Subscribe to 0xucs.eth
<100 subscribers
Share Dialog
Share Dialog
What is the biggest narrative in crypto for 2023? Right, it’s all about Real-World Assets (beyond stablecoins) - tokenized treasury bills and equity, yield from private credits, markets for carbon emissions, fractionalized real estate, and so on. There is no doubt that “that ‘real-world assets’ will eventually simply be ‘assets’”, but since we’re still in the early days for the sake of clarity, I’ll allow myself to call it RWA.
For those who are not familiar with the topic, you can go through these papers (they’re all pretty similar tbh): Overview of on-chain RWAs and the forces propelling their growth from Galaxy, State of the (RWA) industry report from FortunaFi, RWA report from Redstone and Chaos Labs. Or check the list of resources from RWA.zyx.
From the early days of Bitcoin, we saw the desire of crypto-natives to separate themselves and their permissionless habitat (what we call today “web3”) from authoritarian legacy financial systems. The pre-RWA era of web3 had a pretty straightforward narrative, which can be described as “don’t trust people - trust code” or “code is law”. In other words, the ultimate goal for many protocols was to deliver a product (in most cases - a financial product) with trust assumptions minimized as much as possible. But times are changing. Today, both systems have come to the point where they cannot ignore each other and, at the same time, don’t understand how to treat each other:
crypto has proven that the code actually works even in complex financial applications, but the closed self-centered ecosystem is becoming too small - crypto is staking its claim in new fields;
governments publicly accepted that crypto is much more than a gambling crap for a bunch of marginals, but they are still not ready to integrate it into their systems.
Still think that governments do not understand how powerful crypto is? Check this paper from the Federal Reserve - their researchers fear that tokenized assets could transmit volatility from crypto asset markets to the markets for the crypto token's reference assets. Sounds pretty powerful.
Here we are, at this juncture. Intuition tells us that these systems can no longer exist in isolation. Their gradual convergence seems more like a natural evolutionary process than the whims of random actors. However, this convergence cannot occur while fully preserving web3 values, particularly the “code-is-law” dogma. The emergence of Real-World Assets made this evident. If we want web3 to “adopt” a portion of the “Real-World”, some compromises are necessary (to be fair, governments and legacy financial systems are also making concessions concerning crypto). In this piece, I aim to delve into these compromises within the context of onchain private credit markets and address the ensuing questions:
what are private credit protocols all about - do we observe a revolution in private credit space or just minor adjustments to the existing system?
who do we trust when using these protocols - their code, their team, national governments?
is it even possible to keep original web3-values untouched when we deal with private credits in the real world?
We won’t dive deep into specific details like legal frameworks, methods of NAV calculation, etc. (unless it’s necessary to understand the protocol in general), but instead will try to provide a high-level understanding of existing approaches to onchain private credit markets. We’ll discuss Centrifuge, Goldfinch, Maple, Clearpool, and Atlendis. CrediX, TrueFi, Ribbon Lend, and dAMM will be out of the scope since they replicate the mechanics of the mentioned protocols and/or have minimal adoption.
tldr: 1. cause this is the area with so many inefficiencies and, hence, so many opportunities - for both lenders/investors and borrowers. 2. no, we’re not revolutionizing it - in most cases, we’re just bringing existing fintech solutions onchain
Private credit markets in web3 started to get attention in 2020. What goals were they stated? I’ll let protocols speak for themselves:
to organize trillions of dollars of global debt when all private debt moves onchain and to offer an incredibly safe, diversified, reliable yield for stablecoins (Goldfinch);
to connect institutional investors globally with credit opportunities, unlocking risk-adjusted returns through robust underwriting, building on the most advanced decentralized technologies to achieve scale (CrediX);
to reduce the reliance on the intermediaries involved in financing real-world assets (lead manager, managers, lawyers, etc.) and create more open, transparent, and efficient access to finance (Centrifuge).
Wording can vary, but I think you grasped the idea - as always in crypto, we’re empowering financial inclusion, eliminating (hated by everyone) intermediaries, and creating a more open and transparent environment. But what precisely are these protocols doing? How can one formulate the essence of the Goldfinch/Centrifuge/YouNameIt protocol? To understand it, let’s briefly examine how lending in the real world works and what Nigerian startups would do if Goldfinch didn’t exist in this world.
“Kings of lending” in the real world are banks, without any doubt. They originate loans, they collect payments, they manage defaults, and so on. And yes, banks operate under strict regulations, which mandate that they consistently meet rigorous requirements, and adopt a conservative approach to risk assessment. While regulation ensures stability, it also introduces inefficiencies, which is not necessarily bad for us. By this, I mean that inefficiencies create opportunities. And in the realm of lending, this space of opportunities has been (and remains) really huge.
What is the external manifestation of these inefficiencies? It is the situation when only 15% of SMEs in Latin America have access to formal credit even though the default rate of these SMEs is much lower. Or when an individual is denied a bank loan because he lives in a “wrong” district. Or when budding entrepreneurs with innovative ideas are refused financing because they don’t have a long enough credit history. The list goes on, but I think the problem is already pretty clear at this point.
Ok, the problem is clear, but how can we solve it? Remember, the root of the problem lies in banking regulation - to say something like “let’s change it” would be too naive, it’s pretty apparent that, at this point, we must accept banking regulation as an unchangeable fact. This is where fintech-lending startups come into the scene and bring a bunch of different solutions, namely:
let’s aggregate banks’ products in one place, it will ease the borrowing process and make it more efficient (Credible);
let’s help banks with underwriting (within regulation’s restrictions) with our innovative techniques (Upstart);
we underwrite loans better than banks, let’s bring better terms to our borrowers and connect capital from our partners (or from anyone) with these loans (Prosper);
there are so many local companies that underwrite loans better than banks, let’s enable them to sell (securitize) their loans and let investors get exposure to such loans (Lendermarket).
These solutions are often interwoven and always go with comfortable UI, relevant integrations, accelerated lend-borrow processes, and reduced costs for all participants. These above are just a few examples - 5 minutes of googling will bring you hundreds of fintech-lending startups from every possible area (mobile app, software for banks, P2P-platform, and so on) with billions of investments and tens of billions of originated loans. The important conclusion here is that the market is aware of the existing inefficiencies. It also has ideas of how to fix them, and most of them are not about crypto at all. Not because crypto is irrelevant here but rather because the root of the problems lies more in the socio-economic area than the technical side. Fixing these problems is more about questions like “How to match capital with borrowers properly? Which instruments can help investors manage risks? Who and how should manage legal stuff?” rather than “How to cut off intermediaries and make the system more transparent?”.
But let’s get back to crypto. When we’re talking about blockchain companies entering the lending space, it’s essential to understand that their potential positive impact lies in two areas:
general blockchain features. This is what you can see in almost every whitepaper - transparency, intermediaries’ cut-off, and so on. It’s also true in the context of private credits - no fintech lending can offer cash flow automated by smart contracts and non-custodial money storing. But this is true only to a certain extent. Trust assumptions here are drastically higher than in, let’s say, AAVE. We still have to deal with traditional closed financial systems and a bunch of intermediaries (more on it later);
specific platform’s approach. Blockchain itself cannot improve your underwriting techniques or reduce risks for your investors. How will you connect different actors? Why will your borrowers’ performance be better than that of your web2-competitor? How will you manage defaults? In this area, web3-startup is not different from web2-competitor - here, you can’t say “cause we’re on blockchain”.
Do you start feeling why so many protocols providing uncollateralized lending in web3 failed? The reality is that when comparing the approaches of web3 protocols to non-bank private credits, one can hardly find any fundamental differences between them and their web2-relatives. Even when differences do emerge, they often pertain to some distant future where the protocol will be fully decentralized. So… from my perspective, crypto came to the private credits field not to revolutionize it but rather to ehhance it.
tldr: besides the code, you must trust underwriters, a bunch of intermediaries (custodians, loan originators, etc.), local jurisdiction, and socio-economic logic behind protocol (that “liquidator” will appear/that insurance pool will provide enough coverage/etc.)
When you send Bitcoin, you trust the system - that there will be no double spending, that your BTC will not suddenly disappear from the ledger. When you use DEX/liquid staking/lending/CDP/etc. protocols, first of all, you trust the inner logic of the protocol’s design and their smart contracts (with minimum trust in people) - that algorithmic stablecoin will not go to zero cause game theory was misinterpreted, that no rug will happen, that there is no room for hack.
Using centralized stablecoins (USDT, USDC, etc.) is a little more nuanced - you trust not only their inner logic’s design and code but also trust people (legal entity emitting stablecoins, a bank which custody emitent’s reserves, and so on) and jurisdiction (in the case of USDC, it’s USA).
Lending without crypto-collateral is even more nuanced, and often, it’s not easy to understand what are the trust assumptions here. In this section, I’ll try to explain who you must trust and what exactly “trust” is in the private credit field. We’ll look at Centrifuge, Goldfinch, Maple, Clearpool, and Atlendis.
First of all, Centrifuge is rather a platform for loan originators to sell their existing loans from the real world and for investors to get exposure to the corresponding yield venue than for creating an opportunity for people/SMEs to get loans (though the difference between these two phenomena is not always clear). The same web2-projects are represented by Mintos, Lendermarket, and Peerberry.
Centrifuge leverages the well-known in traditional finance concept of securitization. The whole process, from loan origination to repayment/default, looks as follows:
loan originator originates loans in the real world - he evaluates the borrowers’ creditworthiness, negotiates loan terms, and enters into legal agreements with the borrowers;
then borrowers’ obligations are moving to SPV (legal entity separate from the loan originator) - from this moment, SPV is a creditor from a legal perspective, but not an initial loan originator;
SPV “tokenizes” received assets - by tokenization here, we mean a complex of technical (smart contracts) and legal (legal agreements) techniques that ultimately allow us to say “yes, this NFT really represents SPV’s assets (borrowers’ obligations)”;
SPV creates a pool where (1) investors can lend their money to SPV and expect them to be paid back due to debts repaid by borrowers, (2) tokenized borrowers’ obligations act as collateral - in case of default, investors will have a legal right to pursue SPV’s borrowers.
In practice, this process may be more nuanced - sometimes, we have guarantees from loan originators, operating agreements under which loan originators are still managing issued loans, and so on. Also, it’s worth noting that the protocol is asset-class agnostic, meaning that potentially, not only debts but other productive assets can be tokenized and used as collateral.
In any economy, defaults are considered an inherent consequence of lending activities, and the objective of the economic system is to distribute corresponding risks between different actors to ensure the stability of the system as a whole. In other words, defaults among Centrifuge’s borrowers are inevitable - the ultimate goal of the protocol is not to prevent or ban them but rather to establish the protocol mechanics enabling investors to properly evaluate their risk/returns (including frameworks for managing defaults, if needed) and manage their investments. Today, even though Centrifuge has more than $5m defaulted loans, its TVL steadily grows, and there are no signs of disbelief or doubts about Centrifuge protocol in general. I guess it means that Centrifuge is doing it right.
As previously mentioned, Centrifuge employs the TradFi concept of securitization. Delving into its mechanics is beyond the scope of this paper. owever, I encourage anyone interested in the topic to explore Centrifuge’s documentation - they’ve done a really great job adapting a classical financial instrument for onchain use!
I guess you noticed that “trust” here is not just “trust in code”, so let’s summarize the trust assumptions in Centrifuge’s mechanics. Investing in a Centrifuge pool means for the investor that he trusts:
that loan originator holds required underwriting skills - that SPV will not face something like a “90% default rate”;
that SPV/loan originator knows how to manage default situations (remember, we still live more in the real world than in digital - legal contracts are not automatically executable, hence we need lawyers to go to court or collectors to buy defaulted debts);
that SPV/loan originator and all other legal entities and people behind them are not scammers (and will not run away with investors’ money) and have the minimum skills required to run a business in the real world (so their legal entities will not go into bankruptcy because of buying some POKT tokens);
finally, he trusts jurisdiction - that third parties (collectors potentially buying defaulted debts) and courts will treat investors’ tokenized rights as legit legal rights.
Quite a lot of trust assumptions compared to “code is law”, huh?
Goldfinch is much more “borrower-oriented” than Centrifuge, something like Fundingsocieties in web2. Here, it’s unnecessary to have a pool of liquid and potentially tokenized assets to become a borrower, one can just prove his creditworthiness by any means. Let’s take a brief look at the borrowing process:
potential borrower seeks auditors’ approval (checks on borrowers to confirm they are legitimate - no creditworthiness evaluating, just protection against fraud) - currently, the system is not live (Goldfinch team conducts audits);
after approval, the borrower creates a borrower pool (it’s like proposing a “term sheet”) - at the same time, the borrower needs to convince backers (first-loss capital providers) to supply junior tranche (first-loss) capital to the pool (no backers = no money);
backers provide first-loss capital to specific borrower’s pool;
investments from the senior pool are automatically allocated across different borrowers’ pools according to the value of backers’ investments (more backers’ capital = more money from the senior pool).
As in Centrifuge, here we inevitably collide with a bunch of intermediaries (banks, facility and security agents, custodians, etc.).
If Centrifuge is all about transferring the TradFi concept of securitization onchain and connecting investors with underwriters from the real world, Goldfinch is more about creating a semi-automatic underwriting protocol. This means that they’re trying to overcome existing web2 competitors not only by bringing general blockchain features to private credits but also by establishing more robust and effective socio-economic frameworks. In my opinion, some of their ideas are very questionable. Since analyzing Goldfinch is out of this paper’s scope, I’ll just formulate a few questions regarding Goldfinch mechanics:
auditors: why would ordinary people from random countries perform basic due diligence better than local lawyers? Why are the auditor’s tokens slashed when he votes No when the majority votes Yes - is it some form of incentivized conformism?
backers: is it really a good idea for non-backers investors to put their trust in anonymous backers? Why would anonymous actors originate and manage loans better than professional actors in existing web2 platforms?
the basic logic “the more backers who ‘back’ the pool, the more trustworthy the protocol believes the borrower to be”: what could stop a fraudulent borrower from creating a bunch of anonymous backers, funding its pool, getting money from the senior pool, and then running with them (assertions like “auditors will help” and “it’s too difficult and expensive cause of Unique Entity Check” sounds flimsy)?
As for Centrifuge, defaults among Goldfinch borrowers are supposed to be part of the routine. But FIDU priced at $0,77 and 44m FIDU (out of 66m) waiting in withdrawal queue
What are the trust assumptions here? Goldfinch investors (non-backers) must trust:
that anonymous backers from all pools hold required underwriting skills and sanely evaluate borrowers’ creditworthiness;
that someone will manage default situations (since there's no clear mechanism for how investors from the senior pool can be protected in the event of a default, we see here “pure trust” that someone will assist these investors - currently, you can watch what happens in default-situation in real-time);
that borrower and other legal entities and people behind them are not scammers (and will not run away with investors’ money) and have the minimum skills required to run a business in the real world (so their legal entities will not go into bankruptcy because of buying some POKT tokens);
finally, they trust jurisdiction - that third parties (collectors potentially buying defaulted debts) and courts will treat investors’ tokenized rights as legit legal rights.
We see even more trust assumptions than in Centrifuge - some are based on trust in anonymous backers, and some are based on literally nothing. Now we’re even further from “code is law”…
Maple (the under-collateralized part of it, which is abandoned now) could be seen as an underwriter-centered platform, which makes it look much more like a Centrifuge, but with one big difference - loans here were not collateralized by off-chain assets. The borrowing process worked as follows (old docs can be found here):
pool delegates (“experienced professionals”) create and manage lending pools - they conduct due diligence and agree on terms with borrowers;
institutional borrowers can borrow without collateral from lending pools at terms approved by pool delegate;
lenders provide capital to a pool to earn interest from institutional borrowers.
On top of that, some measures existed to reduce investors’ risks (insurance from pool delegates, collateral from borrowers, etc.), but the basic idea was pretty simple - “just give your money to the pool delegate, and he’ll manage everything”. We know how it all ended.
Don’t see any sense in going deeper into Maple’s mechanics. Money was literally given to high-risk borrowers through the high-risk intermediaries with minimal insurance and without any collateralization. I don’t know in which reality this scheme would work.
In my opinion, Maple is an example of a pretty good underlying idea with a terrible realization. Now Maple’s representatives tell us that “no one’s profitable” in under-collateralized lending”, but it’s simply not true. It’s always easier to blame circumstances/market conditions/that curly guy than to accept that you fucked up. If you end up in a situation where tens of millions of dollars are given to high-risk borrowers, and the only defense measure is small collateral represented in your shitcoin (MPL), obviously, it’s something wrong with your approach, not with the market. By the way, a pretty similar (but more generalized and more flexible) design was proposed some time ago by Circle with their Perimeter protocol.
Trust assumptions in Maple were pretty similar as for Centrifuge and Goldfinch - trust in pool delegate’s underwriting and general business skills, in legal entities and people behind them and behind borrowers, and in jurisdictions. But in this case, investors’ trust was mainly (and, unfortunately, vainly) concentrated around pool delegates.
Since other protocols leverage the same mechanics, I won’t torment you with a detailed description of them, but just note two details in Clearpool and Atlendis that are worth attention.
First, it’s Clearpool’s approach to default. Unlike Centrifuge, which delegates the management of defaults to loan originators, or Goldfinch, which relies on unidentified entities, Clearpool's primary model involves selling defaulted debts in open auctions. This strategy seems intuitive and sensible (in most cases, lenders manage defaults precisely in this way - by selling debts to entities that know better how to manage defaulted loans). However, this introduces an added layer of trust: the belief that there will always be a willing buyer for such debts.
Second, Atlendis’ approach to evaluating borrower’s creditworthiness. Rather than placing trust in underwriters (individuals), they utilize an automated risk assessment through the Cred Protocol (Clearpool employs a similar partnership with Credora). It’s much closer to “code is law”, though only time will determine the effectiveness of this approach.
The straightforward answer is "no". Yet, this answer isn't entirely accurate. It would hold true if we lived in a black-and-white world, but our world is not like that.
Recall how I began: by noting that the web3 and legacy financial systems are currently observing each other, seeking ways to forge effective communication channels. This is the foundation for understanding the trajectory of crypto values within the realm of private credits. Subsequent interpretations hinge largely on one's perspective of crypto's interaction with the so-called real world.
In my view, our initial steps will necessitate some compromise on web3-values: we will have a full spectrum of trusted intermediaries emerge and will rely heavily on trust within jurisdictions. There seems to be no other way to meaningfully engage with the real world. As a result, we'll also encounter the full spectrum of consequences stemming from these concessions - uncontrolled borrowers going bankrupt because of buying POKT tokens without any reaction from trusted anons (backers), scammers launching their lending pools, and so on.
Concurrently, a segment of the web3 ecosystem that doesn't require such ties will persistently showcase the merits of foundational web3 beliefs and highlight the flaws of the traditional system. I envision a future where the conventional world acknowledges these truths in various forms - from recognizing tokens as legitimate representations of ownership to integrating protocols within national registries. Eventually, we’ll find ourselves in a position where the difference between web3 and the real world simply doesn’t exist - a trusted intermediary will be a fully onchain DAO, and network states will handle legal procedures. At this point, ‘real-world assets’ will finally become ‘assets’, and the question of preserving web3-values will not be relevant. Whether current protocols will adapt to these shifts and lay a robust foundation for future applications or fade into obscurity, remains to be seen.
It looks like some trust assumptions might persist - in the context of undercollateralized lending, underwriters vetted by investors might still play a role. Or maybe this paradigm will drastically change, and something more powerful than currently existing frameworks will come - reputation-based lending, the concept of shared money, or some other crazy (from today’s perspective) stuff. But that's an entirely different story.
Currently, we're observing a fascinating intersection of crypto with the “real-world”. Private credits are one of the most illustrative examples, showcasing a partial departure from foundational web3-values to pursue pragmatic objectives and foster connections with tangible systems.
From my perspective, in most cases, we’re not revolutionizing anything (as we did during the ICO-boom). In one of its articles, Credix delivered a grounded and precise depiction of what crypto primarily introduces to the realm of private credits - “we leverage blockchain technology to create more transparency in our portfolios, automated payment calculations, secured decentralized custody of assets, and global 24/7 settlement”. It’s a straightforward and, in some ways, boring statement - without any superfluous grandiosity or buzzwords.
But it’s just one of the first steps towards full web3 and real-world convergence. I believe that as we progress, values from both domains will undergo metamorphosis. Notably, the “code-is-law” principle will need to evolve to accommodate the complexity of human relationships. But we must not be afraid or resist these changes - whatever the semantics we employ to define web3 values, fundamentally, these values already form the basis of open, decentralized networks, and it seems unlikely that these networks will vanish. In other words, decentralized networks like Ethereum can tolerate any experiments - you can try to reconcile them with human elements or with legacy financial systems, or even try to bring governments there. The more experiments we witness, the better. It feels that none of that can halt the momentum of web3-values (no matter how they're articulated) as they anchor open networks and continue permeating the (as-yet-named) real world.
More content on my TG-channel (it’s in my native language, but don’t hesitate to click on “Translate to English“)
What is the biggest narrative in crypto for 2023? Right, it’s all about Real-World Assets (beyond stablecoins) - tokenized treasury bills and equity, yield from private credits, markets for carbon emissions, fractionalized real estate, and so on. There is no doubt that “that ‘real-world assets’ will eventually simply be ‘assets’”, but since we’re still in the early days for the sake of clarity, I’ll allow myself to call it RWA.
For those who are not familiar with the topic, you can go through these papers (they’re all pretty similar tbh): Overview of on-chain RWAs and the forces propelling their growth from Galaxy, State of the (RWA) industry report from FortunaFi, RWA report from Redstone and Chaos Labs. Or check the list of resources from RWA.zyx.
From the early days of Bitcoin, we saw the desire of crypto-natives to separate themselves and their permissionless habitat (what we call today “web3”) from authoritarian legacy financial systems. The pre-RWA era of web3 had a pretty straightforward narrative, which can be described as “don’t trust people - trust code” or “code is law”. In other words, the ultimate goal for many protocols was to deliver a product (in most cases - a financial product) with trust assumptions minimized as much as possible. But times are changing. Today, both systems have come to the point where they cannot ignore each other and, at the same time, don’t understand how to treat each other:
crypto has proven that the code actually works even in complex financial applications, but the closed self-centered ecosystem is becoming too small - crypto is staking its claim in new fields;
governments publicly accepted that crypto is much more than a gambling crap for a bunch of marginals, but they are still not ready to integrate it into their systems.
Still think that governments do not understand how powerful crypto is? Check this paper from the Federal Reserve - their researchers fear that tokenized assets could transmit volatility from crypto asset markets to the markets for the crypto token's reference assets. Sounds pretty powerful.
Here we are, at this juncture. Intuition tells us that these systems can no longer exist in isolation. Their gradual convergence seems more like a natural evolutionary process than the whims of random actors. However, this convergence cannot occur while fully preserving web3 values, particularly the “code-is-law” dogma. The emergence of Real-World Assets made this evident. If we want web3 to “adopt” a portion of the “Real-World”, some compromises are necessary (to be fair, governments and legacy financial systems are also making concessions concerning crypto). In this piece, I aim to delve into these compromises within the context of onchain private credit markets and address the ensuing questions:
what are private credit protocols all about - do we observe a revolution in private credit space or just minor adjustments to the existing system?
who do we trust when using these protocols - their code, their team, national governments?
is it even possible to keep original web3-values untouched when we deal with private credits in the real world?
We won’t dive deep into specific details like legal frameworks, methods of NAV calculation, etc. (unless it’s necessary to understand the protocol in general), but instead will try to provide a high-level understanding of existing approaches to onchain private credit markets. We’ll discuss Centrifuge, Goldfinch, Maple, Clearpool, and Atlendis. CrediX, TrueFi, Ribbon Lend, and dAMM will be out of the scope since they replicate the mechanics of the mentioned protocols and/or have minimal adoption.
tldr: 1. cause this is the area with so many inefficiencies and, hence, so many opportunities - for both lenders/investors and borrowers. 2. no, we’re not revolutionizing it - in most cases, we’re just bringing existing fintech solutions onchain
Private credit markets in web3 started to get attention in 2020. What goals were they stated? I’ll let protocols speak for themselves:
to organize trillions of dollars of global debt when all private debt moves onchain and to offer an incredibly safe, diversified, reliable yield for stablecoins (Goldfinch);
to connect institutional investors globally with credit opportunities, unlocking risk-adjusted returns through robust underwriting, building on the most advanced decentralized technologies to achieve scale (CrediX);
to reduce the reliance on the intermediaries involved in financing real-world assets (lead manager, managers, lawyers, etc.) and create more open, transparent, and efficient access to finance (Centrifuge).
Wording can vary, but I think you grasped the idea - as always in crypto, we’re empowering financial inclusion, eliminating (hated by everyone) intermediaries, and creating a more open and transparent environment. But what precisely are these protocols doing? How can one formulate the essence of the Goldfinch/Centrifuge/YouNameIt protocol? To understand it, let’s briefly examine how lending in the real world works and what Nigerian startups would do if Goldfinch didn’t exist in this world.
“Kings of lending” in the real world are banks, without any doubt. They originate loans, they collect payments, they manage defaults, and so on. And yes, banks operate under strict regulations, which mandate that they consistently meet rigorous requirements, and adopt a conservative approach to risk assessment. While regulation ensures stability, it also introduces inefficiencies, which is not necessarily bad for us. By this, I mean that inefficiencies create opportunities. And in the realm of lending, this space of opportunities has been (and remains) really huge.
What is the external manifestation of these inefficiencies? It is the situation when only 15% of SMEs in Latin America have access to formal credit even though the default rate of these SMEs is much lower. Or when an individual is denied a bank loan because he lives in a “wrong” district. Or when budding entrepreneurs with innovative ideas are refused financing because they don’t have a long enough credit history. The list goes on, but I think the problem is already pretty clear at this point.
Ok, the problem is clear, but how can we solve it? Remember, the root of the problem lies in banking regulation - to say something like “let’s change it” would be too naive, it’s pretty apparent that, at this point, we must accept banking regulation as an unchangeable fact. This is where fintech-lending startups come into the scene and bring a bunch of different solutions, namely:
let’s aggregate banks’ products in one place, it will ease the borrowing process and make it more efficient (Credible);
let’s help banks with underwriting (within regulation’s restrictions) with our innovative techniques (Upstart);
we underwrite loans better than banks, let’s bring better terms to our borrowers and connect capital from our partners (or from anyone) with these loans (Prosper);
there are so many local companies that underwrite loans better than banks, let’s enable them to sell (securitize) their loans and let investors get exposure to such loans (Lendermarket).
These solutions are often interwoven and always go with comfortable UI, relevant integrations, accelerated lend-borrow processes, and reduced costs for all participants. These above are just a few examples - 5 minutes of googling will bring you hundreds of fintech-lending startups from every possible area (mobile app, software for banks, P2P-platform, and so on) with billions of investments and tens of billions of originated loans. The important conclusion here is that the market is aware of the existing inefficiencies. It also has ideas of how to fix them, and most of them are not about crypto at all. Not because crypto is irrelevant here but rather because the root of the problems lies more in the socio-economic area than the technical side. Fixing these problems is more about questions like “How to match capital with borrowers properly? Which instruments can help investors manage risks? Who and how should manage legal stuff?” rather than “How to cut off intermediaries and make the system more transparent?”.
But let’s get back to crypto. When we’re talking about blockchain companies entering the lending space, it’s essential to understand that their potential positive impact lies in two areas:
general blockchain features. This is what you can see in almost every whitepaper - transparency, intermediaries’ cut-off, and so on. It’s also true in the context of private credits - no fintech lending can offer cash flow automated by smart contracts and non-custodial money storing. But this is true only to a certain extent. Trust assumptions here are drastically higher than in, let’s say, AAVE. We still have to deal with traditional closed financial systems and a bunch of intermediaries (more on it later);
specific platform’s approach. Blockchain itself cannot improve your underwriting techniques or reduce risks for your investors. How will you connect different actors? Why will your borrowers’ performance be better than that of your web2-competitor? How will you manage defaults? In this area, web3-startup is not different from web2-competitor - here, you can’t say “cause we’re on blockchain”.
Do you start feeling why so many protocols providing uncollateralized lending in web3 failed? The reality is that when comparing the approaches of web3 protocols to non-bank private credits, one can hardly find any fundamental differences between them and their web2-relatives. Even when differences do emerge, they often pertain to some distant future where the protocol will be fully decentralized. So… from my perspective, crypto came to the private credits field not to revolutionize it but rather to ehhance it.
tldr: besides the code, you must trust underwriters, a bunch of intermediaries (custodians, loan originators, etc.), local jurisdiction, and socio-economic logic behind protocol (that “liquidator” will appear/that insurance pool will provide enough coverage/etc.)
When you send Bitcoin, you trust the system - that there will be no double spending, that your BTC will not suddenly disappear from the ledger. When you use DEX/liquid staking/lending/CDP/etc. protocols, first of all, you trust the inner logic of the protocol’s design and their smart contracts (with minimum trust in people) - that algorithmic stablecoin will not go to zero cause game theory was misinterpreted, that no rug will happen, that there is no room for hack.
Using centralized stablecoins (USDT, USDC, etc.) is a little more nuanced - you trust not only their inner logic’s design and code but also trust people (legal entity emitting stablecoins, a bank which custody emitent’s reserves, and so on) and jurisdiction (in the case of USDC, it’s USA).
Lending without crypto-collateral is even more nuanced, and often, it’s not easy to understand what are the trust assumptions here. In this section, I’ll try to explain who you must trust and what exactly “trust” is in the private credit field. We’ll look at Centrifuge, Goldfinch, Maple, Clearpool, and Atlendis.
First of all, Centrifuge is rather a platform for loan originators to sell their existing loans from the real world and for investors to get exposure to the corresponding yield venue than for creating an opportunity for people/SMEs to get loans (though the difference between these two phenomena is not always clear). The same web2-projects are represented by Mintos, Lendermarket, and Peerberry.
Centrifuge leverages the well-known in traditional finance concept of securitization. The whole process, from loan origination to repayment/default, looks as follows:
loan originator originates loans in the real world - he evaluates the borrowers’ creditworthiness, negotiates loan terms, and enters into legal agreements with the borrowers;
then borrowers’ obligations are moving to SPV (legal entity separate from the loan originator) - from this moment, SPV is a creditor from a legal perspective, but not an initial loan originator;
SPV “tokenizes” received assets - by tokenization here, we mean a complex of technical (smart contracts) and legal (legal agreements) techniques that ultimately allow us to say “yes, this NFT really represents SPV’s assets (borrowers’ obligations)”;
SPV creates a pool where (1) investors can lend their money to SPV and expect them to be paid back due to debts repaid by borrowers, (2) tokenized borrowers’ obligations act as collateral - in case of default, investors will have a legal right to pursue SPV’s borrowers.
In practice, this process may be more nuanced - sometimes, we have guarantees from loan originators, operating agreements under which loan originators are still managing issued loans, and so on. Also, it’s worth noting that the protocol is asset-class agnostic, meaning that potentially, not only debts but other productive assets can be tokenized and used as collateral.
In any economy, defaults are considered an inherent consequence of lending activities, and the objective of the economic system is to distribute corresponding risks between different actors to ensure the stability of the system as a whole. In other words, defaults among Centrifuge’s borrowers are inevitable - the ultimate goal of the protocol is not to prevent or ban them but rather to establish the protocol mechanics enabling investors to properly evaluate their risk/returns (including frameworks for managing defaults, if needed) and manage their investments. Today, even though Centrifuge has more than $5m defaulted loans, its TVL steadily grows, and there are no signs of disbelief or doubts about Centrifuge protocol in general. I guess it means that Centrifuge is doing it right.
As previously mentioned, Centrifuge employs the TradFi concept of securitization. Delving into its mechanics is beyond the scope of this paper. owever, I encourage anyone interested in the topic to explore Centrifuge’s documentation - they’ve done a really great job adapting a classical financial instrument for onchain use!
I guess you noticed that “trust” here is not just “trust in code”, so let’s summarize the trust assumptions in Centrifuge’s mechanics. Investing in a Centrifuge pool means for the investor that he trusts:
that loan originator holds required underwriting skills - that SPV will not face something like a “90% default rate”;
that SPV/loan originator knows how to manage default situations (remember, we still live more in the real world than in digital - legal contracts are not automatically executable, hence we need lawyers to go to court or collectors to buy defaulted debts);
that SPV/loan originator and all other legal entities and people behind them are not scammers (and will not run away with investors’ money) and have the minimum skills required to run a business in the real world (so their legal entities will not go into bankruptcy because of buying some POKT tokens);
finally, he trusts jurisdiction - that third parties (collectors potentially buying defaulted debts) and courts will treat investors’ tokenized rights as legit legal rights.
Quite a lot of trust assumptions compared to “code is law”, huh?
Goldfinch is much more “borrower-oriented” than Centrifuge, something like Fundingsocieties in web2. Here, it’s unnecessary to have a pool of liquid and potentially tokenized assets to become a borrower, one can just prove his creditworthiness by any means. Let’s take a brief look at the borrowing process:
potential borrower seeks auditors’ approval (checks on borrowers to confirm they are legitimate - no creditworthiness evaluating, just protection against fraud) - currently, the system is not live (Goldfinch team conducts audits);
after approval, the borrower creates a borrower pool (it’s like proposing a “term sheet”) - at the same time, the borrower needs to convince backers (first-loss capital providers) to supply junior tranche (first-loss) capital to the pool (no backers = no money);
backers provide first-loss capital to specific borrower’s pool;
investments from the senior pool are automatically allocated across different borrowers’ pools according to the value of backers’ investments (more backers’ capital = more money from the senior pool).
As in Centrifuge, here we inevitably collide with a bunch of intermediaries (banks, facility and security agents, custodians, etc.).
If Centrifuge is all about transferring the TradFi concept of securitization onchain and connecting investors with underwriters from the real world, Goldfinch is more about creating a semi-automatic underwriting protocol. This means that they’re trying to overcome existing web2 competitors not only by bringing general blockchain features to private credits but also by establishing more robust and effective socio-economic frameworks. In my opinion, some of their ideas are very questionable. Since analyzing Goldfinch is out of this paper’s scope, I’ll just formulate a few questions regarding Goldfinch mechanics:
auditors: why would ordinary people from random countries perform basic due diligence better than local lawyers? Why are the auditor’s tokens slashed when he votes No when the majority votes Yes - is it some form of incentivized conformism?
backers: is it really a good idea for non-backers investors to put their trust in anonymous backers? Why would anonymous actors originate and manage loans better than professional actors in existing web2 platforms?
the basic logic “the more backers who ‘back’ the pool, the more trustworthy the protocol believes the borrower to be”: what could stop a fraudulent borrower from creating a bunch of anonymous backers, funding its pool, getting money from the senior pool, and then running with them (assertions like “auditors will help” and “it’s too difficult and expensive cause of Unique Entity Check” sounds flimsy)?
As for Centrifuge, defaults among Goldfinch borrowers are supposed to be part of the routine. But FIDU priced at $0,77 and 44m FIDU (out of 66m) waiting in withdrawal queue
What are the trust assumptions here? Goldfinch investors (non-backers) must trust:
that anonymous backers from all pools hold required underwriting skills and sanely evaluate borrowers’ creditworthiness;
that someone will manage default situations (since there's no clear mechanism for how investors from the senior pool can be protected in the event of a default, we see here “pure trust” that someone will assist these investors - currently, you can watch what happens in default-situation in real-time);
that borrower and other legal entities and people behind them are not scammers (and will not run away with investors’ money) and have the minimum skills required to run a business in the real world (so their legal entities will not go into bankruptcy because of buying some POKT tokens);
finally, they trust jurisdiction - that third parties (collectors potentially buying defaulted debts) and courts will treat investors’ tokenized rights as legit legal rights.
We see even more trust assumptions than in Centrifuge - some are based on trust in anonymous backers, and some are based on literally nothing. Now we’re even further from “code is law”…
Maple (the under-collateralized part of it, which is abandoned now) could be seen as an underwriter-centered platform, which makes it look much more like a Centrifuge, but with one big difference - loans here were not collateralized by off-chain assets. The borrowing process worked as follows (old docs can be found here):
pool delegates (“experienced professionals”) create and manage lending pools - they conduct due diligence and agree on terms with borrowers;
institutional borrowers can borrow without collateral from lending pools at terms approved by pool delegate;
lenders provide capital to a pool to earn interest from institutional borrowers.
On top of that, some measures existed to reduce investors’ risks (insurance from pool delegates, collateral from borrowers, etc.), but the basic idea was pretty simple - “just give your money to the pool delegate, and he’ll manage everything”. We know how it all ended.
Don’t see any sense in going deeper into Maple’s mechanics. Money was literally given to high-risk borrowers through the high-risk intermediaries with minimal insurance and without any collateralization. I don’t know in which reality this scheme would work.
In my opinion, Maple is an example of a pretty good underlying idea with a terrible realization. Now Maple’s representatives tell us that “no one’s profitable” in under-collateralized lending”, but it’s simply not true. It’s always easier to blame circumstances/market conditions/that curly guy than to accept that you fucked up. If you end up in a situation where tens of millions of dollars are given to high-risk borrowers, and the only defense measure is small collateral represented in your shitcoin (MPL), obviously, it’s something wrong with your approach, not with the market. By the way, a pretty similar (but more generalized and more flexible) design was proposed some time ago by Circle with their Perimeter protocol.
Trust assumptions in Maple were pretty similar as for Centrifuge and Goldfinch - trust in pool delegate’s underwriting and general business skills, in legal entities and people behind them and behind borrowers, and in jurisdictions. But in this case, investors’ trust was mainly (and, unfortunately, vainly) concentrated around pool delegates.
Since other protocols leverage the same mechanics, I won’t torment you with a detailed description of them, but just note two details in Clearpool and Atlendis that are worth attention.
First, it’s Clearpool’s approach to default. Unlike Centrifuge, which delegates the management of defaults to loan originators, or Goldfinch, which relies on unidentified entities, Clearpool's primary model involves selling defaulted debts in open auctions. This strategy seems intuitive and sensible (in most cases, lenders manage defaults precisely in this way - by selling debts to entities that know better how to manage defaulted loans). However, this introduces an added layer of trust: the belief that there will always be a willing buyer for such debts.
Second, Atlendis’ approach to evaluating borrower’s creditworthiness. Rather than placing trust in underwriters (individuals), they utilize an automated risk assessment through the Cred Protocol (Clearpool employs a similar partnership with Credora). It’s much closer to “code is law”, though only time will determine the effectiveness of this approach.
The straightforward answer is "no". Yet, this answer isn't entirely accurate. It would hold true if we lived in a black-and-white world, but our world is not like that.
Recall how I began: by noting that the web3 and legacy financial systems are currently observing each other, seeking ways to forge effective communication channels. This is the foundation for understanding the trajectory of crypto values within the realm of private credits. Subsequent interpretations hinge largely on one's perspective of crypto's interaction with the so-called real world.
In my view, our initial steps will necessitate some compromise on web3-values: we will have a full spectrum of trusted intermediaries emerge and will rely heavily on trust within jurisdictions. There seems to be no other way to meaningfully engage with the real world. As a result, we'll also encounter the full spectrum of consequences stemming from these concessions - uncontrolled borrowers going bankrupt because of buying POKT tokens without any reaction from trusted anons (backers), scammers launching their lending pools, and so on.
Concurrently, a segment of the web3 ecosystem that doesn't require such ties will persistently showcase the merits of foundational web3 beliefs and highlight the flaws of the traditional system. I envision a future where the conventional world acknowledges these truths in various forms - from recognizing tokens as legitimate representations of ownership to integrating protocols within national registries. Eventually, we’ll find ourselves in a position where the difference between web3 and the real world simply doesn’t exist - a trusted intermediary will be a fully onchain DAO, and network states will handle legal procedures. At this point, ‘real-world assets’ will finally become ‘assets’, and the question of preserving web3-values will not be relevant. Whether current protocols will adapt to these shifts and lay a robust foundation for future applications or fade into obscurity, remains to be seen.
It looks like some trust assumptions might persist - in the context of undercollateralized lending, underwriters vetted by investors might still play a role. Or maybe this paradigm will drastically change, and something more powerful than currently existing frameworks will come - reputation-based lending, the concept of shared money, or some other crazy (from today’s perspective) stuff. But that's an entirely different story.
Currently, we're observing a fascinating intersection of crypto with the “real-world”. Private credits are one of the most illustrative examples, showcasing a partial departure from foundational web3-values to pursue pragmatic objectives and foster connections with tangible systems.
From my perspective, in most cases, we’re not revolutionizing anything (as we did during the ICO-boom). In one of its articles, Credix delivered a grounded and precise depiction of what crypto primarily introduces to the realm of private credits - “we leverage blockchain technology to create more transparency in our portfolios, automated payment calculations, secured decentralized custody of assets, and global 24/7 settlement”. It’s a straightforward and, in some ways, boring statement - without any superfluous grandiosity or buzzwords.
But it’s just one of the first steps towards full web3 and real-world convergence. I believe that as we progress, values from both domains will undergo metamorphosis. Notably, the “code-is-law” principle will need to evolve to accommodate the complexity of human relationships. But we must not be afraid or resist these changes - whatever the semantics we employ to define web3 values, fundamentally, these values already form the basis of open, decentralized networks, and it seems unlikely that these networks will vanish. In other words, decentralized networks like Ethereum can tolerate any experiments - you can try to reconcile them with human elements or with legacy financial systems, or even try to bring governments there. The more experiments we witness, the better. It feels that none of that can halt the momentum of web3-values (no matter how they're articulated) as they anchor open networks and continue permeating the (as-yet-named) real world.
More content on my TG-channel (it’s in my native language, but don’t hesitate to click on “Translate to English“)
No comments yet