[Discussion] Direct deposit DAI module (D3M)

MakerDAO is the decentralized minter of the Ethereum ecosystem, pioneers in Decentralized Finance (“DeFi”). The MakerDAO protocol minted the DAI stablecoin asset, which acts as a hedging tool and medium of exchange directly on-chain.
The cornerstone of DeFi is overcollateralization–always having more collateral than assets minted–which means that there is always room for incentivization of a third party to pay back debt on behalf of borrowers, keepers being repaid and retributed by borrowers collateral through an action mechanism.

The MakerDAO protocol proved itself resilient to a 94% drop of the value of ETH in 2018 during the SCD era and the black Thursday event in March 2020.

The MakerDAO protocol is heavily overcollateralized–each makerDAO vault is an independent entity in a mutualist system, with its own collateralization ratio and thus liquidation price. On this basis, the overall DAI collateralization risk should be analyzed according to system global collateralization and critical collateral price levels creating levels of liquidation thresholds.

Data analysis of DAI collateralization


5.5B$ of ETH collateralized for 1.53B$ of DAI minted. ~360% collateralization on ETH-A vaults

DAI has seen significant growth since Ethereum’s “DeFi Summer” of 2020, growing from a 289M DAI supply to a 3.1B$ supply in 7 months ~+800%

The current “global MCD DAI collateralization” is currently historically low due to recent debt ceiling increases at around 230%. Typically 3 USD worth of collateral backs each DAI when considering the protocol as a whole.

Exploring D3M

Currently, there is simultaneously a stablecoin liquidity shortage in DeFi as well as high demand for stablecoins in DeFi protocols to be used for earning rewards or leveraging/borrowing, which creates high interest rates that potentially could have a negative impact on ecosystem growth as a whole.

DAI “liquidity crush” also creates a significant premium between “cost of money” defined by makerDAO Stability Fee and supply rates on protocols such as Aave.

In theory, the free market should create itself an equilibrium as the premium gap is an earning opportunity for interest rate arbitrage actors.

In reality, market immaturity, more lucrative investment opportunities, “Yield Farming” by leverage loops deposit into DeFi and Debt ceilings limits the natural balancing of premiums between mint cost and supply earnings. Moreover, DeFi is constantly evolving and new products further increase the demand for Dai across the whole network.

The D3M is an exploration of an alternative path to bring more liquidity in secondary DAI venues by minting DAI backed by aDAI and depositing it directly in secondary markets to release liquidity shortage pressures.

The D3M module is a potential new tool for the MakerDAO to stabilize DAI peg, increase DAI attractiveness for borrowers and create new pathways of revenues for the MKR ecosystem.

Additionally, the Aave Community is deploying new liquidity pools across various L2/chains, which means more DAI can bridge into pools to transact affordably and in an inclusive manner, while still ensuring the security of the MakerDAO collateral in L1.

Essentially, Aave could become the distributor for DAI on every L2 and empower new growing communities as the need for stablecoin as a medium of exchange grows across all chains and L2s.

DAI backed by aDAI could be minted leveraging the overall protocol overcollateralization, with limits in size and timeframe set by global protocol collateralization target ratios and secondary market rates target ranges.

The focus of the D3M is to mint & Burn DAI in exchange for aDAI to stay within the Target Max Variable borrow rate of DAI in Aave Liquidity Pools.

Mechanics of D3M

The D3M module mints and directly deposits freshly minted DAI into the Aave V2 protocol, and by doing so, the D3M vault module collects an equivalent amount of aDAI (which grow over time based upon the fact that the Aave DAI pool generates interest).

The D3M has a target rate for limiting the size of the deposit and ensuring DAI to be the most competitive asset to be borrowed on Aave compared to other stablecoins.

The target rate is fixed in such a way that the variable borrow rate does not exceed a threshold value.

This threshold value can be set by the MakerDAO community to ensure that DAI is the most attractive stablecoin for Aave borrowers.

Interest rate curve model allows for precise sizing of mint/burn events for more information on the interest rate curves on Aave, please refers to the dedicated documentation.

The size of the vault could be experimental and later adjusted by the MakerDAO governance as there is more market data. Eventually D3M has the potential to become a global way for the MakerDAO community to inject DAI liquidity across DeFi and even incentivise risk-averse protocols by adjusting liquidity between protocols.

D3M would be the natural step for MakerDAO to expand DAI across DeFi and ensure market positioning as new stablecoins are evolving, including Compound Cash. Working together with the Aave community could position both the MakerDAO and Aave communities uniquely to ensure DAI liquidity across DeFi and decrease the liquidity fractionalization on L2s in mid-term.

What is backing aDAI?

aDAI is an automatically-generated, native token to the Aave protocol, which is issued to a user who supplies DAI into the Aave protocol. aDAI is astandard ERC-20 token, which is natively balance-increasing such that the DAI pool on the Aave protocol generates interest to all suppliers to the DAI pool. he aDAI smart-contract continuously updates the balance of all aDAI token holders (including D3M Vault) to reflect this accrual by increasing their aDAI balance.

As with every deposit into the Aave protocol, there’s no lock-up period and withdrawal can be requested at any time within the limits of available DAI liquidity. There is no slippage associated with aDAI and every aDAI is redeemable for one DAI at a 1:1 ratio. Suppliers of aDAI are covered by the Safety Module (managed by Aave Governance) in case of shortfall event. DAI itself is backed by the systemic overcollateralization of the lending market, where overcollateralization is based on the Aave Risk Framework. Additionally, Gauntlet is performing on-going analysis on the V2 borrowing market’s Market Risk and other related risks.

Additionally, there’s currently ~1.6B$ worth of AAVE and AAVE/ETH in the Safety Module with a 30% slashing parameter activated, meaning that nearly up to 0.5B$ of deposits are protected in case of a shortfall event.

Benefits for MakerDAO

  • External source of direct income by collection of aDAI interest
  • MakerDAO governance would have direct impact on risk assessment by adjusting the D3M minting across DeFi
  • Expanded supply of DAI including L2s
  • DAI attractiveness for DeFi borrowers on secondary markets
  • If Aave community starts a Liquidity Mining program, MakerDAO will collect AAVE assets and earn governance rights in Aave Governance

The technical implementation for D3M module is described in MIP-50 posted by @hexonaut. Kudos to @marczeller and @statelayer for contributing to put the discussion proposal together.

39 Likes

Just wanting to voice my personal support for this idea. The implementation is not too much different from just adding an aDAI vault, but we get the added benefit of targeting a desired interest rate on Aave. This will allow Maker to continue having the lowest rates in the space. There is also no messing around with the Stability Fee and vault owners - we just get all the interest automatically.

That being said, this addition is not without risks (as with all collateral additions). I’ve outlined what I perceive to be the risks and security considerations in MIP50. In particular by more closely integrating with Aave we expose ourselves to both smart contract and liquidity risk from their platform. We have to ask ourselves whether the benefits outweigh the risks. I believe they do.

I posted on the Aave forum as well if you want to discuss with the Aave community. Looking forward to the tighter cooperation between these two DAOs.

19 Likes

Is there an opinion on how deep/liquid a market must be to make use of this and maintain use of this?

I like the concept, but we also don’t want to end up taking on risk that’s not properly managed in size and quality.

I’m all ears and don’t have an opinion myself, but would like to hear some others’ thoughts

Very nice.

I assume the upper limit of the aDAI vault will be Collateral_Value-Dai-Minted?
This is potentially a huge sum, improving the economic efficiencies in the whole space. We will need some rule on how to reduce supply should the crypto markets crash.
Use is however not limited to Aave, but could be expanded to other platforms as well I guess. But using Aave now gives us an immediate L2 access.

Very nice indeed.

2 Likes

Thanks for this post, this is a very interesting proposal.

This sounds very good.
Would it be possible to describe roughly how this protection actually work? What if the price of AAVE drops (just together with a shortfall event)?

The limit is the debt ceiling same as any other collateral. We can enforce the max borrow up to the debt ceiling, but no more.

2 Likes

One consideration.
The target rate (R) will likely have to be greater or equal (\geq) than all SF of all collaterals of MakerDAO: R \geq \{ \max_{c}SF_c | c \textnormal{ is a Maker collateral}\}.

For example, now we have SF=6% on BAT-A.

If we set R on Aave at 5%, this will make Aave more competitive than MakerDAO when borrowing with BAT. So we need R\geq 6%.

I think this prospects a direction where we can get rid of small-cap collaterals like BAT (and many many more to come over the next few years) and leave their ‘management’ to Aave.

We should need to remain the best vanue for borrowing with very large DC collaterals such as: ETH, WBTC.

Of course this means that we will have to manage the risk of Aave failing.

So:

  1. POSITIVE: we can lend a lot more DAI for small-caps coins like BAT, without consuming the precious time of the devs and risk-teams and governance (voting time, fatigue).
  2. NEGATIVE: We are getting more exposed to Aave risks.
6 Likes

We really are speed running the evolution of traditional finance. What a time to be alive.

5 Likes

I personally am not a fan. That’s a lot more risk by being dependent upon them for large swaths of potential business, and leaves us as a middleman who can be cut out or squeezed. Perhaps for really small-cap collateral, that’s ok. But I fear that leaves us really undiversified in risk both from the perspective of being at the mercy of Aave and increasing reliance upon ETH for our own line of business. We need lots and lots of different revenue streams, but don’t want to just hug an external partner while they pick our pockets.

Correct me if I’m misinterpreting :slight_smile:

3 Likes

Let me take again BAT-A as an example.

With current 2,510,763.88 / 3,000,000 utilisation/DC, I claim that it is not worth our time anymore to spend: devs time, risk-teams time, governance-time on it.

Let’s delegate this to Aave. Basically people borrowing DAI from Aave instead of Maker are those having altcoins not supported directly by Maker.

Ultimately there will be a lot of them. We can’t handle them (not cost effective), imho. It’s better to deal with just one entity (Aave) on which we can perform a high-level risk evaluation (again, imho)

4 Likes

So are you referring to new collaterals when you say this? Or existing? Or both?

I am referring mostly to new collaterals: there are hundreds of shitcoins around, and people want to borrow DAI with them.

I am also (but less so) talking about current small-DC collaterals like BAT-A. I am not sure it is a good investment of time for MakerDAO’s risk teams/governance/dev to discuss and analyse and update BAT-A on a total of 3m DC.

We should only focus on those large-DC collaterals that make us real money: ETH, WBTC, UniV2, …
→ less coins/tokens, deeper risk evaluation and safer(*) SC-code

(*) I know it’s already safe :slight_smile:

3 Likes

I’ll need to think more, but for now you have me convinced. Especially as things like digital yuan coming can put us in a position to both be ForEx and privacy at the same time. Throw in some way to do euroDAI, and you’re correct the opportunity cost of picking up a few bucks is huge

5 Likes

Yes, MakerDAO should focus on being the lender to other lenders in order to minimize the cost of onboarding new/small collateral vaults. Instead, focusing on our core competencies ETH, WBTC, RWA, etc. Aave’s backstop of staked liquidity also goes a long way to insure the integrity of the entire ecosystem (maybe we should do likewise). It costs too much time/resources to be adjusting tiny little vaults’ Stability fee, debt ceiling. and so on. So outsource this altcoin lending to others willing to do it.

3 Likes

Worth noting that we can target a desired interest rate with a traditional aDAI vault as well - there would just be a little bit of an interest rate spread to accommodate vault owners’ cost of capital and profit margin.

Eg if we made an aDAI vault with 110% liquidation ratio and 8% stability fee, this would soft cap the Aave DAI rate somewhere in the 9-12% range (equivalent to a ~10-30% return on a leveraged aDAI vault).

In a way I agree, this sort of arrangement can help Maker and DAI address long tail asset borrowing demand. But I don’t think aDAI exposure is equivalent from Maker’s perspective. Secondary lenders like Aave and Compound use a pooled risk model which has higher risk of insolvency - if any individual collateral asset blows up it can be used to borrow the remaining assets. I’m not sure if this asset is suitable to accept at an effective 100% collateral ratio.

Edit: In the example above the Aave supply rate would be soft capped at 9-12%, with the borrow rate being somewhat higher based on the interest rate model and utilization.

6 Likes

@PaperImperium currently the stablecoin supply across all Aave Protocol markets is around 1.5 B USD worth (V1, V2, AMM and Polygon L2). DAI market share is around 13,74 %. Being able to mint directly from D3M would mean that the borrow rate of DAI would decrease and we would see swaps and loans positions opened in DAI instead of other stablecoins.

From technical risk respective, the V2 of the Aave Protocol is audited by 5 different independent auditors and formal verification is done by Certora. More on the audits here. On the market risk and stress testing, Gauntlet should publish their official report within the next days that covers V1 and V2 markets. Since D3M is bespoke feature and changes the dynamics of MakerDAO to become lender to lenders, I think it’s better to be as much risk-averse as possible and start from conservative debt ceiling.

@iammeeoh in Aave Protocol first line of collateralization is the asset overcollateralization in the protocol provided by the borrowers. However, if there is some kind of short fall event such as exploit or failed liquidation, the protocol has Safety Module that backstops the deficit. In practice, Aave community members stake AAVE into the backstop module and receive incentives to allow 30% of their stake to be used per every potential slashing event. If the slashed amount does not cover the whole deficit - then there is Recovery Issuance similar to MakerDAO model which aims to make the protocol as whole. MakerDAO was used as an inspiration to build this part of the module. The Gautlet Market Risk Assessment report also covers the Safety Module and its dynamics. One of the Main concerns is that the SM consist mostly of AAVE currently, however the Black Thursday events works as a precedence that even minting new supply did not affect MKR value substantially and might guide that backstopping protocols with protocol’s tokens works in practice.

@PaperImperium on the revenue, the D3M factually create a new revenue stream to MakerDAO since it would mean that liquidity could be injected even when it would be most needed and risks can be fairly mitigated by capping the total exposure on D3M, I would see it as a way to enable more DAI liquidity as well for assets that might be smaller caps that could be financed against in secondary market protocols like Aave Protocol. Interesting value proposition is also that with Aave markets, DAI could be distributed across L2s where Aave Protocol is being deployed meaning that DAI would have more advantage of being not only the medium of an exchange but the yielding stablecoin of that particular network, which would empower users that do not afford to use DeFi in the current gas costs.

10 Likes

Ok. I’m tentatively convinced. I’m open to hearing more, but you have definitely addressed my top-of-mind, initial reaction!

1 Like

Uh, I mean, sort of. MKR went from $600 to $200 in that event. It did recover (obviously) but we still sold a bunch to recapitalize at a much lower price than we wanted to. One could argue that the minting was priced in as soon as the market realized what had happened.

If the safety module needs to be used, it’s likely that AAVE (and possibly ETH as well) has already dropped by some amount, either in the precipitating event or as a result of the shortfall.

It’s correct that it will work to a point, but that point is heavily dependent on the perceived ability of the system / protocol to recover and is probably less than one might expect given the token’s market cap.

I don’t have a strong opinion on the proposal itself, but just wanted to comment on that particular statement. Appreciate you coming by with this proposal!

5 Likes

Yes, which meant a 400m (from 600m to 200m) value drop …
All caused by a loss of only about 12m DAI.

So I think:

is debatable. At the same time, backstopping with protocol’s tokens is the core proposition of MKR in MakerDAO…

So yeah, I guess we need to iterate on that and make sure it will work better in the future.

3 Likes

Does Aave have reserves / surplus buffer that helps absorb losses?

Also, is there a description of the security and risk mitigation process when doing new collateral onboarding?

To some extent I think having multiple markets that are isolated from each other already helps to some extent, since the main V2 market can focus on having the safe and slow onboarding of higher value assets, and then other, more experimental markets, like the AMM market, can move faster and take more risk. Maker can just allocate its debt ceiling appropriately to take less exposure to the riskier markets

2 Likes