# Redemptions Research

> Redemptions are forced liquidations while the system issued asset is below peg to improve stablity

## Overview

A borrowing system with redepemtion mechanics allows its issued collateral to maintain a soft peg to the redemption value given by the system. Imposing a fee on redemption lowers the minimum threshold for redemption value, for example, a 0.5% redemption fee sets a price floor of the synthetic asset as 99.5% of the represented asset. The net effect of a redemption is to repay debt in the system and remove equivalent collateral to remove issued assets from the market. The reduced supply of minted assets in the market should drive the price toward peg when the market price falls below `1:1 * (1 - Redemption Fee)`.

>Example using Ethereum @ $1000 and Bitcoin @ $10000

| Collateral | Debt | Assets | Collateral Ratio | NAV | Action |
| --- | --- | --- | --- | --- | --- |
| 10Ξ | 0.5₿ | 0.5₿ | 2 | 10k - 5k + 5k = 10k | n/a |
| 7.5Ξ | 0.25₿ | 0.5₿ | 3 | 7.5k - 2.5k + 5k = 10k | Redemption of 0.25₿ |

A redemption in the system removes assets from the market and deleverages participants. The delta in net asset value assuming price remains equal is zero, however the composition of the assets have changed. This can impact the user experience of a given trade as your position is adjusted during redemption.

### Easing or Reducing Redemptions

During volatile periods of liquidity, or lackthereof redemptions can happen frequently due to larger, faster depegging. An example of such events are the UST incident that saw the depegging on chain of several USD equivalents. While this event is a panic, it cannot be discounted.


[Source](https://dune.com/21shares_research/stablecoin-monitor)

It is clear even with the most common methods or trusted forms of sythetically issued assets, depeg risk during crisis is possible and redemptions would be possibly substantial based on the size of the system.

Exploring current implementations for redemption mechanics as well as possible alternatives can allow us to explore the following areas:

#### ICR (Internal Collateral Ratio) Based Redemption

Liquity, as an example, targets the lowest collateral ratio troves for first redemption. This strategy pays back the loans of the most highly leveraged troves unwinding their position, and effectively reversing a portion of their trade position.

Two observations can be made on this position:

- User has a higher conviction on this trade than others
- User is a higher risk to the system than others

Targeting the lowest collateral ratio in the system increases the TCR the most per issued asset redeemed trivially. This behavior is most beneficial when the system is in a critical state or needing to enter recovery mode as it actively derisks the system while helping issued assets return to peg.

The measurable benefits to the system would need to be quantified with respect to peg and safety of the system as exploring other heuristics for redemption would not yield as high of a "safety per redemption" score as riskiest troves first.

##### Alternative Approaches

- Socialized redemption across all open troves
- Trove deleveraging to a Maximum Redemption Collateral Ratio
- Starting with the least risky troves
- Unwinding all troves to an equivalent collateral ratio (cannot redeem past x CR)
- Subsequently redeeming in a socialized manner
- Alternative approach to above
- Start with riskiest troves
- Unwinding all troves to an equivalent collateral ratio (cannot redeem past x CR)
- Subsequently redeeming in a socialized manner
- Step wise redemption unwinds
- Redeem each trove in batches up to y CR at a time to prevent full liquidation

##### Feasibility

The following are unknowns as to whether features are feasible to implement in an efficient manner:

- Ability to apply socialized redemptions, at time of redemption, to specific troves
- Ability to dynamically swap between redemption strategies

#### Protocol Fees

Borrowing protocols studied at a general level utilize several fee types to modify user behaviors: origination, borrow, liquidation, and redemption fees. These fees have an impact on stability, and thus the frequency of redemptions.

##### Origination Fees

Origination fees are fees deducted from a borrow when a user borrows the synthetic asset against their collateral initially. A dynamic origination fee can be used to disincentivize new borrows as a larger fee makes the position increasingly less favorable.

An example of origination fees being high would be in the case of a collapse in the price of the syntetic asset. Redemption would begin pushing a dynamic borrow fee higher - thus disallowing (hopefully) the synthetic asset redeemed supply from returning to the market.

The origination fee may act as a deterrent to users to utilize the protocol so it is important to think about the attractiveness of a loan with such a fee given the intended user or use case for the loan.

###### Possible Use Cases

- Short Asset
- Farm with Asset
- Cross Market Long (Degenerate Trading)
- Borrow synthetic asset
- Deposit in secondary money market
- Borrow stables
- Buy more tokens

The only case where an origination fee seems to be a reasonable price to pay is for farming opportunities, or in demand borrow of the synthetic asset - a role which is anyway played by the system itself.

Thus, for most participants likely this fee is more of a nuisance than something that makes sense and also potentially hampers proliferation of the system synethetic asset.

###### Alternatives

- Reduce or remove the origination fee

##### Borrow Fees

Liquity as an example has no fees associated with their borrowing model. This decision means there is no natural demand for the synthetic asset. Such a decision leads naturally to a lack of demand for the asset on the open market outside of redemptions below peg without a secondary incentive on the issued asset, farming for example.

The nature of liquidity pools offering farming rewards is the general drive for demand on the asset but in the long term may not prove to be a sustainable source of demand based on the need to compensate liquidity providers for their service and holding that asset.

A borrowing fee incentivizes natural demand for the asset from the borrowers ensuring a more speedy or potentially more consistent demand for the asset. The properties of said asset would impact its desirability to borrow along with the rate at which borrow is available.

Money markets like Compound and Aave generally have low BTC borrow rates - sometimes as low as 1-2% due to demand. A noteworthy statistic is that while on both markets using BTC as collateral has decent amount of assets, the utilization on the borrow side of those markets is quite low.

An interesting investigation would be to check the utilization of BTC borrow on protocols such as GMX for any kind of increased usage in that type of market.

###### Conclusion

Initially the idea of an organic demand for a synthetic BTC product seems like a good idea. The rates, however, would have to be exceptionally low. The main difference is the risk of centralization posed by an asset, however when borrowing in a short position that risk plays in the favor of the borrower.

Borrow fees may organically help the demand of the asset, but given the state of the market demand on Ethereum for BTC - it cannot be assumed it will be a substantial demand source.

##### Liquidation Fees

Liquidation fees impact redemptions by causing external actors to purchase the synthetic asset off the open market to ensure the safety of the system. Liquidation fees, or bonus, is paid to the liquidators to ensure healthy system and derisking of the protocol.

This should lead to market demand and aide in peg stability. The fees should not be removed as they incentives protectors of the system and benefit the main goal of redemption reduction as well.

The value of the stable coin is given by the ability to liquidate by it as the stable coin discount is part of the premium you'd get in participating in liquidations (+ the liquidation incentive).

##### Redemption Fees

Redemption fees ensure a soft price floor buffer to prevent constant redemptions. The fee is directly proportional to the floor as a soft peg. An earlier example shows that a 0.5% redemption fee soft pegs an asset to 99.5% of the value of the asset it is a synthetic of.

This redemption fee can be dynamic to attempt to disincentivize mass redemptions or to slow the pace of redemptions while providing a stability to the borrowed asset.

This fee is important as a throttle on redemption speed, however a parameter configuration of 0.5% may be too aggressive in an illiquid or nascent system - especially with low liquidity if the goal of the system is to allow a user to sell the borrowed asset for leverage or a trade.

###### Modifications

The user experience of redemptions present some non ideal scenarios where a users position may be deleveraged at unfavorable rates if the sythetic asset peg is lost.

It presents an issue between causing many slight inconveniences to users vs. larger position changes to specific users.

A few ways to avoid this behavior could be:

- Changes to redemption base fee paramter to allow for larger depeg
- Time delay on redemptions to allow users to capture the redemption against troves willfully
- Time delay and fee decay on redemption parameters
- High starting redemption fees
- After X hours redemption fee decay starts
- Minimum redemption fee in place to keep mechanism as a soft peg
- Dynamic redemption volume increase to fee persisted from original implementation
- Fee restoration at or above peg

#### Protocol Stability Module (PSM)

Liquidity pools allow for market price discrepancies to cause a deviation from peg of a synthetic asset. Assets may be backed by sufficient collateral but due to market demand this is not being followed.

A decentralized PSM can present a serious challenge with respect to the decentralized aspect of the system. Using BTC as an example, a system backed by ETH and minting a sythetic as eBTC has no decentralized (or sufficiently desired) equivalent redeemable for BTC to trade for.

Using the Maker DAO PSM as a comparison, the ability to swap for USDC <> DAI during certain peg conditions reduces the impact on value lost when interchanging the currencies due to market conditions. An interesting use case to avoid redemptions and allow for eBTC, for example, to regain peg without such a market impact would be the ability to redeem eBTC for a bitcoin variant directly backed by Bitcoin, similar to the USDC <> DAI relationship allowing the swap to an asset backed by the actual pegged equivalent.

In practice, this would allow eBTC to be redeemed directly for BTC on the Bitcoin chain utilizing, for example, Ren Protocol as an intermediary for such a transaction. This renBTC would be collected by the PSM during times eBTC is over peg due due to farming related demand or inadequate liquid supply with imbalanced demand for closing positions possibly.

### Conclusion

- Dynamic origination fees starting at zero and increasing during periods of redemption may be a good compromise in lieu of a static origination fee while still getting the benefits during periods of redemption
- Low borrow fees are likely required to remain competitive in the given landscape, how those borrow fees scale or change is up for debate
- Dynamic borrow fees is the default based on utilization for money markets
- Tiered borrow fees based on CR is the standards for Maker DAO
- Protocol stability module is an idea that reinforces the decentralized nature and soundness of the bitcoin based synthetic should the asset used for swaps be deemed acceptable
- The PSM would ideally be able to alleviate some initial depegging seen by the asset prior to redemptions being allowed
- A protocol stablility module with assets to allow for swap might be an interesting prequisite for allowing the redemption fee to drop at all while there are still assets avialable for a pre redemption swap
- Redemption heuristics could stand to be more egalitarian especially within a redemption
- Only redeem a trove stepwise up to specific collateral ratios per redemption
- Preserves the invariant of removing the riskiest positions, but in a more controlled way