Reducing MKR Borrow Cap

Hello Compound community!

I’d like to gauge sentiment on reducing the cMKR borrow limit (currently set to 25,000 MKR). I think this could help mitigate risks to the Compound protocol and wider defi ecosystem, without negatively impacting protocol usage and financial KPIs.

Background

MakerDAO controls the DAI stablecoin system through a governance mechanism directed by MKR token holders. Broadly, there are two governance mechanisms available for MKR holders to influence:

  • Governance module: This component has administrative control over the Maker protocol. The majority of assets and admin privileges are safeguarded behind the “pause proxy”, which imposes a delay period before passed governance proposals can be enacted similar to Compound’s governance timelock. During this delay period, proposals can be cancelled with a new proposal that overtakes the queued proposal’s voter support.
  • Emergency shutdown module: This component allows MKR holders to burn their tokens to trigger global settlement (currently requiring a 100k MKR threshold of tokens burned to be triggered). All outstanding debts are settled against collateral at current oracle prices, and DAI becomes redeemable for a pro rata share of collateral (but may begin to shift in value from $1 due to market movements).

The use of a timelock delay (currently set to 2 days) with possibility to cancel malicious proposals significantly mitigates risk of attack on the governance module. But the emergency shutdown module has no delay on activation by design - the intent is that it could be used in case of a critical bug or oracle problem, where standard governance would be too slow to react. So the risk of malicious triggering of emergency shutdown has become the primary attack vector for Maker governance.

Maker has tried to mitigate this risk by raising the emergency shutdown threshold (minimum amount of MKR required to trigger ESM), from 50k to 75k and now to the current 100k threshold (roughly 11% of circulating supply). Further increases might be untenable, because they would reduce the ability of MKR holders to trigger shutdown in cases where it’s really necessary (eg. critical bug that could lead to collateral being stolen/minting unbacked DAI).

Maker primarily evaluates these thresholds from the perspective of rational attackers - eg. could an attacker earn a profit by maliciously shutting down the system? An attacker’s costs include cost of acquisition of necessary MKR tokens for the attack, while potential sources of profits include short positions in MKR, defi tokens, or other related crypto assets.

The availability of borrowable MKR tokens can significantly reduce the cost of attack - while the borrowed tokens would be burned as part of shutdown, the cost to repurchase MKR and close the loan could be much less after an attack vs purchasing MKR beforehand. By reducing the maximum MKR borrow limit, Compound could help reduce the risk of malicious shutdown. This would in turn make MKR and DAI less risky as collateral assets, which would reduce tail risk faced by the Compound protocol.

Proposal Ideas

Option 1: Deactivate MKR borrowing entirely

Comptroller > _setBorrowPaused > cMKR > True

This would remove the ability to borrow MKR from the cMKR market (similar to setting the borrow limit to 0 MKR). Existing positions would be unaffected, but no new borrowing positions could be opened.

Option 2: Reduce MKR borrow cap from 25,000 to 5,000

Comptroller > _setMarketBorrowCaps > cMKR > 5,000 E18

5,000 MKR is lower than the current supplied amount, so this wouldn’t have any immediate impact on the cMKR market. This would still allow for some amount of borrowing against this market (and more than enough to meet historical borrow demand across Compound and Aave), so Compound can continue to earn MKR reserves.

Cost Benefit Comparison

Benefits of limiting MKR borrowing:

  • Reduce tail risk of MKR and DAI collateral on Compound
  • Benefit wider defi ecosystem by reducing risk to DAI stablecoin system
  • Improve relationship with MakerDAO (partnership potential is already deepening with Compound D3M proposal, and this could further solidify friendly ties)
  • Gain business from MKR whales (many large holders may be reluctant to collateralize their MKR due to increasing governance risk, and reducing/eliminating borrowing could address these concerns)

Drawbacks of limiting MKR borrowing:

  • Reduce reserve growth of cMKR market
  • Marginally reduce utility of Compound money market (eg. users who wanted to borrow large amounts of MKR for non-malicious purposes like trading or market making wouldn’t be able to)

On balance, I think reducing or eliminating MKR borrowing would probably lead to an improvement in Compound’s financials. The cMKR market would see reduced reserve accumulation, but an influx of MKR collateralized borrowers could more than offset this with stablecoin and ETH reserve growth. Considering that the cMKR market has accumulated only around ~$1,000 worth of reserves in the 6 months since launch, this should not be a tough threshold to meet.

Assuming 3% stablecoin borrow rate and 7.5% reserve factor, offsetting the lost cMKR reserve growth (~$2,000 per year) would require only about $900,000 in net new stablecoin borrowing, which seems easily attainable.

Initial sentiment poll (non binding):

Should Compound curtail MKR borrowing?
  • Yes (Option 1): Eliminate MKR borrowing entirely
  • Yes (Option 2): Reduce MKR borrow cap to 5,000 MKR
  • No: Don’t change current 25,000 MKR borrow cap
  • Abstain

0 voters

If this gets significant support from the community, I’ll push this forward through the Compound governance process (including any needed technical or risk review, along with making formal on chain proposal).

4 Likes

I’ve closed the forum poll after 2 weeks. It seems a low borrow limit (5,000 MKR) is preferred to fully disabling borrowing - this would provide most of the benefits while still allowing enough liquidity to meet foreseeable borrowing demand.

I’ll try to schedule time to speak about this on a developer or community call and hear any comments. Stay tuned!

3 Likes

I agree with reducing the MKR borrow cap.

I’m posting this idea so that it does not get forgotten: we should create borrowing and supply velocity limits for problems like these and various other problems. In that, I mean limiting the speed at which certain assets can be deposited or borrowed. If this approach were to be taken, we could slow down the speed of attacks and increase their costs.

2 Likes

This is a great idea, Maker has implemented a similar mechanism with debt ceiling instant access modules, which allow for setting an absolute maximum debt limit while also limiting immediate utilization and daily increases in exposure.

@monet-supply thank you for this proposal. Gauntlet is conducting analysis from a market risk perspective and will provide results in ~2 weeks.

1 Like

Thank you @pauljlei and gauntlet team! Looking forward to this!

1 Like

Gauntlet Analysis:

The design intention for lowering borrow cap is more of a way to limit the losses related to an oracle attack, infinite minting, governance risks, and other technical risks. Gauntlet’s analyses focuses on market risk, which is different from these risks, so we advise that the community ultimately base their decision around their views on these risks listed above.

From a market risk perspective, lowering the borrow cap of MKR can reduce the likelihood of insolvency driven by MKR borrows (all else being equal), although we’d note that under current conditions our analysis already predicts low chance of insolvency driven by MKR borrow. Currently, there’s roughly $20M collateral locked for MKR and essentially no MKR borrowed. MKR has an annual volatility of ~110% and an ADV of ~$20M. Even if the entire MKR borrow cap of ~$50M becomes maxed out, the main market risk would be if one user borrows the entire $50M and the position becomes liquidatable. At that point, given the close factor of 50%, the liquidatable account’s $25M MKR would need to be repaid. Even in this case there is likely to be enough MKR liquidity on the secondary markets to liquidate that position (and prevent protocol insolvency).

To summarize: our analysis predicts that the impact of lowering MKR borrow limit, from a market risk perspective, is likely to be minimal. The community should make their decision from the perspective of oracle, infinite minting, governance, and other technical risks.

2 Likes

Thanks for this review!

With no indication that this proposal would add risk to the Compound protocol, and potential for it to reduce governance risk to Maker (and by extension reduce risk of DAI insolvency), I view this as a net positive from a risk perspective.

I think it could also be positive from a business development perspective - large MKR holders could be more willing to collateralize positions on Compound (increasing stablecoin/eth borrow utilization and reserve growth) knowing that their collateral can not be used in governance attacks.

Next steps:

  • Present to biweekly community or dev call (date TBD)
  • Confirm operations needed to execute this proposal, and have community review to ensure no unexpected results
  • Push proposal live on chain
3 Likes

Agree with this sentiment and next steps. I believe that the Compound community wants what is best for the Maker system and as such will readily lower the MKR borrow cap due to the arguments by Monet.

3 Likes

Happy to say this proposal passed and was executed a few days ago :slight_smile:

Thanks to everyone who voted or participated!

3 Likes