[Proposal] Increasing the capital efficiency on Anchor

[EJ_Lee]

Terraform Labs.

Governance proposal : Anchor Protocol

TLDR

• Increase bLuna Max_ltv from 50% to 60% to enable users to borrow 20% more while maintaining the same liquidation risk. This change would increase the overall capital efficiency on Anchor and would ensure Anchor’s sustainability long-term.

Max_ltv determines maximum borrowing using a specific asset. If the Max_ltv increases, borrowers can borrow more with the same collateral value. This means that when the Max_ltv value increases, the capital efficiency for that specific asset improves. By increasing Max_ltv , we expect the borrowing demand to increase. Increasing Max_ltv from 50% to 60% means we allow users to borrow 20% more maintaining the same liquidation risk.

However, capital efficiency has a crucial trade off, which is protocol safety. The higher the capital efficiency (or higher the Max_ltv value), the higher the safety risk of the protocol.

So the goal of this proposal is to gain maximum capital efficiency while ensuring the safety of the protocol.

The best way to get the optimum number for safety is to test the worst case scenarios we can obtain from historical price data. To obtain the optimum number that satisfies both conditions, we use two types of price data. The first source of price data is from the bLuna on-chain oracle and the second one is from the Binance Luna-USDT pair 1min candle.

1. bLuna on-chain oracle data

• Data length : from Anchor genesis (March 17th) to May 31st
• Maximum price drop(%) : -15.398932921948353%

2. Binance LUNA/USDT pair 1min candle

• Data length : from Luna/USDT pair listing to May 31st
• Maximum price diff. (%) : 25.699463%
• Assuming that this price diff. is a negative price diff.

The worst case scenario would be: at price X, the user borrows until the LTV hits right before max_ltv and price drops y%(Which is worst price drop percentage that we can find in historical data), with the liquidator taking that loan with a max_premium_rate , which is currently at 15%.

We will split the simulation results to ‘Safe’ and ‘Default’. ‘Safe’ means that the loan itself is under max_ltv after the liquidation; ‘Default’ means that there is an outstanding loan post-liquidation with no collateral remaining (a.k.a. cannot be repaid through liquidation).

The simulation result is as follows:

		max_premium_rate
	Max_LTV = 55%	15%
price drop(%)	-15.40%	Safe
	-26%	Safe

		max_premium_rate
	Max_LTV = 60%	15%
price drop(%)	-15.40%	Safe
	-26%	Safe

		max_premium_rate
	Max_LTV = 65%	15%
price drop(%)	-15.40%	Safe
	-26%	Default

Considering that this simulation assumes worst case scenarios with very low probabilities of occurrence, the simulation suggests that it would be safe to modify the Max_LTV parameter from 50% to 60%.

The assumption is that current borrowing demand grows 20% more while deposits remain the same — which would signal an increase in capital efficiency. The organic deposit interest that comes from borrowing interest increases from 6.25% to 10.907% (based on July 7th, 2021 5pm KST market data). This generates an extra 20.9 Million UST annual cashflow to Anchor, which supports Anchor’s longevity.

[Chinoman10]

Great analysis!
Considering the worst-case scenarios, this is a no-brainer.
Will this be implemented (if passed) right after the vote ends, or will it still take a few days after the vote has (hopefully) passed first?

Thanks!

[MannyMarc]

Max LTV is 50% now, but you can borrow up to 40% only. Will these 40% change if the proposal is accepted and how? It stays unclear regarding this point or i missed it.

[NightWatch]

This is also what I was wondering. The proposal states one can borrow 20% more, but will that now mean back to 50% LTV while liquidation occurs at 60%?

[josephsavage]

Anchor protocol sets the 40% borrow limit as relative to the 50% Max LTV.

With an update to the Max LTV, the borrow limit will automatically change the same amount. Theoretically this would actually increase borrowing by 25% (from 40% of collateral to 50% of collateral is a 25% increase in amount borrowed).

In practice, many borrowers (myself included) manage default risk by setting our borrow limits much lower, so the actual boost to amount borrowed would likely fall somewhere between 20-25%

[e-gons]

This seems like a sensible choice. However, I wonder if it is appropriate to run our worst-case analysis by assuming the worst case we have seen in the past. In the spirit of antifragility, I would be interested in seeing the analysis assuming a maximum price drop that is 2x the worst observed.

[Decrypto]

Hello, I love the idea and am going to vote in favour, though I have a question.
What is the furthest this theory/simulation can be pushed?
Is to increase the LTV by said amount in order to increase the max borrow amount as much as we can do for said results? If not, why not push further?

May be a speculative or silly question to some but I don’t understand the workings so well.

Thank you.

[Spaydh]

@e-gons I agree, but x2 might be a bit too much: unless I got something wrong, it’d correspond to a -52% drop in price within a single minute on Binance, to which the current LTV ratio would return a default anyway. I am in favor of adding some factor to the worst observed scenario, though.

@Decrypto Well, the current simulation already suggests a limit to the LTV ratio based on the trade-off between security and attractiveness: at 65% and beyond, in the worst case scenario (in which liquidations are extremely likely to occur en masse anyway), there is a risk of default for the protocol. That means that auctioning-off borrower’s collateral does not fully cover the outstanding debt, which thus becomes incumbent to the protocol and its users. Even if we consider that in such dramatic event, the yield reserve could exceptionally be used to soak up the losses, that’s still:
a) precisely what we should avoid if we want the protocol to be stable and reliable
b) an existential threat, as the multiplication of defaulting loans could result in the protocol going bankrupt.

So yeah, IMO the question becomes: “At 60% max_ltv, how far can the price drop in a single minute on Binance before a max_premium_rate liquidation returns a default?”

[e-gons]

@e-gons I agree, but x2 might be a bit too much: unless I got something wrong, it’d correspond to a -52% drop in price within a single minute on Binance, to which the current LTV ratio would return a default anyway. I am in favor of adding some factor to the worst observed scenario, though.

Fair point. I’ve seen stocks go down 50% in a minute but it is extraordinarily rare. I do reiterate my stance that as a general practice, risk management should be done assuming a scenario that is worse than the worst previously observed.

[hogsss]

yes! and ty for bringing it up:)

[puddles]

Anchor Protocol posted that this is now live, if so they need to change the LTV number on the borrow tab as it still states 50% not 60% or am I missing something here?

[inkymaze]

Volatility and slippage need to be taken into account. Tail events are what break protocols. This is how we do it. Gauntlet

[CryptoMike]

It’s not live now, proposal is still running

[cryptobb]

Anyone knows when this is going live?