@Nicola and @Alex North, 4 Feb 2022
Filecoin storage providers need to pledge collateral in FIL when onboarding sectors. Sourcing pledge collateral is an impediment to providers’ growth. Some providers would like to borrow FIL in order to fund the pledge for onboarding new sectors. Off-chain, custodian-mediated programs for this exist already, and the FVM will enable on-chain lending like
When providers earn block rewards, 75% of those rewards are also held as a kind of collateral, and vest to the provider over a period of 180 days.
We are proposing network loans: a way for the network to lend FIL to miners for pledge collateral. This is a relatively small protocol change that could be delivered to the network quickly without significantly impacting other ongoing work.
- Miners can borrow funds for pledge collateral from their own future income
- Including new/small miners with little up-front capital or reputation
- The network earns a return on loaned capital
- Economic impact only in the direction of locking up more circulating tokens for longer
- Retain consensus security, etc
The Filecoin network lends funds to providers for use as pledge collateral, and then captures their rewards to pay back the debt. The funds are lent from rewards the provider will earn in the future.
There are two types of network-provided loans: under-collateralised loans (for new/small providers) and collateralised loans (for established providers).
Under-collateralised network loan
- A miner can commit sectors with less than the required initial pledge lockup.
- The initial pledge is currently 10x more than the pledge required for consensus security, so we could allow as little as 10% actually pledged.
- The shortfall between the notional and actual pledge is a collateral debt that the miner owes to the network, tracked by the miner actor.
- While a miner has collateral debt, it incurs interest. The interest rate is dynamic and has two parts:
- a network interest rate, which is a function of the ratio of collateral debt to collateral posted for the network as a whole (more debt → higher interest)
- a miner risk premium, which is a function of the individual miner’s collateral debt to collateral posted
- As a miner earns block reward, all vested rewards are first redirected to:
- pay down fee debt [this is already the case today]
- pay down the interest accumulation, by burning the funds
- lock up as pledge collateral, until miner’s collateral debt is reduced to zero.
- A miner defaults on their loan if they cease operation (i.e. earning rewards) while in collateral debt. Their posted collateral will be burnt as sector termination fees.
- The “loaned” funds are not recovered by the network in this case. The has defaulted on a service obligation, rather than a financial one. The network loses the pledged service of maintaining the sectors.
This type of loan is ideal for new miners with little up-front capital or earned rewards. Interest rates would likely be higher than for collateralised loans.
Collateralised network loan
- A miner can use their earned-but-vesting rewards to pay pledge collateral for new sectors, essentially borrowing from their own future income.
- Logically equivalent to taking an under-collateralised loan, and then paying it off over time with the vesting funds. Simpler in operation to just move the funds directly.
- The loan is collateralised because we can secure the funds immediately.
- The miner pays (burns) an up-front interest charge when taking the “loan”.
- The rate may be the network interest rate, with miner risk premium = 0.
The idea of using vesting funds as pledge has been proposed before. This proposal casts it explicitly as a loan, and the network earns revenue from it.
Since the minimum sector commitment (180d) matches the vesting period, this results in funds being locked up strictly longer than if they were allowed to vest. If this condition didn’t hold, we would need to continue to track the vesting schedule of the borrowed funds.
See also the logical extension
- Helps with collateral bottleneck, but not sealing throughput
- Using vesting FIL is not as good as using free funds. May reduce buying/borrowing pressure from lenders. Doesn’t put idle tokens to work.
- If treating unearned rewards as collateral, might we as well treat vesting ICO tokens as collateral too? It’s more predictable as income.
- Runs some risk of undermining active lending programs, which would reduce locking up of free FIL. Changing the underlying protocol has an “unfair advantage”.
- We should set the network interest rate quite high to leave room for real lending programs
Original raw notes, obsolete
Type of loan
- Undercollateralized loans:
- Miners commit sectors with at least a minimum percentage of initial pledge
- Takes on a “pledge collateral debt”, pledge collateral they owe to the network
- As the miner earns, all (?) rewards are first paid to reduce collateral debt
- (optional) The interest rate of the loan is calculated based on how much (as a fraction) total pledge collateral the miner has put down on existing sectors (and some other network parameter). I.e. the ratio of collateral debt to collateral pledged. The rate is paid by the miner by burning it.
- Rate has two parts: network wide rate based on network wide debt ratio, and individual risk rate based on miner debt ratio. Must get more expensive to borrow as debt increases.
- What happens if a miner defaults?
- Only miner pledge faults: If miner puts 1 out of 10 FIL, then 1 FIL is burnt
- (this won’t happen) Network defaults: If miner puts 1 out of 10 FIL, then 10 FIL are burnt (where are the 9 FIL taken from?)
- Collateralized loans with future rewards
- Miners commit sectors using their future (vesting) rewards as pledge collateral
- Pays an interest charge up front for the loan (the rate is calculated based on the loan taken and some network parameter)
- Future rewards can’t be withdrawn before vesting
- Future rewards used for initial pledge can’t be used for faults (* check if this is safe when there are enough faults such that the miner goes in debt)
- Alex: ... until other funds are exhausted, and then miner goes in fee debt