Last Resort Minting Mechanism

Protocol
1

One of our primary goals and design principles in the Celo ecosystem is to minimize the risk for vulnerable stablecoin end-users. That is to transfer as much risk as possible from stablecoin users to the reserve and ultimately CELO holders.

Celo stablecoins are one of the if not the safest crypto stablecoins available, mainly due to the high level of over-collaterization from diversified assets. As I’m writing this, there are ~150 million USD in stables outstanding, backed by ~215 million USD of non-CELO assets and ~330 million USD of CELO, resulting in an over-collaterization ratio of ~3.8. And this after last week’s crypto market turmoil. Even if CELO alone would go to zero, all stablecoin users could still redeem their funds. Over the next few years, an increase in the allocation of natural assets in the reserve will further strengthen the stability, especially in times of crisis.

Our margins of safety are high, but the past week has shown how things can change. Maybe there are ways we can increase the robustness & reliability of the protocol in worst-case scenarios even more. With worst-case scenarios I mean a possible loss of funds to stablecoin users by heavy under-collaterization and bank-run like scenarios.

The core questions I am asking myself are what can happen should the reserve become under-collaterized? Is the protocol well prepared for different scenarios? Is there anything else we can add to the protocol to make sure that stablecoin users don’t lose in that case?

Some other protocols have emergency seigniorage token minting mechanisms in case of shortfall events or under-collaterization built into their protocol. Examples would be Maker MKR or AAVE.

I’d like to kick off the discussion whether we want to have and should start developing such a feature ourselves. I would understand such a feature as a mechanism of last resort: If the stability protocol would be about to fail, or to reduce the chances of even running into such a scenario, the protocol could borrow from it’s future value to try to survive.

What could an emergency mechanism look like? In case of under-collaterization, it would allow the protocol to mint CELO tokens and send them to the reserve. It could basically refinance the reserve by diluting or even hyper-diluting CELO.

Trade-offs

There are a couple of advantages. If well designed, such a feature has the potential to prevent under-collaterization & bank-run like scenarios by creating trust that the stability protocol has a last resort mechanism for such difficult situations. Also it could prevent loss of funds of stablecoin users by mindfully borrowing off from the future and paying back later.

But as with everything, there are risks. If poorly designed, last resort minting could reinforce negative market dynamics, creating feedback loops which worsen the damage to the protocol compared to just sitting crisis scenarios out. Designing such a feature would be a bit about timing a market crisis well, which is hard. If there is already a large supply & demand imbalance in markets and CELO is crashing, such a feature could reinforce this. The only case of such minting in the real world I know of was Iron finance. And well, that didn’t go well.

Would love everybody’s thoughts and perspectives on this.

5 Likes
2

Random thoughts (some good, some probably bad):

  • Timelocked cStable/CELO LP-token bond sales a la Olympus.
  • Some sort of exit queue functionality that increases in duration as the reserve becomes increasingly undercollateralized
  • Some form of de-pegging insurance policy for cStable holders (see example)
  • Something like CELO Temple Defend (basically a private call options market)
  • Something like Alchemix Transmuter where users could lock cStables to borrow a synthetic stable. You could probably automate some sort of purchase of CELO with interest earned from all outstanding synthetic cStable vaults until collateralization reaches parity again.
  • Pause cStable minting until reserve re-reaches collateralization (rate limiting)
  • At a certain point just using non-CELO reserve assets to buy back and burn CELO probably makes sense.

I don’t have much more time to explore this but I created a taxonomy of monetary and fiscal polices that might make sense in different ways in this WP (pages 21-22 and 30-32).

2 Likes
3

A couple of follow up design dimensions I’m tinkering on:

  • At what reserve collaterization ratio the minting sets in: 0, 0.5, 1, 1.5, 2?
  • How fast and how much is minted. Until the collateralization ratio is above 1, 1.5, 2, 2.5 again?
  • What happens once a crisis is overcome? The reserve could buy back CELO tokens and burn them to restore initial network shares. Or the heightened, non-scheduled supply could stay in the ecosystem? We could design an automated, hard-coded central bank monetary policy: Borrow the protocol’s future value in bad times, and pay back the debt by buying back and burning CELO tokens in good times. What humans can’t seem to do well in the real world, smart contracts could enforce.
  • What about the 1 billion CELO supply cap. I think this cap is fundamental to create predictability, long-term deflation when the protocol succeeds, and to reduce ambiguity about everyone’s network share. But if the protocol would be about to die and stablecoin user funds are at risk, why not take chances and loosen that cap? With a respective mechanism burning could happen later to go back to normal times.
1 Like
4

I’m unsure about this idea. I think a coin either has hard cap or it doesn’t; and it’s a great thing that CELO does. As soon as the protocol may print money to keep the reserve afloat, that additional trust is broken.

Imagine what CELO holders will think in case the reserve comes too close to being under-collateralized; they would expect that their holdings are likely going to be diluted soon. This could actually increase sell pressure, decreasing the reserve holdings, increasing the problem. In contrast, it’s unclear if printing could really stop a bank-run which would be a result of widespread disbelief in CELO’s ability to survive.

3 Likes
5

Thanks @Tobi . I think collateralisation is a critical topic to keep talking about and improving.

Technically, by minting CELO and placing it in the reserve, this moves wealth from those holding CELO to the reserve. This makes sense from the perspective of having CELO holders take on stability risk. It is difficult though because of the hard cap on CELO.

Some alternative ideas:
a) Make it so that a portion of locked Celo is confiscated and transferred to the reserve (economically equivalent to the above but only affecting locked celo holders) [probably unpopular because we’ve already said locked CELO is not subject to risk, although minting to the reserve would also be a new risk - although to all CELO holders]. On the other hand I think this may be a better market dynamic for controlling locked Celo rewards than simply having a target locked percentage.
b) Create a second form of locking where your CELO is at risk of transfer to the reserve, but you also get some bonus rewards. I think this complicates things a lot.

2 Likes
6

@Tobi , how’s this going? I’d love to see something happen here on last resort mechanism, cheers

7

@Pinotio.com thanks for your thoughts, this it what I think would be the most promising direction. I want to explore both options, minting and giving users the possibility to participate as stability provider in return for some reward. The necessary tinkering and research is however a blocked: we’re updating our simulation environment in the cLabs economics team for making all these kinds of ideas and decisions more rigorous and understandable. Once this is done, our plan is to make it open source, tinker on such features, and make recommendations on what features to add to the whole stability setup. Particularly about adding lines of defense like last resort minting / stability participation

2 Likes
8

Howdy @Tobi and all:

Worth reading this thread on Luna/Terra about their recent decision to create a $1B Bitcoin collateral reserve: https://twitter.com/WestieCapital/status/1496223635396255747?s=20&t=MuCCzrcYmqy1tkeIFdMKoQ

If you think about Luna as analogous to Celo (and UST as analogous to cUSD), then:

  • The Luna system currently has no well defined reserve collateral to support the price of Luna. It does, however, have the ability to infinitely mint Luna to maintain the peg of UST.
  • The Celo system currently has well defined reserve collateral to support the price of Celo. It does not have the ability to infinitely mint Celo to support the peg of cUSD.

So:

  • With Luna buying a $1B Bitcoin reserve they are doing what the Celo reserve is already doing.
  • By considering the last resort mechanism in this report, Celo would be doing what Luna already is doing.

Miscellaneous thoughts:

  1. Maybe I’m missing something, but it seems crazy to me that Luna/Terra had no well defined reserve until now.
  2. There seems to be $10b+ of UST outstanding (with just this new $1B in collateral of Bitcoin) - so Terra is much less well collateralised than Celo (which is overcollateralised). [Note that the market cap of Luna is $20B, so in theory you could dilute out Luna and shift that $20B in value to back UST. The problem is, in a downmarket there’ll be reflexivity, so if Luna is crashing, there wouldn’t be a way to save UST by minting more and more Luna. This same limitation applies to the last resort mechanism that is being proposed around Celo {which is not to say the mechanism has no purpose, but it’s effect is limited}]
  3. The price of Celo (and UST) should theoretically have a floor that is based on the price of Bitcoin (and Ether for Celo), due to their reserves. [It’s a complicated relationship to calculate what that floor might be, but in theory I think there is some floor.]
3 Likes