BIP8: Proposed increase to Insurance Limits on ETH and LINK liquidity pools

OLD GOV (LEGACY)
1

BIP8 TL;DR:

  • Create extra space for TKN deposits.

  • Enable a limit change from 1 million BNT, up to 20 million BNT minting limits on 2 pools:

  1. LINKBNT

  2. ETHBNT

Vote “for” to enable this change, or vote “against” not to enable it.

Summary

Background

With the acceptance of BIP5 by Governance and the recent contract upgrade, the DAO can now set hard limits for BNT co-investments by the protocol on a per-pool basis. This proposal seeks to exercise this new feature by significantly increasing the co-investment limits on ETH and LINK, from 1,000,000 BNT, to a maximum of 20,000,000 BNT. The passing of this proposal by governance will allow for adjustment of the minting limit within this range, contingent on prudent monitoring of its effects on the system.

Discussion:

The flexible insurance tiers proposal (Insurance Tiers ) was accepted by the BancorDAO on 1st November 2020. The flexible insurance schedule was to allow for improved management of insurance liability associated with assets in the network. Liquidity pools that generate high amounts of trade revenue relative to their arbitrage activity can be boosted with additional protocol co-investment through an increase in minted BNT limits. Conversely, pools that are underperforming may have restrictions imposed on their protocol co-investment. This gives the BancorDAO the flexibility required to manage the health of the system.

This proposal seeks to increase the BNT minting limits on two key pools – ETHBNT and LINKBNT.

The anticipated positive effects are:

1. An immediate increase in TVL.

Increased co-investment limits allow for additional single-sided token staking by both ETH and LINK holders. The pools currently have little space available to accommodate additional single-sided deposits at present. The newfound capacity should attract a commensurate level of single-sided staking, resulting in growth of the total locked value.

2. An increased rate of token burning.

The ETH and LINK pools have thus far proven themselves a minimal insurance liability during the initial phase of V2.1. Trade revenue on these pools has eclipsed arbitrage activity, and the trend is overall deflationary. Providing these pools with additional resources to support continued fee-burning is a convenient option to offsetting the cost of impermanent loss elsewhere in the network.

3. Migration of Uniswap LPs.

The most prevalent base token on Uniswap is ETH. Liquidity providers wishing to migrate their liquidity at Uniswap, to take advantage of Bancor’s IL insurance, face severe capacity limitations for the ETH component of their capital, owing to a lack of space in the pool. This is a pervasive problem that inhibits a large amount of potential erc20 liquidity from pouring into the Bancor protocol. Catering to the needs of Uniswap defectors is expected to have positive flow-on effects throughout the Bancor network.

Conditions:

This proposal asks for dynamic adjustments to the BNT minting amounts between 1,000,000 and 20,000,000 BNT; this proposal is not insisting on an exact figure. Rather, there should be an explicit understanding that there exists a margin for experimentation, whereby an optimal BNT minting limit can be determined. However, these experiments are strictly limited to within the stipulated range. Moreover, it is expected that a reasonable effort should be made to keep the community updated regarding where the minting limits are set, and the observed effects. For example, a dedicated channel on Discord, and the Bancor Announcements via Telegram could be used to provide weekly protocol health reports for these two pools, including token burn rates, increases in TVL, swap volume, and new user addresses.

3 Likes
2

To give some perspective, in the context of our risk assessment project we have looked at the impermanent loss coverage of the various pools. Below is the data from the launch of v2.1 until 4 December (note that recent price moves changed this quite a bit; more on this below). The below shows the cost of IL if everyone withdrew their stake today vs overall fees accrued by the respective pools.

Screenshot 2020-12-16 at 15.40.18
Screenshot 2020-12-16 at 15.40.181914×1190 361 KB

We see that at this point most pools have covered their Impermanent Loss with the fees they had earned. The one pool with significant loss in this dataset was OCEANBNT due to the significant price movement of OCEAN after liquidity had been added. ETH is outstanding, but arguably its importance here is overstated because only about half the transactions change ETH vs BNT; the other half exchange ETH/TKN where TKN is not ETH and arguably the driver of this transaction is TKN more than ETH so a fair allocation would move some of the fees to the respective TKN.

The idea in the proposed insurance limit increase, if I understand it well, is that the community is given the ability to experiment with the liquidity contribution to two key pools – ETH and LINK – on the assumption that an increase in pool liquidity leads to a lowering of slippage (and therefore a lowering of leakage to arbitrageurs) which in turn leads to an increase in volume and fees earned. This, in our view, is a sensible strategy. This does not mean that this is without risk, but, if properly controlled, can increase fee levels and therefore burn over and beyond the additional IL risk taken.

Note that with the big moves from 15 Dec the 4 Dec analysis is somewhat out of date. In particular, IL increased from around 500k BNT to around 1,500k BNT, with the biggest IL being in OCEAN, LINK, REN and then the stable coins. This however is due to BNT moves and should be more than compensated by the increase in BNT value.

4 Likes
3

While I agree with the general idea, I am unsure why we need to increase the limit 20 fold so early on. The system hasn’t had time to prove itself out and thus build confidence through big price swings. It may make more sense for the limit to be dynamic too, instead of being fixed at 1M, possibly by looking at the price ratio of two tokens in the pool (an oracle based solution may play nicely here). ie, BNT can only be minted to match up to 10% of the value TKN in the pool presently before more TKN can be added to the pool.

Additionally, after LM rewards re-staking is implemented, this BIP may prove to be irrelevant as much room should open in the pools. Shouldn’t we wait until we see how that proceeds?

4

The wording in the proposal is important: “this proposal is not insisting on an exact figure. Rather, there should be an explicit understanding that there exists a margin for experimentation, whereby an optimal BNT minting limit can be determined.”

The 20,000,000 number is an arbitrary ceiling; we are not proposing to immediately push to that number. We will start small, and slowly increase it to see how the system responds. The alternative is we set an exact number, and vote every few days to change it after observing the effects.

Agreed - but this proposal is supposed to cover a significant length of time - months to years. In all liklihood, the 20,000,000 ceiling might not get reached anytime time in the next 12 months.

5

I’m in favor of this

It seems that many people saw 20m and got scared. However the idea was to increase the limits incrementally. Having a high threshold gives the flexibility to do this. Each incremental increase (say by 1M or 2M at a time) would be closely monitored and further increases would be discontinued if detrimental effects are observed.

Some have also said why not just test the insurance increase on ETH at first?

Imo we as a community should want to test increasing limits on more than just ETH to get meaningful test results. ETH is an anomaly in the network as it already has very deep user-supported liquidity and, well, it’s ETH.

LINK was chosen because it has shown to produce greater fees than IL liability thus far and therefore poses minimal risk.

We as a community need to move fast to capture market share and this BIP is one way to do that.

6

I definitely agree that it’s time to get a better understanding of the protocol minted BNTs and how they interact and effect the rest of the protocol ecosystem. It is an incredibly powerful but at the same time potentially dangerous tool and we need to start experimenting incrementally and to see how we can best use it to our advantage.

When used properly it has the potential to get over certain liquidity depth thresholds to generate more fees (through being competitive liquidity depth wise on aggregators) and through proper understanding of the risks and levels we should be watching, we can apply these learnings and grow out both TVL and protocol generated revenue substantially in a way no other single change really can.

1 Like
7

Yes, that’s exactly it. I saw 20 million and it raised my eyebrows. I think part of the proposal should include a maximum increase rate. Say only increasing by 2M or 5M at a time.

I think after the discourse and reasons discussed here I am now in favor of BIP8.

8

I am wondering giving the large price increase of BNT what is the increase in BNT supply from newly minted BNT? My understanding is newly minted BNT will be burnt if more than required to cover the IL insurance, but those that got arbitraged will flow out of the LPs.

9

In favor! This proposal has solid reasoning and evidence, and is just increasing future options as opposed to fixing an action. In my opinion, increasing future options is a good thing.

10

Both the ETH and LINK pools deserve some extra dedicated liquidity from the bancor side. It would be really nice if the community could manage the quantity of the tokens in the pools. The 20M made me nervous as others have mentioned and i came to the reasoning that 5M for eth and 2M for link would be nice balance. Maybe those numbers aren’t right but i WOULD trust the community of vBNT to properly manage it.

11

Vey well thanks so much