ARC: Raise AMPL maximum interest rate

ARC Rationale

Following API-12, AMPL borrowing & lending went live on the AAVE v2 market 2021-07-24T20:22:00Z.

Since then, there has been a near 100% utilization rate of lent assets. This suggests the maximum cap of the interest rate curve is not able to reach a high enough value to effectively balance incentives between the borrow side and lending side of the marketplace.

While the AMPL spot market is currently in a relatively extreme condition, the Aave borrowing market should be able to perform efficiently in all market scenarios.

ARC in Short

We suggest the following new variable interest rate model:

  • Optimal utilization = 75%
  • Slope1 = 2%
  • Slope2 = 10,000%

This leads to a piecewise linear curve with two parts and three defining points:

  • Borrow Interest(0) = 0% APY
  • Borrow Interest(75) = 2% APY
  • Borrow Interest(100) = 10002 % APY

Since this will result in overall higher fees coming into the system, in tandem we also suggest lowering the reserve factor from 20% to 10% to share more of the revenue with suppliers. This would be submitted as a separate AIP to decouple these decisions.

We believe a nonlinear interest curve is healthiest long-term and could likely be used by many other assets as well, however this work can be discussed more in the future.

25 Likes

fully support this and very much needed. Currently the cap on rates does not do justice for the lenders. We need to find a happy median to incentivize both parties. Having said that, the INSANE demand for ampl is really great to see. Thanks for posting this @brandon

4 Likes

Hi Brandon,

I’ve been following this AMPL market closely because of the strange way it is acting. It seems like the economics of this lending market is completely broken because of the AMPL team’s custom implementation of the aToken and corresponding debt token. The implementation does not rebase the debt token, leaving a very enticing opportunity for arbitrage against lenders that deposit AMPL into AAVE. A borrower can simply borrow AMPL, watch their balance grow in a positive rebase, and pay back the unrebased debt pocketing the difference. Note that the borrower does not even have to sell the AMPL to benefit, so it’s not even a short position. The borrower profits simply from the fact that the borrowed asset is actually accounted for as a different asset than the loaned one. As long as the borrower repays before a negative rebase, they are guaranteed a very substantial profit at the expense of lenders.

You suggest 10002% APY as the maximum interest rate. This is equivalent to approximately 465% APR, or about 1.27% per day. The max rebase is ~5%, and so the corresponding interest rate that would compensate for the lenders loss is a 5% daily APY, which is completely unrealistic to implement. This is not to mention the fact that AMPL will simply not be borrowed before negative rebases by rational actors. In order for the lending market to function, the borrowed portion of the asset MUST be treated the same as the unborrowed portion. Otherwise, the math simply doesn’t work out. Nobody will be able to withdraw their AMPL in this current situation until the rebase is predicted to be negative.

In my opinion, the AMPL market should be immediately disabled on the front end and the aToken and debt token implementation revised. In this current situation, arbitrageurs are profiting at unsuspecting lenders expense who believe that they are earning a safe yield. They are not. It is currently impossible for lenders and borrowers to have a mutually beneficial relationship.

6 Likes

It’s certainly not broken. The rates just need to be tweaked a bit. Actually everything is working quite well. You also may be looking at it in the perspective of someone looking to maximize their percentage of the AMPL network. Whereas new actors may only be in search for a solid steady yield irrespective of which way the market is going. I would say let’s make these modifications and see how it plays out.

4 Likes

Dear @pakim249,

Wanted to add some color - maybe helpful for the conversation.
AMPL’s rebase rate and APY actually represent very different things, in my opinion - they should not mirror one another or be otherwise entangled.

Why AAVE will be accelerated by AMPL:
AMPL borrows as a stable unit asset. Thus, the borrower always knows they have a fixed unit of debt + interest. Since AMPL the asset has proven to be mean reverting in price - the borrower can choose to take on the market time ‘risk’ of waiting for AMPL to return to a price target that is favorable for them.

The lender - on the other hand - is not seeking to get exactly the same yield as if they hold AMPL.
Rather, they seek to lock in the number of units of AMPL they hold on some of their holdings.

So - since AMPL has no “junior” and “senior” tranches - like some assets that seek price stability - all wallet holders are both the junior and senior tranche. AAVE integration allows these risks to be split:

i) depositors of AMPL maybe keep the numbers of AMPL they have constant, if those AMPL have been lent out.
ii) borrowers of AMPL buy the exposure to changes in supply (by paying interest)

The key unlock for the AAVE platform: in periods when traditional “stables” are in demand (generally market risk-on times), borrowing a stable unit of AMPL may have a lower cost. Then, the AAVE borrower may sell the borrowed AMPL, use proceeds to buy a risk asset of their choice - and simply take the time to wait for AMPL to be at a price (in the mean reversion) that is favorable for them to close their position.

No risk is created or destroyed - rather it is transformed. And here - it becomes an incredibly powerful asset for both lenders and borrowers on the AAVE platform.

Hopefully this is of help // please let me know if I can improve how I think about these matters.
Also: if anyone finds any of this mind bending - I do as well. Elastic finance is going to be super interesting to build and understand.

And of course worth noting: I hold positions in both AAVE and AMPL.

Warmly
Noah

12 Likes

As a lender, I am quitely agree with the ARC, because when the rebase rate is great larger than the interest rate, lender can withdraw the deposite, to push the uitilization towards 100%, and the interest rate will be very high for lender. However the defining Borrow Interest(100) should be mapped with the rebase interest rate to achieve a better game theory between lenders and borrowers

4 Likes

What do you mean by the max rebase being 5%?

Also rational actors can profit from borrowing when there’s an upcoming negative rebase. If they where to borrow at the beginning of a period of negative rebases e.g 95c they then could borrow, immediately sell, repurchase at a later date and pay it back at lower price as you would with a regular short position.

2 Likes

As an AMPL holder I think the AAVE integration creates many opportunities for AAVE as platform itself as well as AMPL holders. I do think that the only reason it hasn’t exploded with more usage is because of the APY cap. Having something closer to Brandon’s solution would provide more utility for both the platform and the holders and users of AMPL imo

4 Likes

I want to throw in that maybe there should be a special info button about AMPL and it’s APY on the Dashboard or a special warning for the deposit as new and unsuspecting users are currently completely getting played by people who understand how to game the rebase system against them and they might only look at the APY and not understand why it’s so high. Just to prevent people from calling AAVE a scam platform because they got drawn in by high APY. I think it’s great to have coins like these on the platform as it gives Aave a lot more utility but
it should really be considered that most users, especially those new to crypto don’t know what a rebase Token is and get burned heavily by this

7 Likes

Hi @brandon

Thanks for your quick reaction

AMPL had indeed a very specific behavior due to the rebasing which affects borrowers but not suppliers, if the liquidity is borrowed. Borrowers can then game Aave Liquidity Pool given rebasing is more profitable then the borrowing costs; leaving suppliers frustrated as they can’t withdraw their liquidity

It’s key to protect liquidity, to do so the model needs to take into account other market opportunities such as rebasing. I can see the maximum daily rebase is recent, standing at 7% daily
https://www.coin-tools.com/ampl/ampl-rebase-history/

Surprisingly, this is a lot more than the 10,002% APY suggested above as mentioned by @pakim249. Why not make the slope 2 parameters even more aggressive to align max rate and max rebase?

The current liquidity issues are a testimony of the high risk of AMPL which had not been quantified accurately in the initial interest rate parameter calibration. At the moment it’s hard to fully grasp the risks of AMPL for Aave - from my perspective, it is one of the riskiest assets thankfully it cannot be used as collateral. For this reason, I am strongly against reducing the reserve factor which is calibrated based on the risk the asset brings to the Aave DAO

12 Likes

I agree! Something must change and the interest rate should somehow compensate the rebase rate otherwise I don’t see how this should work in the long run.

2 Likes

I think 10,000% APY is still low, that’s not even 1.3% a day. If you see the data from rebases, most positive rebases are way higher than that. Every time AMPL has positive rebase above 1.3% there will be 100% borrow in AMPL pool. Maybe 100,000% APY (around 1.9% per day) or even 1,000,000% APY (around 2.6% per day) should be better for experimentation now, otherwise AMPL is still going to have 100% borrowed when positive rebases happen.

In the medium/long term I would like to see different slopes, depending if AMPL is in a positive, neutral or negative rebase.

8 Likes

Interesting comments here. I’ve studied this a bit and it seems like the arbitrage opportunity described by @pakim249 makes it so the interest rate for AMPL will always be naturally super high in the current implementation. Is there a way to counter this long term @brandon ?

On the short term I would support increasing the max interest rate to a very high amount like @Alex_BertoG said to make it fairer for the depositors.

But long-term we need organic borrowers and no one will borrow organically if the rate is at such a high rate. What do we do here?

4 Likes

AAVE has an acute need for users who borrow assets. Borrowers are the ones that incur the interest that is paid to depositors (lenders) in AAVE pools. AMPL has strong borrowing demand and it appears the goal of this ARC is to make the depositor and borrower markets more efficient to better serve AAVE users. I support this ARC. I will now outline some of the factors that allowed me to reach this conclusion.

Borrowers drive the platform because many users are willing to lend any asset type. It is a bit more difficult to facilitate demand for borrowing across all asset types. Some assets have strong borrowing demand and some assets do not have strong borrowing demand. This is reflected in pool utilization rates. Some assets have extremely little activity from borrowers and these are typically volatile assets which are easy to identify when looking at the AAVE markets. Many lenders wish to earn interest on these assets but few have a compelling reason to borrow them and as a result APY’s are low and utilization rates are low. Assets with predictable debt obligations such as USDC, DAI and USDT are excellent for the AAVE platform because they generate healthy interest for AAVE users due to solid borrowing demand. These types of assets offer no upside or downside to borrowers but what they do offer to borrowers is a static and predictable debt obligation and that allows AAVE users to borrow those types of assets with confidence.

AMPL is a unique asset in that it offers AAVE users both the complexity of an asset that can offer upside potential while at the same time offering predictable debt obligation to borrowers. All assets that have a predictable debt obligation have grown to very large pool sizes. For example USDC is over 5.5 billion in market size and assets with predictable debt obligation have very healthy demand to borrow. Assets with predictable debt obligations grow large pool sizes and have healthy utilization rates which is ideal for AAVE and it’s users.

Since AMPL has strong borrowing demand the question becomes how do the parameters need to be adjusted? It is abundantly clear based on the consistent 100% utilization rate that borrow interest rates need to be increased dramatically. To what precise limit is very difficult to answer. While it seems that a limit higher than 10002 % APY could make some logical sense, I support the current ARC as described because the most important factor to preserve in any AAVE pool is the demand to borrow which is impacted by the interest borrowers will owe. The ARC as proposed should strike a better balance between APY for depositors and upside potential for borrowers but not cut too deeply into the upside potential of borrowers to harm that side of the market.

What will likely happen as a result of this ARC as proposed: An increase in APY for AMPL lenders will bring in more lenders as the higher APY will be compelling to a larger group of AAVE users. Borrower demand for AMPL will remain strong as AMPL has a predictable debt obligation yet upside for borrowers will dampen as the utilization rate approaches 100% and the 10002 APY limit.

I could see there being adjustments and refinements in these parameters in the future because it is difficult to perfect the limit at this time. It is clear that the ARC as proposed is moving in the direction which is correct, but towards a precision that is difficult to know. What will happen with this ARC is that the AMPL market will come into a better balance which will allow the pool to grow generating a better outcome for pool users.

9 Likes

@pakim249 The arbitrage you describe is not a bug, it’s a feature.

Since the loaned amount does not rebase, lenders transfer their exposure to supply-side risk on to the borrower, this is what allows ampl to safely denominate debt.

To reach an equilibrium, the cost of borrowing ampl during positive rebase needs to be the same as the rebase, arbitrage traders will continue to borrow ampl until this is the case. As it stands, the apy cap prevents market from reaching this equilibrium, which is why the utilization rate is 100% currently.

While I support this proposal, i’m inclined to agree with @DFC, I think apy cap should eventually be far higher, as if ampl were to see a sudden increase in demand, it could easily reach 100% utilization again even with 10,000% apy.

9 Likes

I think this ARC should be implented to AIP as soon as possible to protect the stake of lenders, no need to discuss futher details, as least the current ARC can work.

7 Likes

Agreed, the current situation is making it difficult for lenders (cannot remove their ampl because of constant 100% usage rate) and also doubly because the incentive to put any more ampl into the pool is low because of the capped APY

3 Likes

I support the proposed ARC. It’s a good starting point and better than what we have now even if it does not end up being the final solution.

4 Likes

I support this ARC. Others point out that 10,000% still doesnt cover larger rebases and I agree, and I think ampl might need custom fee structures that are elastic and scale up and down in the future but this looks like it can really showcase what is possible and also bring great benefits and borrowing volume for Aave.

Would like to raise one point - you state that currently the reserve factor is 20% (hence 38% to lenders but 48% fee to borrow), and aim to lower it to only 10% - however at 10,000% this is 1000% going out to the reserve - is this way too much , does this take away too much value away from the lender? Because as pointed out the 10,000% only scales to cover rebase to around 2.5% (but possibly even less with 10% taken out? Because the apy is so massive couldnt/shouldnt the reserve just be 1% , or 0.5% or less even? Or again this makes me think that ampl might need custom and elastic fees in the future.

Also would like to add a +1 to seeing this go to vote as soon as possible as the issue of the current lenders only earning 1/100th of the rate of ampl holders during the 2 larger rebase days, and the fact that they are locked in because the utilization is at 100% so cant withdraw (which is a bigger issue).

But amazing to see in this first trial and outstanding to see 100% demand. Congrats to both aave and ampl teams, this is pretty historic.

6 Likes

I’d like to point out that using APY in this discussion is extremely misleading. 10000% APY is about 465% APR. If 10% of this were to go to reserve, this would be about 46.5% APR or about 51% APY, which is pretty far from 1000%.

Also, raising the interest rate cap is not sufficient justification to lower reserve factor in my opinion, as it remains to be seen whether this cap is actually realized. Also, the ecosystem reserve collects aAMPL, not AMPL. This means that the ecosystem reserve would be subject to the same arbitrage that lenders are subject to and have the same liquidity problems.

3 Likes