Decentralizing Credit 101

lending is not simple

Today’s article is about what lending onchain will look like in practice, its pitfalls, and how the ECG handles multiple borrowing denominations.

The mainstream model of DeFi lending, which I complained last time, is based around centralized governance and trusted oracles that observe a global market state. They optimize for convenient user experience over security with results resembling the diagram below.

The ECG model is the opposite, removing all unnecessary features and trusted dependencies in the core model. More choice for the lenders and borrowers also means more responsibility.

Borrowing from the ECG

Without a trusted oracle to trigger liquidations, or a pool of idle capital to facilitate withdrawals managed by an automated interest rate curve, lenders are responsible for “calling” loans when they need to access liquidity or fear the loan may become undercollateralized. Borrowers must correspondingly monitor for their loan being called if they do not wish for it to become subject to liquidation.

While there is no liquidation penalty for the borrower (only as much collateral as is needed to repay the debt is sold) this may be a taxable event or otherwise undesirable for them. Once the loan is called, a borrower can either close their position by repaying, or refinance with a new loan.

When looking at loan terms as a borrower, the interest rate is of course relevant, but the main “risky” parameters to be concerned with are the call fee and call period. If these are too small/short, borrowers may take a loss when CREDIT holders seek liquidity. At any given time there may be more than one lending term per collateral asset. Constrained by the available liquidity, borrowers can seek out the terms they prefer.

We’ll make tools available that users can opt into to make it easier for them to “be their own oracle”, for example:

• an AutoRefinancer where once a loan is called, a tip is offered for anyone to rebalance the loan to a new lending term if one is available

• set up triggers to leverage or deleverage at certain prices as reported by an oracle like Chainlink

• notification system for once a loan is called (via email, discord, etc)

Participating in GUILD Voting

GUILD governance isn’t quite like anything else in DeFi, but the best comparison is Curve gauges, where token holders have a direct say in distributing system resources according to their pro rata ownership. The main difference is that GUILD holders can both gain from making good decisions, and lose from making bad ones.

As an individual GUILD holder, you’ll be able to nominate new lending terms, and vote in the gauges for existing terms. When borrowers pay interest on their loans, voters for the relevant gauge earn a portion of these returns. You can only unstake from a gauge if there is available liquidity. If loans in that gauge are revealed to be below the safe threshold once called, you may be subject to slashing.

We’re also thinking about ways GUILD holders can automate or simplify their governance activities, such as:

• a bot that calls loans that are below the safety threshold (so it can earn the slashing reward)

• a bot that calls loans that the operator is voting for that are too close to the safety threshold (so they can avoid being slashed)

• a contract where GUILD holders can express multiple votes, or votes in a range, and allow a potential borrower to “pull” their votes to where they wish to borrow. One could imagine a model where a holder says “I’ll lend against this collateral at the highest terms available. If you pay the call fee, you can kick existing borrowers and access my liquidity”

• a delegation contract where a GUILD holder can entrust another to their vote, possibly with certain limitations such as the delegate maintaining a certain amount of first-loss tokens

A more passive GUILD holder will want to either pick conservative lending terms, or use some kind of delegation or automation tool to simplify their experience. An active GUILD holder can lend under highly competitive terms, taking on more risk but earning more yield as well.

Multi-Denomination Borrowing

The ECG does not engage in collateral rehypothecation by default. Instead, it is possible to create new credit token denominations, such as an ETHCREDIT or EURCREDIT, which are the first two I would expect to see live. This provides a composable way for GUILD voters and borrowers to opt into collateral rehypothecation where appropriate, and for the guild to attract new savers in the denominations they prefer.

The Life Cycle of a Loan

So, with all that preparation, let’s take it from the top:

1. The Ethereum Credit Guild defines a CREDIT token, coming to an offchain consensus regarding its denomination and monetary policy such as a global debt ceiling.

2. The initial set of lending terms for that token is voted on, setting which collaterals will be accepted and at what rates.

3. GUILD holders vote in gauges to assign their pro rata share of the global debt ceiling among the approved lending terms.

4. A borrower accepts a term, providing collateral and minting CREDIT

5. The borrower sells these CREDIT tokens on an AMM or elsewhere

6. Seeing the CREDIT price drop slightly, someone buys CREDIT tokens

7. Eventually, either

   1. The borrower chooses to repay their loan plus interest. GUILD holders voting for this term get a portion, the rest is retained in the system surplus

   2. Someone calls the loan, and it is liquidated. GUILD holders voting for it get slashed if it is below the safety threshold. If not, the same things happen as in a).

8. Eventually, the lending term expires, and all outstanding loans are repaid.

9. GUILD holders vote to approve a new lending term.

10. Return to step 3.

The Initial Set of Lending Terms

There is an open Request for Proposals regarding the set of lending terms that will be approved for voting at launch. Broadly there are four categories I’m focused on :

• nonvolatile, highly liquid assets

  • collateral like stDAI, DAI/CREDIT LP, USDC/CREDIT LP

  • call fee equivalent to a normal swap fee on a stablecoin DEX

  • very short call period

  • low interest rate

  • low collateralization requirements

• volatile, highly liquid assets

  • collateral like ETH, stETH

  • moderate call fee

  • short call period

  • moderate interest rate

  • moderate collateralization requirements (lower than Aave, Compound et al)

• nonvolatile, less liquid assets

  • collateral like Ondo OUSG

  • moderate call fee

  • moderate call period

  • moderate interest rate

  • low collateralization requirements (lower than Flux)

• volatile, less liquid assets

  • collateral like ARB, OP

  • high call fee

  • long call period (borrowers need more notice to refinance or unwind which may require OTC)

  • high interest rate

  • high collateralization requirements

If you can see yourself minting CREDIT against a collateral asset in your possession, whether on this list or not and whether to provide liquidity or take leverage, I’d love to hear from you in the forum. This post by Fishy is a great example of how to introduce potential new assets or terms.

Next time, we’ll discuss plans for distributing the GUILD token, including how it is unified in the mechanism of lending terms and swaps for surplus buffer growth, use as emergency backstop, or redemptions. We’ll also discuss how CREDIT holders can express duration preferences and come together with GUILD holders to price a yield/maturity curve.