How to Reassemble the Existing Financial Infrastructure
A token model for underwriters to minimize trust in real-world assets
Centrifuge is a protocol for decentralized asset financing which lets businesses finance their real-world assets on-chain. We’re developing a new system of credit, one that is open, resilient to single points of failure, and with equal opportunities for borrowers and investors.
Centrifuge creates the first alternative to banks and the existing credit system.
We present our design proposal for a system to incentivize the valuation of real-world assets and the assessment of risk through a decentralized system of underwriters in order to minimize trust in real-world assets.
This design builds on Centrifuge Tinlake’s two tranche token structure. TIN–the volatile first-loss high yield junior tranche and DROP–the fixed interest lower risk senior tranche. Read more about TIN and DROP here.
Inefficiencies in Traditional Finance
Today, underwriting is mostly done by banks, with little incentive to be efficient or fair. They often are both the underwriter and the investor, and because of that: they don’t need to be competitive. Banks only perform this function when there is enough of an incentive (profit) for them. These inefficiencies leave many businesses unserved and result in exorbitant fees for the SMEs they do cover.
Our proposed underwriter token model, using the TIN token, is designed to provide the service of valuation and risk assessment of assets to investors through an open and distributed network of underwriters. The active participation of underwriters minimizes the trust in these assets and results in greater efficiency, higher quality of assets, and reduced risk to the system. Investors will find more transparency as a result; with clearer definitions of the value of assets and the estimated risk vetted by several parties.
Centrifuge’s decentralized underwriters enable businesses to access capital faster and cheaper without relying on a bank.
Risk & Asset Valuation by Underwriters
There is a conflict of interest between asset originators that bring loans into a pool and the investors that fund them. Asset originators are looking for a lower cost of capital, while investors are seeking to maximize their risk adjusted return. In the traditional financial world a bank would often underwrite the loan (decide on the terms) and be the investor at the same time.
Underwriters are actors with the expertise needed to provide the value and risk for proposed loans. They are responsible for deciding which assets should be financed and under what parameters of valuation and risk, within the boundaries set by the pool. Their task is to ensure the stability of a pool, and their return is a share of the interest and tied to asset performance.
Because we’re building in an open permissionless system, our underwriters can look very different from today’s. They can opt to collaborate with others in the system to complement each other: an expert in fraud scoring could join forces with a large credit data provider to propose assets that qualify in both categories. Underwriters can also try to compete and use proprietary models for their own advantage.
TIN: The Token for Decentralized Asset Underwriting
Objectives
The objectives of the Centrifuge Underwriter token are the following:
- Remove a single point of failure within the system by allowing underwriters to compete but also collaborate
- Incentivize the valuation of assets in Tinlake pools
- Incentivize distributed and broader risk assessment to provide more accuracy
- Create a gatekeeper to assets entering the pool
- Give underwriters skin in the game and require them to take the first loss
Mechanics
With this proposed mechanism, TIN becomes a governance token for underwriting Tinlake pools that controls which loans to accept into the pool. TIN holders vote on pricing and risk assessment by staking towards potential assets. For performing this service, these TIN holders earn underwriting fees.
To participate in a pool as an underwriter, the first step is to deposit DAI (or another stablecoin) into a pool to provide part of the capital needed to originate new loans. The pool will mint TIN in exchange for this deposit. The underwriter can then start staking the TIN to potential assets proposed by the asset originators.
Anyone can become an underwriter by buying and staking TIN tokens. If an underwriter has a strategy that performs better for a pool, they can capitalize on this opportunity. These market dynamics lead to diversity and competition among underwriters. The resulting decentralization is more efficient and safer for both the asset originators and investors.
Economics
As assets are financed, DROP and TIN tokens increase in value according to the Net Asset Value (NAV) of the pool. Asset originators repay loans with interest in DAI into the pool’s reserve and DROP and TIN holders can redeem their tokens for DAI in the reserve.
The Yield Waterfall: Yield is distributed to the different tranches and awards are given to staked underwriters.
Asset Underwriting Mechanism
DAI in the pool’s reserve from investors and loan repayments can be used to finance new assets. In order for an asset to get financed by the pool, first an asset originator proposes the asset with a minimum deposit of TIN. This minimum deposit serves to prevent spam of proposals. In the proposal, the asset originator includes a proposed value, loan amount, and risk score for the asset. Any proposer must own a minimum share of TIN of the asset value being proposed to ensure there is enough at stake for the proposer and align their incentives with those of other TIN holders.
Underwriters stake TIN tokens on proposed assets to be financed, and assets above the minimum threshold, with largest stake, are then financed by the pool according to that risk class. The minimum threshold is set relative to the value of the asset. If a proposal does not meet the minimum threshold of stake, the proposer loses their minimum deposit. However, they can propose the asset with different value, amount, and risk parameters.
With this mechanism, underwriters decide under which terms assets are financed within the boundaries set by the pool. Staked TIN is locked until an asset is repaid. Stake on defaulted assets is slashed based on a percentage of the lost value.
The asset underwriting mechanism is designed such that if no loans get enough stake, bad quality loans do not get financed. There is a minimum threshold of TIN at stake relative to the loan amount of the asset to be financed, and the value of TIN. If there are more assets that qualify than liquidity to finance them, only those with the largest stake get financed in that time period.
Encouraging Active Participation
In order to incentivize active participation of underwriters, on loan repayment new TIN tokens are minted. The newly minted TIN are distributed only to those underwriters that staked towards the asset. As a result, not participating in this asset underwriting process leads to a dilution of your TIN–effectively the fee for underwriting services.
We foresee people to build delegation models on top of this mechanism to allow TIN investors to delegate tokens to underwriters for better asset curation–sharing the upside with underwriters that don’t have enough capital of their own avoiding their dilution due to non-participation.
A Better Way of Credit
The above token mechanism incentivizes competitive pricing for real-world assets and reduces the reliance on asset originators and off-chain data. As a result, investors find more transparency; with clearer definitions of the value of assets and the estimated risk assessed by several parties. Asset originators, investors, and underwriters are systematically aligned to create more liquid and stable pools–a better way of credit.
DROP via Swap on balancer or uniswap or Tinlake … to investors using DAI. We can also imagine the Warehouse providing liquidity on private pool to create secondary market.
Collateral - underwrote assets need to be somehow “premium” otherwise we will only attract high yield under rated debt and increase DROP