Centrifuge and Tinlake are very interesting platforms and I’ve eagerly followed the developments in the last couple of months.
During that time I found that some things regarding Tinlake’s unique value proposition are still unclear to me.
Everything that follows comes from a good place and I hope that jointly answering these questions can be part of the success of these brilliant initiatives.
Challenge #1: How is the risk assessment quality of investment opportunities in Tinlake ensured?
Basically Tinlake allows for a digitalized variant of trading CDOs [=> Please correct me if you see this differently].
This is something that exists already in the traditional finance world and was, among other things, a trigger for the 2008 financial crisis due to opaque and poorly understood risks.
To attract a large number of investors, Tinlake must offer very solid, high quality risk assessments better than existing companies do - how can that be achieved?
Challenge #2: What unique benefits do investors and asset originators get on Tinlake compared against traditional collateralized debt investment solutions?
It is clear to me that a crypto-based gateway to real world assets will generate value at some point due to the automatable connection of smart contracts & DeFi applications - i.e. Centrifuge is a necessary addition to the crypto space imo.
However, I would like to have some more specific examples what these benefits could be for the Tinlake use case.
Here’s what I can imagine:
automated risk management through pooling into different risk classes + automated insurance acquisition
better transparency (& auditability) of asset originator track record
Hey eddie, no I suppose there are many more challenges
However, I for one am focused on the ones which are specific to Tinlake and not the common ones for any new (crypto) project.
Please feel free to add the ones you think deserve special attention - this is what this thread is for!
Challenge #1: You are right, DROP tokens have similarities to CDOs. I think such securitization structures are very important financial instruments to provide liquidity on scale. These structures are not the reason for the financial crisis of 2008. They were the tool that was used but not the trigger. The trigger was as usual too greedy bankers and fund managers, a financial system out of control, and very intransparent securitization structures. Junk investments got a triple-A rating because even the rating firms were only able to check the securitization structure but not the single assets. Tinlake helps here. It makes the securitization transparent. Every single asset in Tinlake is represented by an NFT. Every investor can check asset price (risk score) and performance in a daily updated NAV, which is completely novel. The Asset Originator providing the assets has to have skin in the came with investing in TIN (the junior tranch) and has to take first losses. Tinlake shows fully transparent on-chain what happens today hidden in a Bank in CeFi. You do not need to trust the bank anymore.
Additionally, we work on fully decentralized underwriting, which would further minimize the trust in the Asset Originator as the remaining “central” piece.
Challenge #2: I think Asset Originator and Investors enjoy the usual benefits of DeFi: trustless infrastructure, works 24/7, interest per second, immediate borrowing, investment, value transfer.
There will be more soon. The power of composability and interaction with other DeFi protocols will make investing easier and more secure. You will not need to invest in a single Tinlake pool anymore but invest in another DeFi protocol, which will manage your investments, leverage, and may insure it as it is already possible for crypto assets. Real-world assets represent a unique opportunity for crypto investors to diversify their portfolio to make it less dependent on BTC and the rest of the crypto market (which is usually not much different). I also think long-term DeFi will be both cheaper for borrowers and offering better returns for investors. The margin banks are making today will not completely be assimilated by decentralized service providers (Asset Originators, Underwriters, Issuers, …). It will be an order of magnitude more efficient and will in addition reward all participants if they contribute to this decentralized credit system.
I hope that makes sense for you. Keen to get your opinion. -Martin
Hey @martin,
thank you so much for your extensive answer and also for the link to the decentralized underwriting post - very intersting stuff!
Let me see, if I get this right:
Challenge #1 / risk assessment:
If I understand your points correctly, one important measure to ensure a good quality risk assessment is for asset originators to provide TIN upfront to take first loss, as well as enforcing a fine-tuned TIN/DROP ratio per asset pool. That is certainly a valuable mechanism to incentivize special scrutiny for the (asset originator’s) risk assessment. However, that mechanism alone could still be systematically gamed by an asset originator, which is why you have included a whitelisting mechanism for asset originators. That shows that an investor still needs to put trust in the asset originator’s competence in terms of risk assessment. Here, additional incentives (e.g. analytics dashboards on Tinlake DApp) could help by analyzing asset originator track record and thus driving more investors to
asset originators which have proven to be very precise with risk assessment in a given context.
Your second point is that the securitization structures are now openly visible and single assets can be examined more in-depth. However, what does that mean in detail and what data will (realistically) be published onto the blockchain? Surely any intimate data of the risk assessment (solvency, cash flow, etc.) will fall prey to privacy issues and not be public? Therefore, how can we - for the single assets - generate risk scores that are trustworthy?
Challenge #2 / Tinlake UVP vs. traditional solutions
Thanks - I completely get the composability argument and agree that it must be more efficient long term.
Yes, your understanding is correct and your conclusion too. Asset Originators remain currently the weakest and still centralized point in the process. TIN is protecting it and blockchain makes it harder to game it though. The underwriter token can solve it. It is distributing the trust in the AO to all underwriters. They are voting and staking on assets. Incentives are an additional token reward from staking, penalties are that their stakings get burned. Underwriters can be a decentralized investment adviser. You will see their past performance, area of expertise, and reputation on-chain and can just follow them as a DROP investor or DeFi composable. Important is that underwriters are mitigating the conflict between Asset Originators/ Borrowers (borrow cheapest) and Investors (earn the highest yields). They have skin in the game and can earn extra money if their assets perform. They need to balance risk and return.
It is more than “openly visible”. The infrastructure is permissionless, decentralized, and censorship resistant. Examined means, every single assets performance is on chain. Just check the asset tab in a Tinlake pool. Even if you only know that it is an invoice or a bridge loan (asset type), you know the terms and performance. That is already a huge advance if you compare it with existing CDO structures. TIN token holder can ask for all asset details. The NFT is notarizing the asset details in a precise-proof. TIN token holder can even operate their own Centrifuge node, can be added by default as a collaborator to the NFT, and can access the assets details through the Centrifuge P2P network in real-time. That is where the underwriting chips in. The underwriter can create their own risk models, AI driven or whatever, fully automated. Makes sense? The infrastructure is already there. We «just» need to grow it as a community!
One other huge benefit for both investors and even more so for AOs is the decoupling from central banks, mainly in a sense of interest rate risk. In theory, DeFi should not be effected by Fed and other central bank rate hikes (which are coming) and cost of capital should not change when rates increase in the legacy financial systems.
hello, newbie here. I recently started looking at the project and it seems exciting. My initial questions are very similar to OP, perhaps someone could answer:
1. On risk quality and NAV calculation
I’m keen to better understand NAV calculation mechanics & assumptions. As per https://developer.centrifuge.io/learn/pool-valuation/ underlying collateral will be priced using DCF approach. In a nutshell, DCF approach requires two key ingredients - cashflow schedule and discount rates.
a. Please correct if wrong - I guess cashflow schedule is estimated using term sheet (document) which has been published on-chain?
b. Who, how and where determines “appropriate” discount [risky] rate for the collateral assets?
c. How NAV pricing at origination step differs from ongoing NAV pricing during the life of the pool?
d. How are cash proceeds handled - are they sent pass-through as interest / principal payments to senior / junior tranches of the pool (respecting waterfall)? Or there could be some cash reinvestement options
e. Are borrowers allowed prepay early? If so, is prepayment risk taken into account for NAV calculations?
f. What is the procedure to determine delinquent (non-performing) assets? How does the recovery mechanism work - timing horizon; how this information will be communicated to the pool / investors?
g. And finally, since underlying collateral are real world assets - are all of the above calculations audited?
2. On benefits for borrowers
What are the benefits for borrowers to draw liquidity from Centrifuge bridge rather than going to traditional commercial banks? Is there a risk of adverse selection (i.e. do borrowers go DeFi because they could not secure a better deal at commercial bank)?
trying to find answers…
1a) Cash flow is generated, and on-chain reported/calculated for every asset per NFT representing the asset.
1b) There the average discount rate for the pool is an estimate. The discount is set per NFT/asset. Every single asset has its terms with principal, maturity, and discount; and it can be different per NFT. Maturity at least actually is different for almost every NFT.
1c) The NAV is continuing on the real performance of the financed NFTs, which can be tracked on Tinlake btw.
1d) All payments for an NFT (interest and/or repayments) will go into the NAV and drive the price of the DROP (fixed APR) and TIN (remainder) token of this pool. The DROP and TIN price is reflecting the current NAV. Investors can redeem at any time and then the cash reserve and next borrower payments are used to satisfy it.
1e) yes. There is no pre-payment risk or late-payment risk as long as the NFT is paid back in full (the interest is APY compounding per second), meaning no default.
1f) You can check it in Tinlake. Those NFTs will be automatically written off following a write off schedule which is usual for this asset class and impact the NAV (starting with TIN only of course)
1g) That smart contract code is open source, the code audits as well. You can find it all on github. Tinlake makes it fully transparent and every user can audit the smart contracts doing all of the above.
2) I guess code is long-term more efficient, faster and by order of magnitude cheaper than banks. Especially if it is not closed (per bank or fintech or network) but a public decentralized protocol.
I hope that helps Ty!
thank you! few follow ups:
1b) “The discount is set per NFT/asset” - yes, exactly this is what i’m keen to learn. who sets this discount and how is it estimated. in traditional finance borrower’s discount rate is a function of [risk-free] interest rate and borrower spread (aimed to compensate for credit risk). I’m very keen to learn how borrower credit quality is evaluated.
1e) sorry, what i originally meant by prepayment risk is whether borrowers have the right (option) to pay back their debt early. This will directly impact the NAV calculation of the asset - shorter life of the instrument means less interest cashflows collected. Kind of similar to prepayment in MBS, CLO… effectively investor is selling the option on interest rates to the borrower.
1g) sorry, i meant financial audit of real assets. Nevertheless - thanks for pointer to code base - i’ll take a look
2) agreed, in theory centrifuge approach should be way more efficient & transparent than traditional banks, imho subject to credit risk of financial assets is evaluated appropriately (love it or hate it but that’s partially the reason why banks charge higher rates - they do have tools to estimate credit risk of real world).
side note, centrifuge project reminds me of Intex (https://www.intex.com/main/solutions.php) which is a top product for securitization analytics. If centrifuge could achieve similar moat, and on top of that handle trustee responsibilities on chain - then it’ll moon. good luck!
yeah. let’s take it forward. many thanks for your interest.
1b) “The discount is set per NFT/asset” - yes, exactly this is what i’m keen to learn. who sets this discount and how is it estimated. in traditional finance borrower’s discount rate is a function of [risk-free] interest rate and borrower spread (aimed to compensate for credit risk). I’m very keen to learn how borrower credit quality is evaluated.
that is the protocol. depending on the ts&cs setup in the pool the incoming NFT is rejected or accepted and priced if accepted using predefined score cards. underwriters - typically junior investors - set the stage here. the coming underwriter token will improve it further.
1e) sorry, what i originally meant by prepayment risk is whether borrowers have the right (option) to pay back their debt early. This will directly impact the NAV calculation of the asset - shorter life of the instrument means less interest cashflows collected. Kind of similar to prepayment in MBS, CLO… effectively investor is selling the option on interest rates to the borrower.
you are right if you compare it with existing static debt instruments. all these pools are revolving and there it does not really matter. the money is just reinvested.
side note, centrifuge project reminds me of Intex (https://www.intex.com/main/solutions.php ) which is a top product for securitization analytics. If centrifuge could achieve similar moat, and on top of that handle trustee responsibilities on chain - then it’ll moon.
thanks for sharing. interesting! sorry, that I disagree here. I think you look at it with the web2 eyes. web3 does not need it. I would even more say that web3 trying to establish the know web2 moats will fail. you either be centralized or decentralized. a mix does not work. …also the reason why DeFi will eat CeFi
Ty!