CIP-3: Compound V2 to V3 Migration

Preamble

CIP: 3
Type: Protocol Enhancement
Title: Compound V2 to V3 Migration
Description: Community migration plan from V2 to V3
Author: Gauntlet
References: Compound V2 → V3 Migration Planning

Abstract

Following the community’s preference for V3 Migration, Gauntlet provides a migration path for the community.

Specification and Rationale

Below is Gauntlet’s initial migration proposal for the Compound community. During the migration process, we can adjust the plan in response to changes in TVL, utilization rates, and general market conditions. Note these dates are tentative and subject to delay depending on how governance proceeds. The ~10 day buffer between the end date of one proposal and the start of the following proposal is to allow time to post in the forums, create the proposal, and wait for execution.

2023/02/20-2023/03/06 2023/03/17-2023/03/31 2023/04/11-2023/04/25 2023/05/07
Levers
USDC Comp distribution migration (v2 → v3) Transfer 50% v2 2023/02/13 value to v3, for a total of 241.2 tokens per day. v2 supply and borrow incentives will drop to 120.6 per day. Transfer 75% (+25%) v2 2023/02/13 value to v3, for a total of 361.8 tokens per day. v2 supply and borrow incentives will drop to 60.3 per day. Transfer 90% (+15%) v2 2023/02/13 value to v3, for a total of 434.16 tokens per day. v2 supply and borrow incentives will drop to 24.12 per day. Transfer 100% (+10%) v2 2023/02/13 value to v3, for a total of 484.4 tokens. v2 supply and borrow incentives will drop to 0 per day.
v2 USDC Reserve Factor (currently 7.5%) 15% (+7.5%) 17.5% (+2.5%) 20% (+2.5%) 22.5% (+2.5%)
V3 Supply caps Increase current v3 supply cap for ETH to 350k tokens & WBTC to 12k tokens. No changes to COMP, UNI, & LINK. Will reassess if utilization increases Increase if appropriate depending on supply cap utilization Increase if appropriate depending on supply cap utilization Increase if appropriate depending on supply cap utilization
Storefront Price Factor (currently 50%) 60% (+10%) 65% (+5%) 62.5% (-2.5%) 55% (-7.5%)

Below we analyze the goals, execution plans, and mitigation strategies for each of the proposed parameter changes. We are using data from 2023-02-03.

v2 & v3 net USDC supply & borrow APYs

To incentivize v2 USDC suppliers to migrate to v3, the net USDC supply rate (supply APY + supply COMP distributions) on v3 must be greater than that of v2. Similarly, to incentivize v2 USDC borrowers to migrate to v3, the net USDC borrow rate (borrow APY - borrow COMP distributions) on v3 must be less than that of v2.

Below are the supply APYs for v2 and v3 as of 2023-02-03:

v2 v3
USDC supply APY 1.57% 2.41%
USDC COMP supply distribution APY 0.79% 0%
USDC net supply APY 2.36% 2.41%

The net v2 USDC supply APY is 2.36%, and the v3 USDC net supply APY is 2.41%. So USDC suppliers are currently only slightly incentivized to migrate to v3.

Below are the borrow APYs for v2 and v3 as of 2023-02-03:

v2 v3
USDC borrow APY 3.30% 4.10%
USDC COMP borrow distribution APY 1.53% 2.71%
USDC net borrow APY 1.77% 1.39%

The net v2 USDC borrow APY is 1.77%, and the v3 USDC net borrow APY is 1.39%. So USDC borrowers are currently moderately incentivized to migrate to v3.

Updating the v2 & v3 USDC COMP distributions and v2 USDC reserve factor can further incentivize USDC suppliers and borrowers to migrate to v3. This is also motivated by the fact that v3 COMP APY would decrease if usages increase and we don’t migrate rewards.

USDC COMP distribution migration

Execution

We aim to migrate 100% of v2 USDC COMP distribution (supplies & borrows) to v3 by the end of the migration. This will increase the v3 USDC net supply APY and decrease the v3 USDC net borrows APY, thus incentivizing v2 USDC suppliers & borrowers to migrate to v3.

At first, we propose to migrate 50% of v2 USDC COMP distribution (supplies & borrows) to v3 to provide an initial impetus for users to migrate. We then propose to gradually migrate the remaining 50% USDC COMP distribution over in accordance with the above schedule.

Mitigation plan

If we see a large swing in v2 and v3 utilization rates for an extended period of time, the community can adjust USDC COMP borrow & supply distributions accordingly to help balance the markets.

USDC reserve factor increase on v2

Execution

Increasing v2 USDC reserve factor decreases v2 USDC supply APY, thus further incentivizing USDC suppliers to migrate from v2 to v3.

The current USDC reserve factor is 7.5%. At first, we propose to moderately increase USDC reserve factor to 15% and then gradually increase it to 22.5% by the end of the migration in accordance with the above schedule.

Note that a small percentage of v2 USDC suppliers use their USDC borrowing power to borrow v2 assets, which may serve the new purpose of the v2 protocol. Therefore, it may be an acceptable outcome for these users to remain in the v2 protocol. Even though they will gain less net supply APY on their USDC as a result of our changes, their usage suggests that they will be less elastic to these changes, given they most likely value their USDC borrowing power more than net USDC supply APY.

Mitigation plan

Similar to the COMP distribution plan, the community can adjust the v2 USDC RF accordingly to help drive the desired v2 utilization rate while monitoring how the intended v2 USDC suppliers respond to these changes. Also, the community can revert the v2 RF increases if we see users migrate from v2 to other protocols instead of v3 and shift focus to making v3 more appealing instead of making v2 less appealing.

Supply cap increase

Below are the v3 supplies and supply caps for ETH and WBTC as of 2023-02-03:

Supply Supply cap Supply cap utilization
ETH 100,650 ($166.43M) 150,000 ($248M) 67.1%
WBTC 3,588 ($83.47M) 6,000 ($139.61M) 59.8%

Goal

Increasing supply cap increases the protocol’s risk, but it’s necessary to grow the protocol.

We aim to accelerate the migration of assets with lower volatilities and larger liquidities from v2 to v3.

Execution

We suggest initially increasing v3 supply caps for ETH and WBTC to 350k (~$580M) and 12,000 (~$279.22M), respectively. Note these caps are very close to the current v2 supplied amounts for each asset.

Assuming a successful migration in the first two weeks, we should see an increase in ETH and WBTC supply. Given current market conditions and similar user health factors, our simulations indicate the protocol should still be able to absorb the greater quantity of possible liquidations. Often, the insolvency risks we see in our simulations are due to users who supply lower-liquidity assets such as COMP at high borrow usage, which isn’t as much of a concern with ETH and WBTC.

As of 2023-02-01, WETH has 24h trading volume of $889,896,262 and ±2% market depth of ~$42m from trusted exchanges on Coingecko, and WBTC has $103,018,115 volume and ~$20m market depth. Based on the 24h volume and quote at 3% slippage, liquidators should be able to sell blocks of ~40m of WETH and blocks of ~20m WBTC with roughly 2% slippage, well-calibrated with a 5% liquidation bonus * 60% storefront price factor = 3% discount rate.

As the migration proceeds, we can assess increasing supply caps for lower-liquidity assets. These assets, however, generally pose greater insolvency risks, and we would perform further liquidity analysis first.

Mitigation plan

If market conditions and/or user positions change such that our simulations indicate that these larger supply caps pose outsized risk, the community can decrease the supply caps. Additionally, the community can increase the storefront price factor to further incentivize healthy liquidations, as discussed in the next section.

Storefront price factor

Execution

We suggest initially increasing storefront price factor from 50% to 60% (+10%).

Assuming the migration is successful, we should see increased v3 asset inflow and a resulting increase in overall protocol risks. Increasing storefront price factor will further encourage turnover of absorbed assets from Comet to liquidators and help facilitate healthy liquidation cycles in the system.

However, increasing storefront price factor does not monotonically decrease the system risk of the protocol. Although increasing it facilitates liquidations and prevents the protocol from holding non-stablecoin assets over time, large storefront price factor inhibits reserve growth. Therefore, assuming the initial migration plan is successful and the system is relatively stable, we aim to slowly decrease the storefront price factor to ensure healthy long-term reserve growth.

Mitigation plan

If the protocol isn’t accumulating enough reserves as a result of our storefront price factor increase, the community can revert the changes.

Next Steps and Timeline

We welcome the community’s feedback and plan on taking this to vote on 2023-02-13. We are targeting first round of proposals going live on 2023-02-20, but this is subject to delay depending on how governance proceeds.

By approving this proposal, you agree that any services provided by Gauntlet shall be governed by the terms of service available at gauntlet.network/tos.

7 Likes

Hello @pauljlei,

Thanks for moving forward with this topic. I find that the strategy fits well the Compound community’ strategy.

I just want to question the V2’s reserve factor increase. As some members of the community mentioned, including myself there, the strategy should not include actions making the V2 natively worse. Moreover, as far as I know, increasing the reserve factor doesn’t decrease the risk of the protocol. In a first approach, the V3 improvements along with the COMP rewards migration may already have a big impact of the V2 to V3 migration.

5 Likes

I tend to agree with @MathisGD here; if the intent is to encourage migration, the COMP distribution updates are the most appropriate lever. If the intent is to explicitly discourage continued use of V2, then the reserve factor increases are an appropriate lever. But there are plenty of reasons that folks may prefer to continue using V2. Yes, it fragments liquidity, but the providers of that liquidity have different goals (and possibly risk preferences).

Changes that encourage new users to select V3 make sense; changes like the reserve factor increases that are only punitive to users who choose to stay in V2 are less desirable.

4 Likes

Hey @pauljlei,

Thanks for putting this together. I also agree with @MathisGD and @allthecolors, I think using COMP incentives as a soft lever to migrate users initially is preferred. We can maybe revisit the idea of increasing reserve factors if we believe incentives aren’t having a substantial impact on migration.

Thanks!

2 Likes

Thanks, @MathisGD @allthecolors @WintermuteGovernance and all for your feedback. From the discussion on the community calls, we were under the impression that the community viewed Compound V2 and Compound V3 USDC Comet as having different use cases. In particular, Compound V3 USDC’s primary use case is to be the protocol geared towards borrowing stables. And for Compound V2, the primary use case is not geared towards borrowing stables but rather focuses on other use cases (borrowing volatile assets against volatile assets, shorting assets, etc.).

Under this assumption, increasing USDC RF for Compound V2 would not necessarily “make Compound V2 worse” because the goal of V2 is not for users to borrow stables. E.g., if the goal is for users to borrow stables, then USDC yield for suppliers is very important in order to bootstrap USDC supply. But if the goal is not for users to borrow stables, then the primary use case of USDC on Compound V2 would be using USDC as collateral to short other assets, in which case, the supply yield of USDC is not as important for V2 users. To summarize, increasing the USDC RF for Compound V2 would not make Compound V2 worse but rather align the USDC asset on V2 with its updated primary use case while incentivizing Compound V3 to be the primary protocol where users borrow stablecoins.

Our original proposal included RF changes. If the community is still against increasing RF, we are happy to update the proposal, given that this is primarily a strategic (not risk) related decision. But we hope that this clarifies for the community the tradeoffs.

3 Likes

Echoing @MathisGD 's comment (thank you also to @WintermuteGovernance for your comment) from the perspective of an integrating protocol, I don’t think it’s correct to say that Compound v2 has only one use case. While it’s true that lending USDC on Compound v2 offers the possibility of using it as collateral to short other assets, it’s also true that Compound v2 has many more integrations built atop it than Compound v3, including aggregators, leveraged token strategies, interest rate swaps, and more. I’m all in favor of migrating COMP incentives to v3, but doubling the reserve factor seems to penalize these other use cases needlessly.

3 Likes

Gauntlet Update

Following community feedback, please see below an updated step 1 of the migration plan to align with the community’s strategic preference.

Following this initial stage, we will monitor how conditions evolve, which will inform the next stage of migration.

  • Decrease v2 daily USDC supply COMP rewards from 241.20 to 211.20 (-30)
  • Decrease v2 daily USDC borrow COMP rewards from 241.20 to 211.20 (-30)
  • Decrease v2 daily DAI supply COMP rewards from 241.20 to 211.20 (-30)
  • Decrease v2 daily DAI borrow COMP rewards from 241.20 to 211.20 (-30)
  • Increase v3 daily USDC borrow COMP rewards from 161.41 to 281.41 (+120)
  • Increase v3 ETH supply cap from 150k to 350k
  • Increase v3 WBTC supply cap from 6k to 12k
  • Increase storefront price factor from 50% to 60%

Analysis

Note: Assumes COMP price is $50.

Current v2 COMP distributions

Daily Supply Daily Supply (USD) Daily Borrow Daily Borrow (USD)
DAI 241.20 $12,060 241.20 $12,060
USDC 241.20 $12,060 241.20 $12,060
USDT 34.74 $1,737 34.74 $1,737

Compound distributes roughly $50k COMP/day (~$18.25M COMP/year) to stablecoin users in Compound v2.

Current v3 COMP distributions

Daily Supply Daily Supply (USD) Daily Borrow Daily Borrow (USD)
USDC 0 $0 161.41 $8,070

Compound distributes roughly $8k COMP/day (~$2.95M COMP/year) to stablecoin users in Compound v3.

Current v3 daily reserve growth

Data is from 2023-02-15.

Current borrows: $104M

Current utilization: 68%

Below is a table of various daily reserve growth and earn APRs given the current borrows in Compound v3 at various utilization rates. Note that v3 daily reserve growth is only positive between 50% and 85%.

Utilization Daily reserve growth Earn APR
0% -$4,200 0.03%
50% $28 0.03%
68% (current) $1,800 2.21%
86% -$40 5.0%
90% -$1,520 6.60%
95% -$2,860 8.60%
100% -$3,720 10.60%

Given the current v3 state, the daily reserve growth is $1,800/day.

On the low end of the utilization curve, the max loss in reserves would be $4,200/day at 0% utilization.

On the high end of the utilization curve, the max loss in reserves would be $3,700/day at 100% utilization.

The reserves scale linearly with borrows.

Notable v2 non-recursive USDC borrowers

Address Supply tokens Supply Borrows
28617487 WBTC, ETH, BAT $146.2M $66.2M
6e5665bd ETH $27.7M $14.1M
64639cf8 ETH, WBTC $19.6M $10.4M
7499bba9 ETH $22.2M $10.0M
ec64ca41 ETH $39.2M $9.1M
14c3cfdd ETH, WBTC $14.3M $7.0M

Supply (USD)

Borrows (USD)

Notable v2 non-recursive DAI borrowers

Address Supply tokens Supply Borrows
fbe5ddc3 ETH $36.0M $24.7M
aa43ba42 WBTC $10.7M $4.9M

Supply (USD)

Borrows (USD)

Notable v2 non-recursive USDT borrowers

Address Supply tokens Supply Borrows
e9605af5 ETH, WBTC $19.1M $7.3M

Supply (USD)

Borrows (USD)

Goal

For the first phase of the migration, we aim to incentivize some of the largest non-recursive v2 accounts to migrate to v3. The above accounts supply solely ETH and/or WBTC, except for the largest account, which also supplies some BAT. There is currently $104M worth of borrows in v3, with 161.41 daily COMP borrow distributions, for a borrow distribution of 2.77%. To incentivize an additional $80M worth of borrows, we would add an additional 120 daily COMP borrow rewards to v3, and evenly decrease v2 USDC & DAI supply/borrow rewards. The initial increase in rewards would result in a borrow distribution of 4.83%, which would result in a positive 4.83% - 3.86% = 0.97% net borrow APR. Theoretically, the protocol would again reach equilibrium at $184M borrows, at which point the utilization would also be 68%, with the same earn/borrow APR, and the borrow distribution again diluted to 2.77%.

Risks

Increasing v3 COMP borrow rewards will initially result in higher utilization, which could result in short-term reserve losses if the utilization exceeds 86%. However, even in the most extreme case of 100% utilization, the daily reserve loss would only be ~$3,720 (in actuality slightly higher given increased borrows), which is low relative to the current reserves of $638k and low relative to the $58k daily losses in COMP distributions. And it seems likely that within a short time, suppliers would flock to the protocol to take advantage of the high supply APY.

Other risks of this plan include:

  • Assumes borrowing demand exists.
  • Could cause increased unforeseen market downturn risk if the health factors of new users are risky.
  • Users from other protocols could migrate before v2 users get a chance.
  • Requires setting higher initial supply caps to support the increased borrows.
  • At a certain point, recursive borrowing in v2 will cease to be profitable, at which point some DAI and USDC recursive borrowers will close or reduce their positions, which will also affect v2 supply/borrows, utilization curves, and distribution APYs.

Analyzing v2 recursive borrowers

Notable v2 recursive USDC borrowers

Address Supply Borrows
bc84d1eb $71.7M $51.7M
adc355ef $43.2M $36.6M
81c793f3 $20.6M $17.1M (USDT)
8717d99a $14.8M $12.3M
feecf8e1 $14.2M $11.0M (USDT)

Supply (USD)

Borrows (USD)

The top 5 recursive USDC borrowers account for $164.5M USDC supply (27.9% of total USDC supply), $100.6M USDC borrows (32.3% of total USDC borrows), and $28.1M USDT borrows (24.4% of total USDT borrows). Note that if these recursive borrowers leave v2, the USDC and USDT distribution APYs will substantially increase.

The current DAI net earn APR is 2.34%, and the net borrow APR is 2.01%. If we decrease v2 USDC supply & borrow daily COMP distribution by 30, the new net earn APR will be 2.26%, and the new net borrow APR will be 2.17%, still somewhat incentivizing recursive borrows. Note that as some of the recursive borrowers leave, others will be reincentivized.

Notable v2 DAI recursive borrowers

Address Supply Borrows
e0a9070c $160.6M $112.4M
10772c4e $82.9M $68.6M
6b4a437d $22.7M $15.0M

Supply (USD)

Borrows (USD)

The top 3 recursive DAI borrowers account for $266.2M DAI supply (47.5% of total DAI supply) and $196.0M DAI borrows (67.5% of total DAI borrows). Note that if these recursive borrowers leave v2, the DAI distribution APYs will substantially increase.

The current DAI net earn APR is 2.18%, and the net borrow APR is 1.82%. If we decrease v2 DAI supply & borrow daily COMP distribution by 30, the new net earn APR will be 2.09%, and the new net borrow APR will be 2.00%, still somewhat incentivizing recursive borrows. Note that as some of the recursive borrowers leave, others will be reincentivized.

Conclusion

If we assume that users are elastic between borrowing USDC and DAI, the v2 reward distribution shouldn’t matter as much as the quantity.

To incentivize some of the top non-recursive v2 USDC borrowers to migrate over, we recommend initially adding 120 daily COMP distributions to v3 USDC borrow rewards and decreasing v2 DAI & USDC borrow & supply daily COMP distributions by 30.

In anticipation of the increased v3 USDC borrows, we recommend increasing the ETH and WBTC supply caps to 350k and 12k, respectively, and increasing the storefront price factor to 60% to further incentivize liquidators.

Next Steps

  • Targeting an on-chain vote on 2/26/2023.

By approving this proposal, you agree that any services provided by Gauntlet shall be governed by the terms of service available at gauntlet.network/tos.

4 Likes

On-Chain Proposal has been published below. Voting begins in 2 days. We thank the community for their participation in this initiative.

1 Like

The proposal has passed and been executed. We thank the community for their participation in this first stage of migration.

4 Likes