Data on Tether (USDT) from Coin Metrics.
2018 has been a big year for stablecoins. From concerns around Tether’s insolvency to a proliferation of transparency-oriented “Tether-killers”, to discussions around the viability of crypto-native stablecoins, the topic has found consistent mindshare in the crypto community. A considerable amount of research and writing has been dedicated to examining the stability mechanisms that each of these stablecoins employ (some great overviews here, here and here), so I’ll focus on an aspect that hasn’t received as much attention: the business behind stablecoins.
It turns out that stablecoin issuance can have significant upside across a range of mechanisms for capturing value. This upside is driven by seigniorage.
While there are diverging definitions in the literature, at a high level, seigniorage is the net revenue derived from money creation and the introduction of this money into the economy. Historically, seigniorage was the difference between the purchasing power of commodity money and the cost to produce it (e.g. the profits earned from issuing metal coins at a price above their melt value) and represented a significant source of income for monarchs and governments.
In modern economies, the majority of money is created by commercial banks through the act of lending. Commercial banks produce revenue primarily from functioning as financial intermediaries and money creators. While the lines are blurred with respect to the revenue attributable to each function, it’s generally understood that money creation is the primary income driver, as it allows commercial banks to lend in excess of their deposits. Without getting lost in semantics, commercial bank seigniorage can be thought of as some (likely majority) portion of the interests earned on banknotes held less interests paid on account deposits.
Stablecoin issuers, on the other hand, function exclusively as money creators and the seigniorage they earn depends on their issuance mechanics and network growth. Thus, when investors back a stablecoin project, seigniorage potential is a key valuation driver. Let’s explore this potential across the three major categories of stablecoins and discuss possible competitive implications.
Fiat-backed stablecoin systems accept fiat currency or other non-crypto assets as collateral, store it with custodians and issue tokenized IOU vouchers (stablecoins) on a 1:1 basis. The issuing company can generate revenue as a function of outstanding stablecoin supply, fiat collateral-base or transaction volume (on-chain or on exchanges). The following are the main ways these types of stablecoins capture value in practice:
Charge fees on issuance/redemption:
Issuers that employ this model charge a fee for issuing and redeeming their stablecoins in exchange for underlying fiat collateral. Revenue here is a function of the in/out flow, which is driven by changes in demand — primarily for access to crypto markets, with stablecoins serving as a major on/off ramp. Interestingly, fees in this model are limited by nature of the stability mechanism, as the spread between the stablecoin price and the underlying collateral must be greater than the fee to create arbitrage opportunities for traders. To encourage a tight spread, the issuer must limit its revenue potential from this strategy, which may explain Tether and TrueUSD’s choice of a 0.1% issuance/redemption fee (Tether actually has higher fees for redemptions, on the order of 0.4–3%). Assuming annual in/out flow of $10B and 0.1% issuance/redemption fee, this strategy could generate $10M in revenue.
Engage in market-making on exchanges, earning as a function of the bid/ask spread and exchange volume facilitated by the issuer’s trading desk. The bid/ask spread reflects the depth of order books on exchanges and depends on liquidity provided by the company, traders and other 3rd party market-makers. Importantly, bid/ask spreads tighten with increased trade volume and can be lower than 0.01% in a sufficiently liquid market. It’s also worth noting that market-makers run by the issuer are not differentiated from other market-makers, suggesting that this is not a primary monetization strategy as much as it a necessary function for the wellbeing of a stablecoin ecosystem.
Invest a portion of collateral-base in short-term treasuries and money market funds. This strategy is driven by the size of the collateral-base, interest rates on short-term investments and the reserve ratio — the ratio of reserves held as fiat vs productive assets. 1-month treasuries currently yield 2.2% annually, making this an attractive strategy. With a stablecoin market cap of $10B and a reserve ratio of 50%, this strategy would yield ~$110M in annual revenue. Note that this is an order of magnitude more lucrative than charging fees on issuance/redemptions, provided that annual in/out flow is comparable to market cap (not necessarily a fair assumption).
By no means do the above strategies encompass all possible ways to capture value in fiat-backed models — they are just the strategies I’ve seen employed so far. Of the above, short-term lending on a fractional reserve seems to be the most lucrative strategy. It’s important to note that fiat-backed stablecoin issuers can engage in multiple monetization strategies at the same time, though not all types of issuers share the same incentives to monetize their stablecoin systems.
Fiat-backed stablecoins are used most on centralized exchanges, serving as a desirable trading pair, safety asset and onboarding tool. Because exchanges receive meaningful utility from fiat-backed stablecoins, they could in theory justify issuing them without needing to monetize directly. This makes it challenging for non-exchange issuers to compete on user-imposed fees with Tether, USDC, GUSD, etc. For this reason, market-making and short-term lending may be the only viable long-term monetization strategies for fiat-backed stablecoins.
Crypto-backed stablecoins move away from centralized custody and provide a decentralized alternative that still maintains claims on collateral. As we depart from the simpler implementation of the fiat-backed approach, the design space expands and monetization strategies become more complex. In addition to the stablecoin that they issue, crypto-backed projects typically feature a second cryptoasset — a “volatility-coin” — that is intended to incentivize behaviors that benefit the system. While the issuing company or DAO maintains the flexibility to engage in market-making activities, the volatility-coin tends to be the asset of interest from an investment perspective.
Crypto-backed stablecoins maintain reserves of on-chain collateral: collateral in the form of cryptoassets, held in smart contracts. The on-chain collateral typically consists of ether or volatility-coin, but in the future might include wrapped bitcoin, utility tokens, security tokens and even other stablecoins. The volatility-coin could have a range of functions as a utility token or might be a security token that holds rights to certain income generated by the system. The ways volatility-coins presently do or are proposed to capture value include:
Rights to transaction fees:
Fees on on-chain transactions are paid proportionally to volatility-coin holders. The volatility-coin can also serve as the collateral backing the stablecoins. Given the potential volatility of the collateral-base, stablecoins are issued on an overcollateralized basis. The volatility-coin represents a percentage of future transaction fees and its value can be modeled accordingly. This ties the strength of the stability mechanism to the economic activity on the network. On-chain transaction volume tends to be at least an order of magnitude larger than the stablecoin market cap (the velocity of USDT is ~44, calculated using coinmetrics’s adjusted transaction volume), suggesting that fees associated with transaction volume can generate significant value for volatility-coin holders.
Havven is the only live example using this model and collects 0.15% on on-chain transactions, which are paid to HAV holders. Given $10B in stablecoin market cap and a velocity of 44, on-chain transactions would be worth $440B on the year and HAV holders would accrue $660M in revenue. As a benchmark, at peak market cap of ~$3B, Tether did about $300M in on-chain volume per day — which translates to roughly $110B annually. Clearly, transaction fees can be highly lucrative, but the strategy imposes avoidable costs on users, which may be to the detriment of widespread adoption.
Rights to interest on collateral:
The Maker system issues stablecoins as collateralized debt positions (CDPs) with an annual interest rate of around 2.5%. The volatility-coin in the system is MKR, which has multiple value capture mechanisms, one of which is indirectly receiving the economic equivalent of part of the interest payments from CDP owners. This portion is called the “stability fee” and will change upon launch of Multi-collateral Dai. The interest is paid in MKR, which is subsequently burned, bearing resemblance to a stock buyback that should manifest in price appreciation of MKR tokens. The outstanding debt is the market cap of stablecoins (DAI), meaning MKR holders can approximately expect fee proceeds equivalent to some percentage of the DAI market cap each year. Given a $10B DAI market cap and a 0.5% return to MKR holders, MKR holders should expect to indirectly receive $50M through MKR burning. Note that MKR holders are also punished through dilution if CDPs become undercollateralized, which has not happened yet to my knowledge but could in a severe market crash. It might be appropriate to model this risk as a reduction to the effective interest rate.
Claims on collateral:
The Reserve system sells both stablecoins and shares for collateral assets (tokenized bonds, properties, commodities). The collateral assets can be redeemed at any time in exchange for either stablecoins or shares. The trick here is that stablecoins represent fixed claims in fiat terms and appreciation of the collateral assets will ultimately be paid to shareholders as dividends. The system will initially be overcollateralized by the contributions of shareholders, but the collateralization ratio will trend towards 1:1 over time as the supply of stablecoins grows. If during that time the value of the collateral appreciates, the stablecoins will be worth less of the collateral assets than they paid in, with the difference being paid to shareholders.
For example, imagine there are 10 shares and 10 stablecoins that were each acquired in exchange for 1 GoldCoin. The collateral-base would consist of 20 GoldCoin and the collateral ratio would be 1:1. Now suppose the value of GoldCoin doubles. The collateral ratio now is 2:1 and each stablecoin holder can redeem only for 0.5 GoldCoin as they have fixed claims in fiat terms. The system can now pay out half of the collateral, 10 GoldCoin, to share holders to return to a 1:1 ratio (each share holder receives 1 GoldCoin in dividends). This is effectively a leveraged investment in the collateral asset, where the amount of leverage is determined by the ratio of stablecoins to shares. Thus, profiting from shares here requires that the underlying collateral assets appreciate and a meaningful amount of stablecoins are issued.
The volatility-coin gives the owner governance rights over system parameters and protocol changes in the future. In Maker, MKR holders will be responsible for voting on key risk parameters that affect the solvency of the system as well as proposals for protocol upgrades. This is the least concrete value capture mechanism as it is extremely challenging to quantify the value of governance. Clearly, if a network facilitates a meaningful level of economic activity, changes at the protocol level can have a material impact on network participants. Network participants therefore have an economic incentive to protect their interests — a volatility-coin with voting rights may capture some of the value of this incentive.
A potential mental model for estimating the value of a given governance token: how much less is your business worth if you switch to the next-best network? That ought to be the upper bound for what a business should be willing to pay to protect its interests through the governance process. The lower bound might be the least amount that the average stakeholder can contribute and still get the service they need.
In order to avoid positioning them as directly competitive, I want to draw a distinction between the market opportunities for fiat-backed and crypto-native/decentralized stablecoins. Fiat-backed stablecoins serve as an on-ramp and banking API for the legacy financial system to interact with crypto finance, whereas decentralized stablecoins will service crypto-native applications such as DEXs, decentralized financial products and dApps. At risk of stating the obvious, trustless execution of smart contracts simply can’t rely on tokens that can be frozen. Free of major external dependencies (ignore the oracle in the room for now), crypto-backed stablecoins may be more easily and reliably integrated into the emergent infrastructure for programmable money and smart contracts.
In all likelihood, we’ve barely scratched the surface of potential value capture mechanisms for crypto-backed stablecoins and their associated volatility-coins. Of the above, tangible rights to fees or interest payments are the easiest to value, while the value of unique instruments of leverage and governance rights are harder to quantify. While the volatility-coin introduces conceptual challenges around valuation and adoption, it also serves a distinct purpose missing from the fiat-backed model: incentive alignment among stakeholders. Crypto-backed stablecoin systems borrow from Bitcoin the idea that a community can be incentivized to evangelize one network over another by sharing in its upside. This may ultimately enhance network effects for crypto-backed stablecoins beyond what is possible in the fiat-backed model, giving them a strong competitive edge in the long-term.
A more experimental variant of decentralized stablecoins, algorithmic stablecoins adjust supply to stabilize price without any collateral requirements. They are designed to stabilize price in a way that is both crypto-native and capital-efficient, which in theory makes it easier for them to scale than crypto-backed systems. Like crypto-backed stablecoins, they commonly feature a volatility-coin that may capitalize the system in early days and ultimately serves to align stakeholders. Algorithmic stablecoins are by far the most lucrative category, as they accept capital inflows without providing redemptions, turning almost all money paid into the system into seigniorage. The following are the main value capture mechanisms that have been proposed:
The basic idea behind seigniorage shares (first proposed by Robert Sams in 2014) is to have two classes of users, stablecoin users and seigniorage shareholders (volatility-coin holders). When demand for stablecoins increases, new stablecoins are issued to offset demand and keep price flat. The new stablecoins are issued proportionally to seigniorage shareholders based on the percentage of shares they own. When demand decreases and price falls, shares are put for sale in exchange for stablecoins, which are subsequently burnt from supply. A potential investment strategy might consist of buying shares every time they go for sale to keep from getting diluted; the result would be a fixed percent share of all future stablecoin issuances. Thus, if the stablecoin market cap grew to $10B, seigniorage shareholders would earn $10B. Carbon most closely resembles the original seigniorage shares concept, while Basis and Fragments deviate from it in several key ways which affect how value is captured.
Basis deviates from the seigniorage shares model by introducing a third class of tokens, bonds, that are sold to burn stablecoins and pay out a fixed return. Bonds in Basis pay out before shares, according to a bond queue, and are given a 5 year expiration date meant to jumpstart demand if needed in the future. The implication is that bondholders will serve as the first money in when the system needs to burn stablecoins, with shareholders ultimately being the last line of defense. The shares are finite in supply, suggesting that the shareholders will have an incentive to pledge capital to support the system by buying bonds if no other buyers emerge, as they capture all upside after bondholders are paid out.
Fragments also deviates from the seigniorage shares model by combining the stablecoin with the volatility-coin — issuing new stablecoins directly to the existing stablecoin holders. The result is that users’ wallet balances change both as the system adds new stablecoins and removes them. Stablecoins in the Fragments system are both stores of value and units of account, as they accrue purchasing power with the growth of the system but are denominated in fiat. With Fragments, users and investors both hold the same token, which could more closely align their interests in the long term.
Variable interest rate deposits:
Another way to increase supply is to create an algorithmic central bank that expands money supply through floating interest rates. In the Feron system, users deposit stablecoins into a smart contract that pays interest. In times of low demand, interest rates are increased, driving more deposits which lower velocity and bring price increases; in times of high demand, interest rates are lowered, increasing velocity and decreasing price. In this model, an investment strategy might be to strategically allocate to the deposit contract at times of high interest. A disproportionate amount of new supply will accrue to those with good timing, creating a fairly interesting new asset for traders.
Across both of the algorithmic strategies we are considering, all supply increases are paid in full to holders of volatility-coin (whether that is broken into bonds/shares or combined with the stablecoin). This is in stark contrast to other models we’ve looked at, that capture 2.2% or less of the market cap increases, though it’s worth noting that if a seigniorage shares model wins major global share, ultimately the market cap increases will converge to something low like GDP growth. In times of high growth however, algorithmic stablecoin systems clearly have the most potential for value capture.
Because of the difficulty of maintaining a strong peg in systems without redemptions, some of these algorithmic models don’t produce true stablecoins but rather create assets designed to decrease in price volatility as they mature. Both Fragments and Feron don’t explicitly change supply in real time, opting instead to change it slowly over time — trending towards stability. This arguably puts them in a separate category (“low-volatility coins”) but I included them in the algorithmic category because their value-capture mechanisms are similar in magnitude to the seigniorage shares model. This has very meaningful implications for the target market though, as these fall more into the store of value bucket than medium of exchange. As such, these are not directly competitive with other stablecoins in the short-to-medium term, at least until their mechanisms meaningfully curb volatility, which requires massive buy-in from hodlers and would likely take many years.
The biggest benefits dollar-pegged algorithmic systems like Basis have over crypto-backed systems is their capital-efficiency and increased profit potential. Capital-efficiency should make it easier for the system to respond to sudden growth in demand, while the profits can be shared with many stakeholders who should be motivated to support and drive the ecosystem forward. These benefits, however, come at the cost of being growth- or faith-backed. It’s not readily apparent whether the benefits outweigh the costs, with profitability coming at the expense of increased susceptibility to collapse during a black swan event. This tradeoff is likely to be a headwind for algorithmic stablecoins within the broader decentralized stablecoin opportunity, as the market should favor risk-aversion when it comes to “stable” cryptocurrencies.
Stablecoins are productizing money, turning it into a profit-motivated competition for who can create the best form. With a tremendous opportunity to create more useful money for the world, stablecoins are proliferating with a variety of mechanisms for capturing upside. From an investment perspective, this has become a crowded space very quickly, with many untested, new and experimental ideas coming to life.
Each category of stablecoin differs in the ways value is captured. On the fiat-backed side, sustainable value capture is capped at a low percentage of the stablecoin market cap. On the crypto-backed side, some value capture mechanisms resemble those of fiat-backed systems, but the design space opens up and allows for new concepts like governance tokens, where potential upside is unknown. Within the algorithmic category, value capture mechanisms have the highest near-term upside due to the elimination of redemption rights. This helps explain why despite only representing 23% of all stablecoin systems, algorithmic projects have raised 50% of the total $350M funding allocated to stablecoins so far.
The most interesting development in the business models of stablecoins is the introduction of volatility-coins among crypto-native/decentralized stablecoins. Volatility-coins allow for a much broader sharing of the value surplus that money creates. For the first time in the history of money, seigniorage can be distributed to a group as broad as all users of money. This could have profound implications for the future of money and we are very much in uncharted territory.
On the competitive front, I expect fiat-backed models to dominate in the short term — they are the conceptual cousins of the fiat money that we are comfortable with already. Over time, as the open financial system grows, DEXs gain market share and dApps gain adoption, I suspect that crypto-native stablecoins will slowly eclipse fiat-backed ones on the merits of censorship-resistance and value sharing, with Maker’s Dai currently leading the way. But in order to get there, fiat-backed stablecoins will be a crucial on-ramp.