I would like to use this thread to discuss the pros / cons / and risks associated to adding UMA on lending protocols (e.g., Aave, MakerDAO) for UMA token holders to be able to borrow against their UMA holdings as well as being able to lend out their UMA tokens.
If someone could borrow UMA on a lending protocol the bad things they could do would be abusing the DVM by submitting a bunch of disputes to earn inflation rewards or borrow so much that they have more than 51% to manipulate the outcomes on liquidations/disputes/governance votes. If the latter were to happen, UMA could increase the CoC through its buy and burn mechanism.
The DVM is not automatically monitoring the CoC > PfC and automatically buying and burning UMA. This process is currently manual. Maybe this is a requirement before listing UMA on a lending protocol?
(Cross-post from another thread I just made on this subject. Let’s consolidate discussion here!)
There are many reasons why someone may want to borrow against their UMA tokens, including asset diversification, tax minimization, acquiring leverage for yield farming, taking a levered long position on UMA, and so on.
However, there are security risks to making UMA tokens available to borrow on protocols like Aave. An unlikely scenario is that enough tokens could become available to borrow that someone could acquire > 50% of the voting tokens and corrupt the DVM. This is something we could see coming from a mile away and figure out how to address, so let’s ignore that possibility for now, though acknowledging that it meaningfully decreases the Cost of Corruption.
A more realistic risk is someone borrowing tokens to boost their voting rewards so much that it would be to their advantage to raise spurious disputes to the DVM, since their voting rewards would be greater than the final fee, and as a borrower of UMA token holders, they have no long-term skin in the game. They could raise disputes, borrow tokens before the snapshot, sell their rewards, and pay back their loan for a profit.
UMA as Collateral on Maker and UMA
An alternative to making UMA available to borrow on lending protocols is making it a collateral type on Maker, covering the common use case where you want to lock up collateral to borrow Dai. Intuitively, I think this would increase the Cost of Corruption, since it removes UMA from circulation, and in the event of an UMA price dip, borrowers would need to buy UMA tokens to lock up as additional collateral or pay down their debt to avoid getting liquidated. This would add an additional layer of price defense before we get to the buy-and-burn outlined in our documentation.
Another alternative along the same lines is making UMA available as a collateral type within the UMA protocol itself . This is more useful, since you can mint pretty much any kind of synthetic, instead of only being able to mint Dai. It has a similar security benefit, removing UMA from circulation and requiring borrowers to buy up UMA in the event of a price dip to shore up their collateral or pay down their debt. Fees would be paid in UMA tokens, which would go into the store.
These UMA tokens in the store would be useless for a buy-and-burn, for obvious reasons, but would be removed from circulation. I guess you could burn them in the event of a price drop, but I’m not sure the burn would serve any real purpose.
At first glance, I can’t see any risks to the DVM of having UMA as a collateral type, only benefits. I think that having UMA as a collateral type would provide an additional layer of price defense before we would have to resort to a buy-and-burn from the store.
We would get similar benefits by having UMA as a collateral type for MakerDAO.
Edit: There’s also a positive feedback loop if UMA is a collateral within UMA. More UMA tokens locked up as collateral means fewer UMA tokens on the open market, which could lead to a higher price. Both the higher price and the number of tokens locked up as collateral (and unavailable to buy or borrow) increases the Cost of Corruption, which expands the number of synthetics that the UMA protocol can support, which increases the value of the UMA token, etc., etc. This is in addition to the existing value protection where if the CoC is getting too low, the tokens in the store can be used for a buy-and-burn.
We would decrease the Cost of Corruption by making UMA available to borrow on protocols like Aave, though the risk may be manageable, and a number of governance tokens are on Aave already. I think the risk of a borrower raising spurious disputes to the DVM to cash in on voting rewards is a realistic attack, though.
If we wanted the ability for people to borrow UMA-like tokens on Aave, and UMA was a collateral type within the UMA protocol, we could set up a uUMA synthetic that tracks the price of UMA but conveys no voting rights. The collateralization ratio could probably be close to 1:1, or maybe exactly 1:1, since the synthetic price literally tracks the price of the collateral. Need to think about that a bit more.
This uUMA could then be made available on lending platforms like Aave without increasing the risk to the DVM. In fact, there may be a continuous increase in demand for UMA to lock up as collateral to mint uUMA that can be loaned to people who want short-term price exposure to UMA and aren’t concerned about not being able to vote. So, like locking up UMA to mint other synthetics, this probably increases the security of the protocol by removing actual voting UMA from circulation.
Edit: If there is borrower demand for uUMA, this might lead to upward price pressure for the UMA token from people buying UMA to mint uUMA to lend to other people and earn interest. Since a higher UMA price means a higher Cost of Corruption, this is a security benefit.
Edit x2: A cool thing about this is that, if we think this is a good idea, it’s extremely easy to execute. Adding UMA as collateral and a price feed could be done in a single voting cycle and you could deploy a uUMA synthetic immediately after that and try to integrate it with lending protocols.
UMA Short Positions
One thing to think about is that allowing UMA to be used as collateral within the UMA protocol also allows people to short UMA. I don’t know if this creates a risk to the DVM. At first glance, I think it’s actually a security benefit, since UMA locked up as collateral is not available to purchase for corrupting the DVM when the price drops, and someone looking to close out their short will need to buy UMA tokens to pay back their debt, further reducing the tokens available to an attacker.
I was told that we could campaign to AAVE that $UMA only be added as a collateral only, and couldn’t be borrowed.
I’m a big fan of having it added to AAVE with full lend/borrow ability.
I think this is an excellent point. I’m not sure how to navigate it!
I suppose the interest rate for borrowing UMA might naturally settle at a level where the lender captures most or all of the voting rewards for borrowed tokens, so this may not be an issue in practice.
uUMA is quite a neat idea.
However I’m not sure that Comp/Maker/Aave would accept a synth as collateral rather than the actual token. Its an added risk for them to bear.
Another option for a synthetic UMA would be that rather than tracking the price of UMA directly, it tracks the TVL in UMA contracts. There is a pre-existing relationship between the price of UMA and UMA product TVL because of the PoC>CoC inequality, but that isnt linear. The “UMA premium” over the base measure is the voting rights that the UMA token confers.
This would be a cool synthetic but I wonder who would take the short position by minting the synth when UMA’s TVL is on a steep upward trend, and actually increasing faster than UMA’s token price. If $UMA were actually trading at a premium to the increase in TVL, I guess it would make sense as a hedge, but the recent token price (1.4X?) is lagging the recent TVL growth (8X) by a significant amount.
That trend has its limit tho, because of the CoC>PoC.
I just saw that Opium have kindof done the reverse of uUMA,:their base token has no voting rights, but can be wrapped for voting.
This happens to be the point I wanted to make regarding this aspect. However, since the issue here is more that the token value is always going to be greater than the TVL, it may open a way to arbitrage the token if it is based on TVL. As much as I try and think of an alternative, the token price is the easiest way to go about this
this observation aged poorly