Should the DAI sitting in the xDAI Bridge be deposited into the DSR?

  • This post is to launch the discussion to deposit all the 25M DAI in the xDAI bridge into the DSR.
  • This proposal is to pass all yield onto a Savings xDAI contract.
  • Users on Gnosis Chain can mint GC sDAI with their xDAI which will accrue the DSR of all the DAI in the GC Bridge.
  • If less than 100% xDAI is deposited in the Savings xDAI contract in GC, then the yield will be larger than the DSR. For example, if the DSR yield is 3.49% and only half of the circulating supply of xDAI is in the Savings xDAI contract, the yield will be 6.98%. Theoretically, the DSR yield will always be higher on GC than Mainnet, as only if all xDAI were staked would we achieve rate parity.
  • The 25M DAI currently sitting in the bridge represent ~0.9M DAI yearly to incentivise GC. With this proposal, TVL is likely to increase.
  • DSR yield is risk-free if you are already holding DAI. All the risks derived from the collateral are borne by all DAI holders, regardless of their depositing in the DSR. We have written a research piece here.

Context

MakerDAO has raised the DSR to 3.49%. In the past 30 days, 7M DAI have fled out of the xDAI bridge, as can be seen on this dashboard. We need to do something to stop the bleeding and recover some deposits.

Implementation

The DAI currently in the bridge and any future DAI bridged to GC will be deposited into the DSR, resulting in pot shares being allocated to the bridge. These pot shares accrue in value based on the chi in the pot contract (rate accumulator).

To account for GC shares the Savings xDAI contract would have to issue its own shares relative to the DAI in the DSR. For composability purposes, the Savings xDAI contract should be ERC 4626, as they have an internal rate provider which can help to boost all current yields on GC. For example, from Balancer’s perspective, it makes a lot of sense to integrate sxDAI for their boosted pools, and making it 4626 should make the integration very simple.

We need to have a discussion about the best way to implement the passing of the yield. From the top of my head, I can think of a flow like the following:

  • xDAI bridge deposits all DAI in the DSR.
  • Every new deposit goes straight to the DSR, and every withdrawal is taken from there. At some point, we could implement a buffer to make withdrawals and deposits cheaper.
  • Every day, bridge validators calculate the increase based on the yield obtained from the DSR and mint the corresponding xDAI in GC. Rewards are sent to the sxDAI contract, increasing its totalAssets and therefore increasing assetsPerShare. A small withdrawal fee of 0.01% (roughly, a day of yield) should be charged to avoid deposits right before the accrual of rewards. These fees can go to the Gnosis DAO or remain in the contract, accruing to sxDAI holders.

Other alternatives to pass on the yield

While this proposal explores passing the yield purely to depositors of Savings xDAI, there are other options available, namely:

  • Charge a performance fee on Savings xDAI, that goes to the Gnosis DAO
  • Distribute part of the yield among GNO validators
  • GNO buybacks
6 Likes

Seems to be a good idea to me but I am not sure if we should use the whole bridge supply for this.
Have you made a calculation how much tx cost for bridging will increase if this would be implemented?

mmm that’s a good point, maybe a good workaround for this is to batch the deposits. Once per day, when bridge validators mint extra xDAI and send them to sxDAI, net deposits into bridge can be deposited into the DSR. This shouldn’t change the bridge-in cost significantly. Bridge-out would vary based on the undeposited DAI. We could also leave a buffer undeposited.

These are the current deposit/withdrawal costs:

  • It currently costs ~30k gas to deposit in the bridge (~0.0003 E @ 12 gwei)
  • It currently costs ~160k gas to deposit in the bridge (~0.002 E @ 12 gwei)
  • It currently costs ~180k gas to deposit in DSR (~0.003 E @ 13 gwei)
  • It currently costs ~160k gas to withdraw from DSR (~0.002 E @ 12 gwei)

As you point out, this would increase deposit costs significantly so batching deposits make a lot of sense, but I’m not so sure about withdrawals.

Big fan of the proposal.

A few comments:

There are ideas that at some point Maker/Spark could bridge DAI into a lending market on Gnosis to ensure that DAI can be borrowed at a low rate. For this DAI it would be important to not earn the DSR (only once it is actually in circulation)

There a number of ways to do this and the proposal is still in an early stage - so should not be a blocker for now - but just something to keep in mind.
The xDAI bridge already has the functionality to create xDAI on Gnosis from different tokens. So in theory we could create a wrapped DAI version for those direct maker infusions that do not go into DSR.

n this proposal we will need to bridge the information about how much interest has been created from Ethereum to Gnosis. This would be a great first use case for our new bridge design: Hashi

Quite likely other chains will be interested in such a solution as well and they might be able to do this based on Hashi.

At this stage where both Gnosis chain block space and defi activity is quite low we need to focus on giving as much as possible value to our users. Giving them a very attractive sDAI solution is a great move!

Thanks for the numbers! So even a withdraw from bridge will double in price if I understand it right.
Not sure how this might impact the use, but it will make difference especially for the small fish which we need to catch to get more usage.

Wouldn’t it be possible to give ppl the choice if the like to bridge to use sxDAI or xDAI which gave them either a cheaper bridging solution or they pay a little premium in tx fees but get an asset that appreciate in value.

Another solution could be to put only e.g. 75% of DAI in the DSR to always have enough liquidity to bridge. There has already been a thread about placing bridge funds to work some time before (Should GnosisDAO manage native bridge funds in a way that improves the Gnosis Chain ecosystem while managing risk properly?), some comments there apply also to the actual proposal.

Yeah, I think what would probably be best is to deposit something like 75% or so, that’s what I meant with leaving a buffer undeposited to process deposits and withdrawals more cheaply. It’s surely something we should take into account when defining the specs.

2 Likes

In addition to a buffer (which is a good idea to explore), other bridges based on AMM would still enable small fish to bridge xDAI to mainnet.