Yes fully agree, in both cases market will find its equilibrium.
Even though I understand it is counter intuitive. It seems AAVE utility is a function of (LTV,SI) with either (50%,0) or (0,N) but there might be a universe in between which might optimise capital efficiency and share the risks between more AAVE holders.
N depends on the amount staked.
And while writing this message I realised actually we can virtually create our own AAVE utility function at the portfolio level as user.
If I have let’s say 100 AAVE and decide to stake 40% and to keep as collat 60%. I am building a sort of AAVE utility function (30%, 40% x N). It might be slightly different as a global one, but virtually close enough.
Then, in order to maximise the staked amount at any time (security) it could be in the UI/UX, similar to the stable/variable borrow rate button, could we have a stake / un-stake to make it smooth with some percentage amounts nicely displayed** instead of a vault on an other web interface as traditional DeFi.
**Why you may ask, because if there is a FYI token offering 1000% APY, I’d love to quickly jump on the tractor and come back to secure/stake as soon as the deal is done. (Two options: vesting few days stake to avoid this type of behaviour or making it cheap(optimise tx fee) and simple to avoid friction and to ensure maximum stake at all time).
Back on the original question and based on Stani’s answer:
Current simplified model 685 AAVE/day for SR.
100% 13M staked – 1,9% APY – 400M USD insurance
75% 9.5M staked – 2.6% APY – 300M USD insurance
50% 6.5M staked – 3,8% APY – 200M USD insurance
25% 3.5M staked – 7,1% APY – 100M USD insurance
10% 1,3M staked – 19% APY – 40M USD insurance
5% 650K staked – 38% APY – 20M USD insurance
Very cavalier statement but I believe in a crypto bull market 38% APY are easily achievable (even with 50% capital available), so staking might be less important than LTV. However, in a bear market this could attract capital the same way as staking rewards worked in 2018/2019.
After further thinking, I then believe more weight should be placed on the SI side than the LP one for the current distribution and really like the quarterly approach.
My two cents