applications in programmability in some stablecoin projects, like USDC’s API. This post can only be a snapshot of such product concerns.

All this complexity suggests that stablecoins should be considered a product category like “power drills” or “desktop computers” or “automobiles.” Ostensibly, each of these is meant to carry out a particular set of functions. Power drills have a narrower purpose, and desktop computers more broad. But when we peruse these products as an interested buyer, we spend time reading reviews, assessing product features, determining what distinguishes instances of these categories. For anyone contemplating a purchase, if one is lucky to have a choice among competing products, it would be very strange to imagine that you could close your eyes and just pick one because, “it’s a power drill; it drills.” Or, “it’s a desktop computer, it’ll run Word.” Or “it says ‘stable,’ that’s all I need.” Product features, details, reviews, and so on, matter.

In the UST mania, for example, many investors focused too closely on one single variable—the interest rate of Anchor. They, unfortunately, did not consider other underlying product features that made UST a riskier bet.

This “product” concept is very simple, but might be one way to achieve a wider understanding of the strengths and weaknesses of crypto projects in general, and with stablecoins in particular. The “stablecoin” label risks framing these products too homogeneously. And that presents deeper risks, such as tragic exposure to stablecoin experimentation.

Further Content

• ⭐️ stablecoins.wtf — metrics and explainers
• George Selgin’s “Synthetic Commodity Money” (2012)
• Jimmy Song has a nice introduction to stablecoins in 2018
• On the Brink’s crypto-dollarization miniseries led by Nic Carter (2020–2021)
• Cointelegraph’s excellent Stablecoins 101 article
• BitMex has a thorough history of distributed (decentralized) stablecoins on its blog from 2018
• Alyze Sam has a clear and detailed history of 10 major stablecoins from 2020.
• Coin Metrics SOTN #152: “Cambrian Explosion of Stablecoins” by Kyle Waters and others (spring 2022)
• Excellent technical discussion of stablecoins from Dankrad Feist (2021)
• Vitalik Buterin’s post “Two thought experiments to evaluate automated stablecoins” (2022)
• Informative discussion between Peter McCormack and Jonathan Wu about stablecoins, regulation and Terra collapse (on “What Bitcoin Did”, 2022).
• “On Impossible Things Before Breakfast” by Ross Stevens, Nic Carter, and Allen Farrington (2022)

Endnotes

1. An arguable feature of this post is that Vitalik does not categorize USDC or other centralized coins as “stablecoins.” This is a game of semantics, but calling only decentralized coins the “true” stablecoins is a risky semantic assessment. Centralized coins are very stable, and though there is risk in that centralization, if users of UST had instead invested their money in USDC they’d still have it all (imagine the effect of misuing Vitalik’s powerful voice in a marketing ploy: “Only these algorithms are the true stablecoins!”). It would seem important to distinguish between centralized vs. decentralized stablecoins as an initial criterion to understand distinct risks.

Caveats, Disclosures

Takens Theorem is an independent data scientist and creator, and writes from the perspective of an interested participant in the space for several years. Importantly: This post is only informational, and should not be considered an endorsement of any particular external website or link, stablecoin or other investment thesis or product.