The Cost of Centralization: Stablecoin risk, yield, and structure

Over the opening month of 2022 there has been a LOT of drama in Crypto. The lead Treasurer of Wonderland.Money, 0xSifu, was revealed to be QuadrigaCX’s co-founder Michael Patryn. On top of that, the other co-founder, Dani, knew it was the case and unilaterally withheld the information from the ‘DAO’. Investors were displeased (duh) when the news broke publicly, and wMEMO sold off to well below its backing price. UST and MIM both then broke peg, caused by waves of liquidations across three chains, and like four or five different platforms. Turns out the Senator’s nerves around ‘Magic Internet Money’ and ‘Degenboxes’ may have been at least somewhat justified.

Not only MIM is affected by this synergy. The prime use case for the Degenbox is its ability to leverage UST up to 9x (via MIM). Since the introduction of the loop on November 8th, the circulating supply of UST has risen from 2.83bn to around 11.25bn - that’s a 300% increase in a quarter year. Certainly, some of this has been clever navigation and structure by Terra founder, Do Kwon, but clearly Abracadabra is influencing the amount of UST circulating. In that time, MIM supply increased to a peak of over 3bn- in large part flowing back to UST. 

The rebound of MIM despite a black-swan event speaks to a key point for DeFi users- fully collateralized stablecoins can hold even under high stress situations. MIM Is fully collateralized by UST, which held its peg admirably after a brief drawdown to $0.92. In turn, UST is an algorithmic stablecoin derived from LUNA, the native token of the Terra blockchain, and which is minted by the Terra Treasury (featuring a $300mn bailout on the above news, and another $450mn planned). That’s already almost 20% of the Market Cap of Anchor in the past two weeks. 

That said, funds are meant to be used in exactly these situations. Do notes that even if the yield reserve is depleted, Anchor will still be able to offer ~15% APR (note: at these levels). So, for a retail investor looking to save, Anchor is still enticing. However, if you’re a larger institution or protocol looking to settle flows and minimize risk are these really your best options? It’s important to note that of the 11.25bn UST in circulation today, almost exactly half are in Anchor. Of those, probably half (based on MIM supply) are through some level of Degenbox loop. That puts a lot of emphasis on the success and risk-management of two isolated protocols.

The evidence seems to support this as well. Frequently, traditional institutions and protocols turn to other, more secularly collateralized coins like USDC, DAI, and USDT. As an investor, you can back whichever horse you believe in. Differentiation in the space occurs through two routes: yield and risk mitigation. UST, through Anchor, and MIM, through Degenbox, are focused on the prior, and offer an enticing yield on cash of up to 100+% APR. Major FIAT-collateralized stable issuers like Circle and Tether are focused on the latter. 

After spending time in the space it’s easy to get desensitized to the overwhelmingly high yields available. It’s important to take a step back and put these numbers in context. Interest rates on cash outside of crypto are effectively zero. This has had the practical effect of crushing margins in many businesses that carry a balance sheet, such as lending operations and banks.

In DeFi, the macro framework is overwhelmed by the incredible ROIC for protocols with a balance sheet. This means that even stablecoins like USDT and USDC, which are collateralized entirely by cash and cash-equivalents, can offer 3%+ staking rates on sites like Aave and Compound. It’s not a perfect comparison, but the last time the Federal Funds rate was above 3% was prior to the 2008 recession. Even 10-year Treasuries are only pricing at peak rates of around 2%. It’s unlikely that we see traditional infrastructure able to match these rates of return anytime soon. That trend will continue to force central and private institutions to turn to digital assets. When they do, their primary concern will not be on yield- it’s on security. 

That trend is extremely clear when we look at recent growth of stables. The two most token s with the most decentralized backing and lowest yield, USDC and USDT, are head and shoulders above their peers in terms of uptake. 

At its core, this arrives at the key questions moving forward for stablecoin issuers: Do stablecoins need to be fully collateralized? Why are decentralized backings superior to centralized? Are some assets superior to others when backing a stablecoin?

The last question is the easiest to answer. Clearly, a coin fully collateralized by BTC or ETH would be better than a coin fully collateralized by your favorite shitcoin. The high volatility and centralization of the latter would lead to massive inefficiencies in price and frequent arbitrage opportunities. So, we arrive at the First Tenet of Stables- some assets are better collateral for decentralized stablecoins than others. 

Fully collateralized coins can be backed by anything. USDC is backed by cash and cash-equivalents; DAI is backed by digital assets (Crypto), but contains 60+% centralized tokens. So, thinking about yields in the context of backing is important. Anchor is able to offer the highest yield because of their centralization and direct partnership with TerraLabs, but it also makes it the most risk prone. By extension, MIM, which is a leveraged position on UST, would be multiples more risky than its parent. USDC offers the lowest yield because it has the highest ‘minting’ cost- purchasing cash (with some offset through cash equivalents). DAI sits somewhere in the middle, semi-centralized, but entirely cryptoasset backed. UST is technically algorithmic, but the relationship between Anchor and TerraLabs is so codependent that it's effectively backed by aUST's yield mechanism until more growth occurs in other protocols on-chain.

At the same time, it’s even more crucial to also put risk in perspective. Think back to the recent black swan MIM event. Even with a complete meltdown of the riskiest major stablecoin, MIM never drew down more than 20 cents. Only the most aggressively leveraged traders were ultimately liquidated. UST (less risky) drew down under 10 cents & has an all-time low of $0.85 from nine months ago. USDC’s bottom is from the same period, having drawn down to just $0.90. With backing updates and increases in sticky circulation, risks of such a drawdown decrease. This leads us to the Second Tenet of Stables- backing quality is risk based on the quality, quantity, and decentralization of the underlying assets. 

Compare stablecoin issuers to their closest traditional finance proxies- banks. Banks have reserve ratios set by the Fed- these are the fancy interest rates you keep hearing about. As they’ve approached zero, banks have been able to lend out increasing amounts of money against their assets. Obviously, token issuers cannot and should not leverage this irresponsibly for any number of reasons, but perhaps a middle ground is feasible. 

We would expect that investors would be increasingly comfortable with greater collateralization ratios as the backing quality of the assets rises. Extending our prior example above, if Anchor, DAI, and USDC all reduced their Collateralization Ratio to 50% overnight- general concern for underlying protocol health would likely be least for Circle’s and greatest for Terra. So, anyone seeking to create a semi-algorithmic stablecoin would need to do so in a way that can compete in yield opportunities, but without sacrificing asset quality and peg integrity. 

To be honest, this report originally started out as a case study on FRAX. The more I dug, the more I realized that part of the main reason for the need for such extreme overcollateralization is simply the tremendous information asymmetric around how to evaluate risk around centralization and backing. That’s likely why we are seeing so much FIAT-stable adoption; if you don’t know, err on the side of caution. 

On the other hand, requiring new mints to be entirely backed doesn’t really make sense either when you think about it. You want to tell me, that Archegos can run enough leverage in SPACs and concentrated international equity positions to cause a meltdown  the size of UST’s market cap, and we can’t collateralize a stablecoin at 75%? 

Frax is the first, and only, large scale semi-collateralized stablecoin. What this means is that rather than basing the entirety of its backing on collateral, like prior examples, or on algorithmic pegs, it utilizes a mixture of the two. In Frax’s case the ratio between collateralized and algorithmic depends on the market's pricing of FRAX. If FRAX is trading at above $1, the protocol decreases the collateral ratio. If FRAX is trading at under $1, the protocol increases the collateral ratio. 

Ideally, this combination allows for the scalability and security of collateralized stables, but without the volatility and difficulty scaling algorithmic assets. Because FRAX can always be minted and redeemed centrally for $1, it maintains the ability of collateralized stables to keep price stable through arbitrage. In the background, this is also supported through algorithmic adjustments in the floating collateralization rate. If the price rises above $1, FRAX decreases the collateralization ratio, which means fewer assets are required to mint new FRAX (example below). In turn, this causes increasing supply and puts pressure on price to return to peg. When trading below its peg, required backing increases and puts downward pressure on supply, and the opposite on price.

Example A: Minting FRAX at a collateral ratio of 100% with 200 USDC ($1/USDC price)

Put in 200 USDC. 100% Collateralization is required to mint. Receive 200 FRAX. 

Example B: Minting FRAX at a collateral ratio of 50% with 200 USDC ($1/USDC price)

Put in 200 USDC. 50% collateralization ratio = $200 USDC / 0.50 = $400 of mintable FRAX.

The collateralization ratio can also be viewed as a measure of how much faith investors have in the algorithmic side of FRAX. If faith is low, say in a period of high volatility, the price of the token would drop. This, in turn, would cause collateralization ratio to increase, and newly minted tokens would be ‘overcapitalized’ relative to prior minted tokens (although all FRAX trade equally, because you can always centrally redeem). On the other hand, if volatility is low and people have increasing faith in the underlying mechanism, we would expect to see the ratio trending downward over time. 

Indeed, since FRAX’s launch, we’ve seen a steady decrease in the collateralization ratio from 100% down to its current level at 84.5%. How  has FRAX inspired such confidence in its backing? The first thing is the composition- DAI, FEI, LUSD, sUSD, USDP, USDC. Unlike DAI, which is backed by stables as well as ‘market assets’ (wBTC, ETH…), FRAX’s is collateralized entirely by dollar stablecoins. It doesn’t fix the centralization issue- FRAX began as decentralized as DAI (~40%), then trended down to 30% in recent months. 

Breaking down the move, numbers were initially strong due to the protocol’s strong involvement with Defi and Web 2.0 protocols, such as Olympus. As the protocol scaled its supply almost 10x, a significantly larger portion of the value had to move into high-liquidity AMOs like Curve’s FRAX3CRV. At the same time, that move necessarily increased their centralization. Half of the pool is FRAX, which they can’t hold as backing and consider themselves centralized. The remaining half is in 3CRV, a mix of DAI (60% centralized, 40% FIAT), UDST (100% FIAT), and UDSC (100% FIAT). 

As the protocol continues to scale organically, trust increases in the strategy, and the collateralization ratio decreases over time. Now, a greater portion of all newly minted FRAX is algorithmic, rather than collateralized. Algorithmic FRAX is decentralized. On top of that, if maximizing decentralization was a near term goal, focusing remaining backed mints around assets like Lending AMOs, which overcollateralize, and stables like LiquityUSD, which requires 110% collateralization in exchange for 0% interest loans.   

FXS is ultimately a pretty straightforward concept that is much more interesting in application, so I won’t spend too much time here. 

At the highest level, you can think of FXS:FRAX as LUNA:UST. As new FRAX is minted, an equivalent amount of FXS, in dollars, must be burned. That means that the quantity of FXS burnt per FRAX mint is contingent on both the number of FRAX, collateralization ratio (CR), and the current price of FXS. When CR=100%, no FXS is burned.

For example: 

Ex. 1: Take $100 USDC to mint FRAX, with FXS at $20, CR=80%: 

1. $100 USDC is deposited at an 80% Collateral Ratio.

2. 1.25 FXS is burned, per the formula in the docs

3. 125 FRAX is minted ($100 USDC / 80% CR)

Ex 2: Take $100 USDC to mint FRAX, with FXS at $50, CR=85%: 

1. $100 USDC is deposited at an 80% Collateral Ratio

2. 0.50 FXS is burned (60% less)

3. ~118 FRAX is minted

From an investor’s POV, the burn mechanism is great. By owning FXS you get direct exposure to the growth of FRAX supply. Plus, when FXS price significantly lags FRAX supply, increasingly large quantities of FXS are being burned. The difference can’t be maintained forever- as long as FRAX’s underlying secular growth is there FXS will respond in kind, and, the longer it takes the more aggressive the catchup.

Interestingly, the divergence between the price of FXS and supply of FRAX is higher than it's ever been. Yes, there are macro effects that can suppress the price impact and reduce the multiple put on the protocol in the short term, but the underlying growth fundamentals are unchanged. 

This takeaway is supported through a detailed look at the sentiment and volume of Twitter activity around FXS. Despite FRAX mints quickly and consistently rising throughout the same period, both amount of discussion on social media, and tone of discussion, fell dramatically in the same period. That suggests that, despite unchanged growth fundamentals on the project itself, the lack of price movement is a product of lack of engagement, rather than underlying issues- reinforced by the recent uptrend in sentiment.

Here’s where things get really fun.

I know you’re probably sitting there thinking that you’ve been having a blast already. But would I really be able to stick to just writing about stablecoin risk and the basic alternatives? Absolutely not. On to the ponzinomics. 

veFXS is a staking system based on the now famous Curve mechanism. By locking FXS for up to four years, users receive up to four times the amount in veFXS, scaling linearly (so, 100 FXS locked for 4 years returns 400 veFXS). You can think of these as being a timer for how long your FXS is locked up for. As the unlock date approaches, the ratio of veFXS:FXS will approach 1:1, generating the highest rewards for staked users who are willing to align with the protocol long term.

Like veCRV, veFXS works to direct liquidity. Through votes, FXS emissions are deployed across a variety of different FRAX staking pools. Each active pool has a minimum and maximum yield range contingent on the number of votes directed to that pool. Pools where farmed FXS rewards are sold off will not have voting power to continue to direct emissions, and will taper off naturally over time. 

All these mechanisms combine to make for very sticky holders. 62% of circulating FXS is locked as veFXS, with an average lock time of 1.5 years. While the base APR of 10.59% isn’t necessarily super exciting on paper, it allows for controlled token emissions schedules with upside optionality through the burn mechanism. It’s also worth mentioning that veFXS APR goes up, just like burn rate, as FXS price goes down due to more being bought back per dollar of profits. 

The last thing really worth digging into is a big one: the Convex/FRAX collaboration. Between their ve-updates and this, FRAX is well and truly intertwined into the Curve Wars. If they become as big a theme over the next year as predicted, we will likely see accompanying supply gains for the stable. 

Announced in late December, the duo settled on the creation of a new token, cvxFXS. Like veFXS, it is earned through an FXS deposit and leaves you eligible for the FPI airdrop (wen). Locked FXS is automatically set with a four year horizon, meaning maximum eligibility. Although accrued rewards aren’t live until FPI’s launch, Convex is holding on to accrued rewards to be distributed. 

cvxFXS’s initial uptake was phenomenal, rocketing up to 2.5 million FXS locked within the first two weeks of launch. Since then, incremental flow has been pretty lackluster. As it stands, Convex’s total FXS holdings stand around 3mn, or ~8% of the total supply of locked FXS. Interesting, given that rewards from veFXS and cvxFXS should be more-or-less identical in the near term. 

In the medium-to-long time horizon, cvxFXS becomes incredibly interesting. Opportunities to use influence to affect FRAX liquidity through the native protocol remain fairly low. On the other hand, large quantities of FRAX are being transacted as one of the most popular base pairs in Curve & Convex-related pools. By aligning itself more closely with the liquidity already deployed on those platforms, cvxFXS should expand the opportunity set of bribes. Other marketplaces, such as Redacted Cartel’s Hidden Hand marketplace will also serve as buyers of these vote tokens, and ensure continued secular demand. BTRFLY has quickly risen to become the second largest holder of CVX, and will continue to amass similar assets. 

In any case, FRAX has found itself in an admirable position simply through staying its course. Some peers fly close to the yield sun in search of DeFi glory. Others stay very close to home in search of traditional finance comfort. It’s likely that the answer, like FRAX, lies somewhere in the middle. 

This report is not investment or trading advice. Please conduct your own research before making any investment decisions. Past performance of an asset is not indicative of future results. The Author may be holding the cryptocurrencies or using the strategies mentioned in this report