Some thoughts:
Structural: Ultimately I think the CFG ethos is aligned around a bottom-up solution that incentivizes all economic parties generate value for one another over the long term. Cheap, quick capital for AOs. Commensurate rewards to underwriters (UWs) for loading pools with good assets. Transparency, protections, and returns for investors. The consistency of each actor in the supply chain feeds into the consistency of the others.
@Fabien Quality control for pools and avoiding adverse selection are obviously important. The potential for Tinlake pools to attract āunbankableā assets will be a concern to current and future investors. This is why we have underwriters who filter loans proposed by AOs. These UWs should be held accountable for their decisions. Thus, the suggestion to whitelist underwriters has been floated. While I think the intent is directionally correct, the implications do not align with the long-term efficacy of the project. Whitelisting on the front-end will give UWs with credentials or status preference. This seems a sort of reversion to the standards of traditional finance, where certain banks dictate terms of asset pools. Instead, Iād recommend that UWs be held accountable by pure incentive. An incentive filter is usually the simplest, least-invasive, and most effective way to align actors in the same direction. Crypto-economic protocols enable a new frontier of incentive mapping, so why not use the design space to its fullest extent? Tinlakeās UW fee design is a very robust way to ensure UWs are acting in the best interest of the protocol. The TIN issuance (UW fee) only upon repayment of a pool incentivizes the approval of good collateral into the pools. Further, the passive TIN holders subsidize active TIN holders, incentivizing a greater and broader set of eyes to scrutinize the assets, making the pools more robust.
An additional incentive for UWs to act in good-faith and underwrite accurately could be a reputation layer, which has been mentioned. Whitelisting on the front-end makes more sense when the feedback loop for underwriting performance is longer, say for 5-10 year mortgage pools. But the nature of the assets in these pools is different. A shorter cycle means more and quicker feedback to UW quality. This generates a dynamic history of performance for UWs. Why not use this āfreeā data to filter UWs. I think that appending a reputation tail to individual UWs promotes an open, diverse ecosystems of UWs while delivering quality control simultaneously. As UWs generate a track record of good performance, they will naturally attract delegation from more passive UW. In conclusion, the issuance mechanism provides an incentive for UWs to approve good collateral in the short-term and a reputation mechanism provides an incentive over the longer term: consistently good underwriting bolsters an UWs future earning potential. We might look to Chainlinkās node reputation system as a reference.
Two Questions: For UWs proposing assets for inclusion in pools, how is the minimum threshold of staked TIN relative to the asset value/loan amount determined? If this threshold is denominated by both (volume of TIN * value of TIN), how does a high TIN price vs. a low TIN price affect the efficacy of the UW process for new loans? Maybe it doesnāt matter, because incentives against including loan with LTVs that are too high are strong enough.
@Fabien The warehouse idea is an interesting one. If a warehouse enables more nimble deployment of capital, it solves a real-world problem for AOs. Part of why they come to the platform is speed relative to what they could access in TradFi. The key is setting the interest accrued by the warehouse at the optimal rate. The additional cost should not be so high that the higher cost of capital to AOs offsets any increase in efficiency. In a world of zero interest rates, I think the appeal of the centrifuge protocol for AOs lies more in relative simplicity/speed than in price, but I may be wrong. You guys are more intimate with these tradeoffs than I am.
@nicolasb I think your suggestion for additional guarantees/protections in the system is a good one for a few reasons. First, insurance against loss is a fundamental building block of any capital market. For a market to grow and mature, insurance-like mechanisms, be it put options, futures, or credit default swaps are an important component. Especially for the institutional variety, they will want to hedge any exposure to underlying asset pools. Second, building insurance-like protections into the protocol could generate additional revenue that could offset other costs, say the cost of a warehouse. If Centrifuge/Tinlake were to build an open API for asset data, convicted third-parties OR UWs could reference such data and write protection contracts against default of the underlying pools. Investors would buy such contracts and pay regular premiums for the protection. The majority of these premiums would go to the writer taking the risk of course, but some could be accrued back to the Tinlake platform for provision of the oracle. Further, this function could be outsourced to an existing decentralized insurance protocol like Etherisc, Arbol, or Nexus Mutual. This symbiotic relationship would bring strength to the protocol and new frontiers for awareness and ultimately liquidity.
I have more thoughts on how Tinlake providing oracle services could be compelling but thatās prob enough ranting for now