An outrageously TradFi look at the state of DeFi credit in early 2023

14 min readMar 15, 2023

Before I became a venture capitalist I was working in leveraged finance. These desks within investment banks help private equity investors structure and finance their buyouts. What surprised me about this job was the complete absence of technology and data science. We had $ billions of loans outstanding, and had decades of detailed financial data of hundreds of companies in our database, but none of it was ever mined to improve our decision-making or increase our efficiency.

Coming from this background, I am naturally drawn to any innovation within credit and lending. And I continue to believe that decentralized finance (DeFi) has the potential to become the most fundamental disruption to financial services that we have seen in a very long time. I would therefore like to explore the current state of lending in DeFi, and whether real progress has been made compared to traditional finance (TradFi).

The following article is structured as follows: after a quick intro on DeFi I will set the scene by sizing debt as an asset class. I will show how large and important debt is within TradFi, before comparing it with the DeFi world. I will thereby consider product availability and outstanding loan volumes across both consumer and corporate lending. I will finish the article with a few thoughts on how to grow DeFi loan volumes and by highlighting some startups that could enable this.

I know that the timing of this blog post is not ideal as I cannot show graphs that point up and to the right. But if you can get past that, I hope I will be able to show you that — even amid a crypto winter — something significant is brewing.

The concept of DeFi is wild. You take financial products, break them down into their primitives (components), and put them on the internet for anyone to use. And in the process you create an instantly-global financial market with no gatekeepers, and a fertile ground for continuous experimentation to build new and better financial products.

DeFi has gone from being almost non-existent in Jan-20 to generating more than $1bn in fees in Jan-22 alone (see below). Within two years, dozens of protocols have been built and deployed. Lots of experimentation has led to some successful, some not so successful outcomes; the inevitable result of an open ecosystem with low entry barriers and no regulation to slow it down.

The chart above depicts the fees generated by the top-9 largest DeFi protocols (excluding Axie Infinity as I do not think it belongs here). In terms of categories, most fees were generated by decentralized exchanges ($4.5bn), followed by NFT marketplaces ($3.2bn), then staking ($1.0bn) and lending ($909m).

Yes, the graph above is parabola-shaped rather than “up and to the right”. This is due to the ongoing crypto winter where activity is reduced. However, the chart does prove that people are willing to use and pay for DeFi products. Moreover, it already looks like activity slowly began to pick up again in Jan-23. I fully expect us to see volumes surpass the Jan-22 levels again in the future as better products will increase the utility of DeFi and these products become more seamlessly integrated into the daily lives of many people.

Now let us zoom into the lending vertical and start by sizing the market opportunity.

Debt is the fuel of our lives

Debt is everywhere. It fuels our governments, our economies and consumers. According to the IIF’s global debt monitor (as of Jun-22), global nominal debt outstanding was c. $300tn, having shown a +5% CAGR since 2016.

The largest borrowers are governments and non-financial corporates, each holding a 29% share in Jun-22. They are followed by financial corporations (23%) and finally households (19%). It is noteworthy that over the past six years, not a single category decreased its outstanding balance.

Taken overall, $300tn is a massive amount of debt. Compare it to the $95tn total value of global equity markets at their recent peak in Nov-20. Or to the c. $3tn total market cap of all crypto at its peak in Nov-21.

It is no surprise then, that DeFi has started to build product in this asset class. In 2019, Compound started facilitating their first loans, and soon after a number of other protocols followed suit.

DeFi lending truly took off in the beginning of 2021 and strongly accelerated throughout that year. It peaked at around $35bn borrowing volume in Nov-21, before dropping amid the broader market turmoil. The three leading protocols today are

MakerDAO

, Aave and Compound, with outstanding loan volumes of $5.2bn, $2.1bn and $728m respectively (3 Feb 23).

The key innovation brought forward by DeFi protocols is that they are able to issue loans instantly and at scale that are fully automated, anonymous and permissionless. Users connect their wallets to a protocol and take out a loan from the available pools of capital without any human involvement. The jury is still out whether a pure-play approach like this can reach significant scale, or whether some permissioning is needed for regulatory purposes. However, it does constitute a significant innovation and there is no equivalent of this in the TradFi world.

Before we get into the details of DeFi lending, its use cases and product variety, let me make a brief detour. Bear in mind also that for the rest of this article I will focus on non-financial corporates and household debt as proxies for real-economy adoption of on-chain debt.

A quick tangent: unsecured versus secured debt

At this point I want to take a quick look at the structural difference between secured and unsecured debt, and how prevalent the two are in off-chain consumer and corporate banking.

Secured loans are issued against collateral. A borrower in need of a loan posts collateral (i.e. securities) that the lender can claim in the event of a default and sell in the market to recoup his losses. The degree of collateralization can be different, ranging from partly secured to overcollateralized. In any case, having collateral de-risks the loan for the lender and hence translates into better terms for the borrower. Mortgages or car loans are typical examples of secured consumer debt. Corporate debt can be secured for example by posting a machine park as collateral.
Unsecured loans are much more risky for a lender as there is no asset to claim in the event of a default. Such loans are instead issued against the credit history and score of a borrower and are often relationship-driven. The higher risk of these loans translates into less favourable terms and generally higher interest rates. Credit cards and student loans are examples of unsecured consumer debt. In the corporate world, most bonds and loans are unsecured.

Taking the US as a proxy (due to good data availability), we can look at the split between secured versus unsecured loans in both categories:

c. 20% of consumer loans are unsecured, which implies a global volume of $11.5tn in outstanding unsecured household debt (source: NY Fed as of Nov-22)
c. 85% of corporate debt is unsecured, which implies a global volume of $75tn in outstanding unsecured corporate debt (source: The Decline of Secured Debt by Benmelech, Kumar, Rajan as of Aug-19)

Globally there is c. $144tn of consumer and corporate debt outstanding. Of that, c. $86tn (59%) is unsecured, with the remainder (c. $58tn) being secured.

The state of play in on-chain loans

I am going to split the next section into consumer versus corporate lending. For the sake of this analysis, I am grouping all protocols into the consumer category unless they have a declared focus on institutional players and run on permissioned access (in which case I will count them as corporate lenders).

Now that we know how big debt is as an asset class and what type of loans there are, let us look at the current state of on-chain loans. Which TradFi products are already accessible through DeFi, and what are the current loan volumes relative to the fiat world?

Consumer loans

As we saw above, on-chain debt started taking off in 2021 with outstanding loans exceeding $35bn by late 2021 / early 2022. Most of this volume was in fact driven by protocols like Aave and Compound that are mostly accessible to consumers (excluding their institutional pools for now). Below I compare TradFi consumer loan categories to their on-chain equivalents, look at the current outstanding loan volume of each and show a selection of active protocols per category.

Sources with clickable links can also be found at the bottom of this post

The largest off-chain loan categories (mortgages, unsecured student loans, credit cards), are all but unavailable in the DeFi world to date. All consumer on-chain loans have so far been driven by a single category: over-collateralized loans secured through existing assets (coins, tokens or NFTs).

There is one protocol working on DeFi mortgages (Moon Mortgage), but they are pre-launch and it sounds like borrowers will have to collateralize the value of the real estate in full with existing crypto assets. This shows how early we are in the development of DeFi debt, and how much upside potential there still is.

The key innovation of the above protocols is that they are able to issue loans instantly and at scale in a way that is completely automated, anonymous and permissionless. There is no equivalent of this in the TradFi world. However, so far all these loans are secured by substantial collateral as there is no trusted relationship or trusted intermediary between the borrowers and lenders. Hence, collateral is needed to protect the lender against fraudulent borrower behaviour.

In fact, when you visit the Compound website, this will be the first thing you see:

It shows that there are currently $859m of loans outstanding (3 Feb 23), backed by collateral worth $2.6bn. The average borrower therefore had to put up 3x the amount of their loan in collateral. Imagine having to lock assets worth $30,000 at your bank before being able to take out a $10,000 loan.

This is obviously not only a very restrictive policy but also one that is highly capital inefficient. Consumers without existing crypto assets are not able to take out loans. And while you can take out a loan backed by assets like NFTs, you can not take one out to acquire them in the first place (i.e. take out a “mortgage” on them).

The current lending protocols are used for a variety of use cases, but mostly for leveraged trading strategies. Very few use cases beyond that are currently enabled by the existing protocols.

Corporate debt

The TradFi corporate debt landscape is a bit more complex than that of consumer debt. Corporate debt can be classified in a number of ways: secured vs unsecured, short-term vs long-term, senior vs subordinated, loans vs bonds, etc. I would have liked to do an analysis along the lines of the use cases above, but the data is just not there. We will therefore have to make do with the Fed’s categorisation of debt:

Sources with clickable links can also be found at bottom of this post

Admittedly, the table above looks bleak in terms of outstanding on-chain lending volumes. However, we have to keep in mind that we are in the depths of a DeFi winter. In Q2 2022,

TrueFi

and Maple each had about $1bn in outstanding loans and volumes were much higher across the board.

The table above does show, however, that protocols are working on a variety of on-chain lending products, both in the secured and unsecured categories:

A number of protocols are working on DeFi bonds; volumes are still negligible, but companies like DeBond, obligate, Arbor Finance or Notional are building the infrastructure to unlock this product.
Bank lending is a broad category, and the one with the most on-chain activity to date, with projects across secured and unsecured debt.
Secured corporate loans are available on platforms like Aave Arc or Compound Treasury; both Arc and Treasury are sub-products of their namesake protocols that specifically target an institutional client base where participants are whitelisted and KYC’d.
The largest and most promising forays into unsecured on-chain lending have been made by Maple,
TrueFi
,
Clearpool
, etc.; borrowers undergo a KYC / onboarding flow and loans are given based on a manual (albeit tech-augmented) credit analysis. However, these platforms are not automated, anonymous or permissionless and are therefore (somewhat) at odds with the ethos of DeFi.
Leveraged loans, which are mostly associated with acquisition finance / buyouts in the off-chain world, are not yet represented on-chain.

Even if the corporate landscape offers a greater product variety than the consumer landscape, it has so far only scratched the surface of what is possible. These protocols are mainly used by market markers and hedge funds to source liquidity in native tokens (so they fall somewhat into the “financial corporates” category). Very few use cases beyond providing trading liquidity have been successfully unlocked so far.

Bonus category: Real World Asset (RWA) Debt

We have not yet considered one particularly interesting group of players: Real World Asset protocols. They are a sub-category of the above but with the speciality of bridging the on- and off-chain worlds. They pool liquidity on-chain and use it to finance assets / activities off-chain. For example:

These loans can be either secured or unsecured, depending on the specific borrower profile and loan type issued.

The big three protocols in this space are

Centrifuge

, Credix and

Goldfinch

. These protocols arbitrage between different local debt capital markets and make real-world yields available on-chain. While all other lending protocols have suffered from the FTX contagion, the RWA protocols have held up well. See below, by way of example, the borrowing volume on Centrifuge over time. The other two have held up similarly well — see here and here.

One can argue that RWA protocols are not truly DeFi, and that would be correct. The liquidity pools are permissioned, investors are accredited and loans are not issued automatically. What we are left with, technically, is a more transparent back-end as compared to TradFi. Commercially, however, it increases the utility of the DeFi ecosystem as excess crypto wealth can be put into productive uses. It also helps arbitrage between stronger and weaker debt capital markets globally.

Although the loan volumes in these protocols are currently low (c. $250m as of 3 Feb 23), this is a space to watch.

So what have we learned so far?

A number of things should have become clear by now:

We saw above that DeFi has found category-market fit over 2021–22; I strongly believe that it is here to stay and will reach volumes that far exceed the Jan-22 peak.
DeFi has started to build lending products for both corporates and consumers; consumer debt volumes are multiple times larger than corporate ones, even if product variety is much more limited for consumers than for corporates.
Unsecured (under-collateralized) loans remain an untapped market opportunity for consumers; corporates already have some access to them, but the untapped opportunities remain enormous.
RWA is an attractive category that could be a trojan horse for DeFi to penetrate deeper into TradFi territory, and thereby attract significantly more capital into the on-chain world.

The on-chain credit opportunity

We saw above that there is a large delta between unsecured debt in DeFi versus TradFi. In the latter, it accounts for 59% of all outstanding consumer and corporate debt. In the former, it accounts for <1% of all loans (as of 3 Feb 23; comparing total value locked in all unsecured protocols versus top-10 secured protocols).

There is a reason for this. But there are also projects working on unlocking this opportunity. Let us take a look.

Lack of trust and credit make unsecured debt challenging

DeFi protocols are trustless, permissionless and anonymous. It means that any wallet can connect to any protocol and start using it immediately, without any gatekeepers in place.

Wallets are anonymous, and there is no link to real-world identities. In fact, one person can have a plethora of wallets. If a protocol were to provide an unsecured loan to a wallet, the borrower would have an incentive to run away with the money. As their identity is unknown, they would have no fear of legal repercussions in the off-chain life. Their wallet might be blacklisted, but they can always abandon it and spin up another one.

The (inefficient) work-around in DeFi is that for anyone to obtain a loan, they have to come up with 3x the amount of that loan in collateral. This is a clear disincentive to run away with the loan, but also limits the utility of on-chain debt.

The challenge is therefore to build a protocol that adheres to the values and efficiency of DeFi, but with the incentives and structures in place to prevent fraud.

Startups building on-chain credit through identity

One, arguably very TradFi, approach to unlocking unsecured on-chain debt would be to link on-chain to off-chain identities in such a way that behaviour in the former has a real impact on the latter. Defrauding a party in DeFi would lead to off-chain prosecution. Defaulting on a USDC loan would drag down the off-chain credit score. On the other hand, repaying a USDC loan in time could also lift the off-chain credit score and thereby help a person build credit in situations where it might be tough to do so otherwise (think of the un- or underbanked).

The decentralized identity layer would enable wallets to remain anonymous, but still have a provable link to an off-chain identity. It would enable protocols (anonymously through zk-proofs) to determine a wallet’s creditworthiness based on on- or off-chain data or a combination of both. These credit scores would then be referenced by lending protocols before entering into a relationship with a borrower. Those with acceptable scores would be able to borrow money in an under- or even fully un-secured fashion.

This would open up the market to those that do not (yet) have crypto assets (which is most of the world). It would also open up the market to new use cases apart from leveraged trading, as DeFi loans would compete head-on with TradFi use cases. In countries where local debt capital markets are thin (e.g. emerging markets), it would enable borrowers to profit from a global pool of lenders, and thereby receive better loan terms.

The good news: a number of startups are working on this

This is one possible path forward, but a polarising one. The idea of linking on- and off-chain identities or introducing gatekeepers to protocols are red flags for many within DeFi.

I am not a purist. I believe we should use the potential of DeFi to benefit everybody, and especially those that are being failed by TradFi institutions. We should think about how we can bring DeFi to the masses, and build products not for the sake of the product, but because they offer tangible advantages over TradFi. If that means making a few (albeit crucial) trade-offs, I believe this to be a price worth paying.

I am rooting for the aforementioned companies and hope that in a few years time we will be taking out on-chain loans to finance our off-chain activities. I hope that this process will be smoother, quicker, cheaper and more transparent than what TradFi institutions are currently offering.

If you share this vision and are building a product that enables mainstream adoption of DeFi credit and undercollateralized lending, I would love to speak to you.

Thanks to everyone who has provided me with feedback on this, especially to Darshan (of Credora) and Julian (of Cred Protocol) for their valuable input.

References

Consumer loans:

Corporate loans:

*https://www.federalreserve.gov/publications/2022-november-financial-stability-report-borrowing.htm