Superfluid Collateral in Open Finance
What happens when collateral becomes liquid?
While the past year was a tough one for the public crypto markets, talented and dedicated teams spent it heads down, shipping what appear to be some of the building blocks of a truly open financial system. As a result, while 2017 was the year of the ICO and 2018 was the year of continued massive private token sales, 2019 is shaping up to be the year of Open Finance.
Most significantly, Maker, which launched at the tail end of 2017, has created a system for minting a USD-denominated stablecoin (DAI) using Collateralized Debt Positions (CDPs). The system has performed incredibly well, with DAI smoothly retaining the dollar peg and CDPs sucking up >2% of all ETH, during a year in which ETH declined as much as 94% from its all-time high.
Maker isn’t the only way to borrow or lend cryptoassets. Dharma allows users to request or offer loans for any ERC20 (fungible) or ERC721 (nonfungible) asset, dYdX enables derivatives and long/short margin trades, and Compound offers money market borrowing/lending for ETH, DAI, and a handful of other tokens.
Decentralized exchange (DEX) protocols (e.g., 0x, Kyber), are now functional, as are 3rd party exchanges and interfaces to access them (e.g., Radar Relay, Easwap), allowing non-custodial trading to become a viable reality, though liquidity is still somewhat lacking across the board. In November, a new completely on-chain DEX launched that makes it possible to provide liquidity and earn fees automatically using basically the same approach as Bancor, but simplified to remove the unnecessary token. Hello, Uniswap!
And, of course, Augur shipped v1 of their long-awaited prediction market protocol, for which both Veil and Guesser have recently launched centralized services that enable vastly improved UXs.
We now have (almost) fully decentralized options for borrowing, lending, and trading cryptoassets, creating derivatives around any asset or event, and even a USD-denominated stablecoin that allows risk-off positions and greatly improved UX without ever needing to directly touch that dirty, dirty fiat.
Collateral Makes the Open Finance World Go Round
One of the core tenets of Open Finance is that of permissionlessness. But if there are no gatekeepers, then how can we be sure that a borrower won’t default on a loan or that a derivative will pay out as intended?
Collateral.
In a nutshell: collateral is posted as a form of insurance for the other participants in the system, so that you can be trusted to take certain actions without anything else being known about who you are, where you live, how competent you are, and so forth. If your actions ever come close to harming the system, you are automatically booted out and some or all of your collateral is handed over to more responsible parties.
In the Maker system, in order to borrow dollars in DAI, you must lock more than 150% of the equivalent value in ETH in a CDP. If your collateral ratio ever drops below 150%, a “Watcher” will step in, and *POOF* — your collateral is liquidated at a 13% penalty to repay your loan. Compound and Dharma employ similar structures to ensure lenders don’t need to be worried that borrowers won’t repay their loans.
Naturally, builders and participants in the Open Finance ecosystem have tended to think of assets used as collateral as just that: assets in use as collateral. Those assets may eventually be released and used for other purposes once a loan is repaid, but for now, their raison d’être is to be collateral.
But what if it doesn’t have to be that way?
Enter Liquid Collateral
There are currently over 2 million ETH locked in Maker CDPs, generating around 78 million DAI. That means at current prices, more than half the ETH in CDPs isn’t even technically required for collateral. It is a completely dead, unproductive asset.
Sowmay Jain, founder of InstaDApp, recently proposed automating the process of sweeping excess ETH out of CDPs and into Compound’s money market protocol to earn interest (the process of moving ETH back to the CDP as needed would also be automated, of course). That is a great first step and one that is relatively straightforward to implement by integrating the Maker and Compound protocols as they exist today.
But what about the rest of the ETH that is sitting in the CDPs, ensuring that they meet the minimum 150% collateral requirement? Why couldn’t that also be sitting in a Compound money market, available for others to borrow and earning the current 0.27% APR?
While a single unit of ETH can’t literally be in two places at once (one of the primary breakthroughs of Bitcoin was in solving the “double-spending problem”), there’s no reason why deposits into Compound couldn’t be made through a “deposit token” contract, which issues an equivalent number of “Compound ETH” or cETH (or cDAI, cREP, etc) ERC20 tokens. These cETH tokens would always be redeemable 1:1 for ETH in Compound. Ryan Sean Adams points out the same approach could be used in the future by staking pools with staked ETH.
Given this reliable, transparent backing, once Multi-Collateral Dai is shipped, it’s not hard to imagine that cETH could be added as a supported collateral type for Maker. Especially with speculation that much Maker activity today is driven by ETH holders who are looking to go long with leverage, it makes sense that there would also be demand for earning interest on the ETH that sits as collateral.
cETH is a fairly basic example of liquid collateral, but what if liquid collateral was actually used to provide… liquidity?
Meet Uniswap
Uniswap is a fully on-chain decentralized exchange. Rather than maintain an order book, Uniswap uses liquidity pools and an automated market maker to determine the price at which an asset can be traded. If you want to supply liquidity to the exchange and earn a proportional share of the 0.3% fee charged on every trade between a given asset pair, you just deposit an equivalent amount of ETH and the relevant ERC20 token. (If you want a deeper explanation of Uniswap, check out Cyrus Younessi’s excellent overview.)
The ETH/DAI pair is currently the second deepest liquidity pool on Uniswap. Both assets are also available on Compound. It would be great for anyone providing ETH/DAI liquidity on Uniswap to also earn interest on their assets by having them simultaneously available on Compound for borrowing. Again, the cETH-type trick would work here for half the equation (cDAI), but it feels like there’s potential to do a more direct integration or re-writing of the two protocols to make this possible.
A different trick with Uniswap that definitely works is using the liquidity pair itself as collateral (e.g., ETH/DAI, as opposed to each of the assets in that pair, ETH and DAI). What does that mean? Well, although most Ethereum wallets don’t show them by default, whenever you deposit liquidity into a Uniswap pool, your share of that pool is actually represented by ERC20 tokens.
Any system willing to accept as collateral both assets in an ETH/XYZ pair should also be willing to accept the corresponding Uniswap pool shares as collateral. Uniswap’s automated market maker function ensures that the combined value of the ETH/XYZ pool share will never drop more than either of those two assets. In fact, all else equal, the value of the ETH/XYZ pool shares will rise over time, based on earnings from trading fees being added to the liquidity pool.
I predict we’ll see Uniswap pool shares used as collateral for millions of dollars in loans in months, not years.
What could go wrong?
Anyone familiar with the world of prime brokerage services will immediately recognize the process described in the preceding sections as forms of rehypothecation: a lender taking an asset posted as collateral by a borrower, and using that same asset as collateral to take out another loan. Except in this case, we may see collateralized collateral collateralizing collateralized collateral… and so on and so forth. Oy.
This creates a daisy-chain situation: if there’s a failure at any link along the chain, all of the assets further down the chain will also fail, but anything further up (i.e., closer to the original underlying collateral) should be fine.
However, it’s extremely unlikely this chain of collateralization would be created in a neatly serialized fashion. All available evidence suggests that financial engineers will slice, lever, mix, and sling every imaginable asset in every imaginable combination to create new products they can sell to each other or the unwashed masses. It will probably end up looking more like this:
Is this really any worse than the legacy financial system? Probably not. Technically, this would all be publicly viewable and auditable, rather than hidden behind a series of closed doors and incomprehensible legal contracts. We should be able to devise systems to track and quantify risk when everything is linked together via public ledgers and immutable automated contracts. We should be able to self regulate, put in place reasonable standards, and refuse to interact with contracts/protocols that don’t demand conservative margins and ensure the collateralization chain doesn’t go more than a couple layers deep.
But given what we know of human nature, do you really think we’ll show restraint when the possibility exists to earn an extra point of yield or pay a slightly lower rate on a loan?
And yet…
There is something undeniably compelling about all of this. If assets can be allocated for multiple purposes simultaneously, we should see more liquidity, lower cost of borrowing, and more effective allocation of capital. Most of the builders I’ve met working on Open Finance protocols and applications are not looking for ways to wring a few extra bips (basis points, not Bitcoin Improvement Proposals… sorry) out of the system; they’re trying to build the tools that will ultimately make every imaginable financial asset, service, and tool available via open source software on the phone of every person on the planet. Maybe we’ll never get there, but based on the hyperspeed pace at which this industry is evolving, if this is all a big terrible idea, at least we’re likely to figure that out while it’s still only a few million nerds losing their shirts, rather than causing the entire global financial system to crash and burn.
In the meantime: superfluid collateral, anyone?
Thanks to Tara Tan, Ali Cottong, Matt Condon, Sowmay Jain, and Dan Held for feedback.
