A logical extension to the
Changes that would come with this are:
- Funds for lending are limited by deposits, no more under-collateralisation
- When a borrower defaults, lenders lose principal
- Lenders earn most of the interest paid by borrowers (the network could still capture a spread as fee)
- The interest rate must float according to supply and demand, rather than be set by policy
This adds quite a bit of complexity to the program, and also introduces some other problems.
Funds at risk without lender diligence or control
Storage providers are not homogenous. The risk of default of different lenders will vary greatly. But network-provided loans make the same deal available to all providers.
With basic network-provided loans, no funds are at risk. The network isn’t lending actual capital, it’s granting a reduced pledge by the provider. When the provider fails, they are defaulting on a commitment to future service, not a repayment of capital. The network can extend the same deal to all SPs and tolerate failed commitments in the name of supporting network growth, without great loss.
With staking, lenders would lose actual capital, yet they had no ability to select which providers to lend to.
The heterogeneity of SP risk informs the
Monopoly lending market
A built-in lending/staking service would tend to form a monopoly, leaving no room for competitive lending platforms. Without providing room for innovation at the base layer, this could also suck the air out of higher level and derivative applications, stopping the DeFi party from ever getting going.
With basic network-provided loans, the interest rate is set by policy, as a function only of demand. We could set the rate quite high, both acknowledging the “risk” of lending to the least reliable provider, but also to artificially leave room for new lending markets to offer lower rates to more reliable borrowers. The network is the lender of last resort.
With staking, the rate must float with demand and supply of funds. With an initial monopoly, it will attract large supply. Because network-provided loans are available to all SPs, there would be no room for another lending platform to offer higher returns by selecting riskier providers and charging them more: the network-provided loans set a ceiling on the interest rate. Alternative markets could only offer a lower return for investors, in return for less risk.
The primary way by which alternative markets could compete would be that if marginal storage providers were so unreliable that the risk of losing principal while staking was quite significant, and so supply was limited. But this would not be a good look for the network.
A possible mitigation to this problem could be that the network charges a very large spread on the lend/borrow rates, so that an alternative market could offer lenders competitive returns even while borrowers pay less.
Securities law
The protocol offering a reduced pledge requirement is quite different from a monetary loan, burning funds is reasonably different from paying lenders an interest rate, and forgiving a commitment is quite different from taking a lenders funds on default. All of these things could present significant risk to PL if we implement them into the network.
Complexity of doing it right
The basic network-provided loans can be implemented well with relatively minor changes to the built-in miner actor.
With staking, a hacky version could be implemented directly into the power actor, but this wouldn’t look much like the money lego of DeFi. It would be limited in flexibility and composability. Doing it properly requires new contracts/actors for the fungible staked tokens, which means working out the standards for such contract interfaces etc. We will do this by the time the FVM launches, but if network-provided loans are intended as a short-term resolution, adding staking will either make it take much longer or add technical debt to the network built-ins.
Post FVM, I fully expect something that looks like staking to be built on top of non-fungible loans to selected miners – essentially yield aggregation. We probably don’t need to push that too hard.