Ensuring Ethereum isn’t on the brink of a second bailout

An exploration into Multi-Collateral Dai and the systemic risks involved

This report assumes an understanding of the mechanics behind MakerDAO. For an in-depth look with related readings view the Vision Hill case study.

Context
MakerDAO is a peer-to-peer smart contract-driven collateralized lending platform that powers Dai, the industry’s largest decentralized stablecoin. The system allows users to lock cryptocurrencies such as ETH in a collateralized debt position, then mint Dai, a synthetic asset pegged to the US dollar. Dai is the lynchpin of most “Open Finance” activity today, and undergirds the majority of the decentralized lending market.

The MakerDAO platform is governed by MKR token holders via a decentralized autonomous organization (hence the “DAO”). MKR holders set the rules surrounding the system’s minimum collateralization ratios (150%), liquidation policies, interest rates, and supported collateral types. Until last week, the only collateral type accepted in MakerDAO was ETH, but that changed with the system’s upgraded support for “multi-collateral Dai” (MCD).

MCD has profound implications on the MakerDAO and Dai ecosystems. It not only changes the nature of Dai by adding additional collateral types with which it can be created but also makes key design changes that impact the economics of the system.

In theory, MCD increases the addressable collateralized debt market and should improve the system’s stability. However, this growth may also become an existential threat to Open Finance given the reliance on Dai and the absence of alternative trust-minimized stablecoins. If MCD allows Dai to scale to the size of legacy lending markets and more decentralized financial continue to be built on top of it, a black swan event affecting Dai could bring the whole system down with it.

Collateral Changes
As the name implies, multi-collateral Dai adds sources of collateral other than Ether. A recent vote was held in which MKR holders ratified changes that include the addition of Basic Attention Token ($BAT) to be locked in a vault (formerly known as a “CDP”) in order to mint Dai. With a market cap that’s only 2% of Ethereum, BAT won’t materially change the composition of debt in most vaults within the system, especially since the initial “debt ceiling” for allowable staked BAT is capped at 3 million Dai. From a risk management perspective, adding ERC-20 tokens such as BAT provides diversification and reduces collateral portfolio risk, although the assets themselves may be of lower perceived “quality” than Ether. That being said, the main benefit of including BAT and early ERC-20’s are that they will play an important role in acting as guinea pigs for wide-scale MCD.

Over time, as more assets are added with properly configured risk parameters, the vision is to add real-world assets (tokenized versions of equities, bonds, real estate, etc.) that would open up the collateral pool into the trillions of dollars.

This issue of adding trust-reliant assets has been controversial in the Maker community. After years of debate, tensions escalated culminating in accusations of a coup and the removal of several executives at the Maker Foundation who wanted the system to remain trustless in perpetuity. While there are still many opposed to the idea, it is clear the Maker team intends to move forward with the addition of non-cryptoasset collateral in the future.

What this would look like
One proposal that has generated interest is the Tokenized Asset Portfolio, which piloted a few months ago on the Kovan testnet a means to tap into the $23 trillion U.S. treasury market. An implementation like this one would require a Sponsor to set up a special purpose vehicle (SPV) that could purchase the underlying asset which would then be held by a qualified custodian, as required by U.S. law. Tokens would then be minted to represent an equity interest in the SPV and used as collateral to mint Dai.

The process is outlined below (from Fluidity):

In single collateral Dai (“SCD”), anyone in the world can send Ether to a smart contract and mint Dai. If the Ether price drops and the loan becomes under-collateralized, anyone can participate in the liquidation process.

The Tokenized Asset Portfolio model differs because it necessitates trusted third-parties to facilitate these processes. While it would add more historically stable assets, it would also open up new attack vectors if a large proportion of debt had a single point of failure. What would happen if the U.S. Government ruled that the entire Maker system was an illegal synthetic asset and seized the SPV. This problem will come into play no matter how real-world assets are added since inevitably there will need to be a known third-party holding the tokenized right to them.

The Maker team hopes these risks is mitigated by the MKR holders who have control over the system’s risk parameters. By limiting the percent of total debt in the system drawn out from any one type of collateral, they can ensure that even in the worst-case scenarios of complete loss of one collateral type, the system can remain intact. If this occurs it should have a positive reinforcing effect on the risk aversion of MKR holders who will be careful in adding more centralized assets.

Dai Savings Rate (DSR)
Prior to the upgrade, MKR holders had one lever to pull to influence the price of Dai – the Stability Fee (SF), in other words, the interest rate that debt holders pay to reclaim their collateral. As the price of Dai changed on the secondary markets, MKR holders voted to adjust the SF in order to restore the peg.

Earlier this spring, the Dai price was consistently below $1 leading MKR holders to act swiftly increasing the rate from 0.5% to over 20%. Intuitively, this increases the cost of holding debt which leads to more repayments. As the supply of Dai decreases, the price trends upwards restoring the peg. While it has been effective in its goal, changing the interest rate in this manner is detrimental to debt holders as their cost of capital changed by a factor of 40 in the span of about three months.

The Dai Savings Rate, a native interest-bearing savings account, was introduced to address this issue. It acts as a savings account for Dai holders who can deposit money and earn interest on their balances. Unlike a traditional bank, Maker doesn’t lend out the money in the savings account. Instead, interest payments are paid through the stability fund. This mechanism removes the credit risk inherent in money-market protocols such as Compound, which should, in theory, always have rates above the DSR. You could then think of the DSR as the de-facto “risk-free rate” for Dai in Open Finance.

By impacting the demand of Dai rather than the supply, the DCR can be used as another lever to bring Dai back to its peg. If there is ample demand for this “risk-free rate” then whenever the Dai price drops below its dollar peg, rather than increasing borrowing costs, MKR holders can opt to increase the DSR which should have a similar effect. The higher rates will lead to more demand for Dai to then restore the peg. However, since the DSR will always be lower than two-way lending platforms, lenders might prefer to take on the additional risk in return for additional basis points. In this case, the DSR will be ineffective as no additional demand for Dai will occur. This would mean the SF would remain the dominant tool to influence monetary policy.

An interesting dynamic at play is the tension between using the DSR as a tool for stability and maximizing the value of MKR. The value of MKR is derived from the SF used to burn it, but the DSR takes a portion of the SF to pay interest to those utilizing the DSR. As the DSR increases, less of the SF goes to burn MKR.

You can think about the whole system like that at a traditional bank where they hold people’s savings and pay out interest with money earned lending to other customers at higher interest rates. Banks want to maximize the interest they charge to people and minimize the interest they have to pay. In this case, the “bank” is really just the large group of MKR holders, so they would love to have a high stability fee and 0% Dai savings rate, but just like a traditional bank that is not sustainable. There will be a delicate balancing act between the two rates where the stability fee will need to account for the risk MKR holders take (this is the “bank’s” revenue) without getting too high that it becomes prohibitively expensive to borrow Dai. On the other hand, the DSR must be high enough to incentivize people to lock Dai so the system can maintain its peg. This is something MKR holders are willing to take since the DSR effectively stabilizes the cost of capital for debt holders. Even though margins per loan decrease, it could be value accretive if it spurs loan growth.

One Dai to Rule Them All
Dai is often championed as the standard of decentralization in the stablecoin world - and for good reason. Its only competitors are fiat-backed assets whose creators may be easier to censor or seize if they run afoul of existing law. This has traditionally afforded assets like Tether an advantage: they’ve been able to scale more quickly since more people trust dollar-backed tokens that they believe sit in a bank account somewhere. As a result, the total market capitalization and trading volume of fiat-backed stablecoins are much greater than Dai.