← Back to research

Defi, are we correctly valuing the systemic risks?

Aug 2, 2023


8 min read



To begin, I should clarify that my position on potential for Defi is overwhelmingly positive. As someone with a TradFi trading background, Defi has solved many real world problems i've experienced and given me access to sophisticated products that would previously have only been available to the most active of institutional investors (whether that is, in fact, a good thing could be a separate article in itself).

In 2021/22 I was yield farming with everyone. Collecting 50,000% APY from numerous protocols. Frequently getting rugged and not caring because i'd make it back in interest in a few hours. Life was great. But of course, it was never going to last.

In an environment where the excessive gains have diminished significantly and prices have followed, one could be forgiven for thinking that Defi is now in a more sustainable place, more secure, more mature. In contrast however, it feels as if the potential shortcomings are more apparent than ever.

The recent Curve hack underlines this. More on that later.


A brief aside, but let's start with Defi's favourite metric, TVL, or total value locked. This is the total amount of assets locked in Defi smart contracts by their market value. It's a metric that's been used to measure the growth of the space and is often used as a proxy for the success of a protocol.

Everyone has their opinion. Mine is that TVL is a weak metric. To name a few of my major reservations:-

  • Lack of standardisation (lends? borrows? yes? no? what?)
  • Double counting (or triple counting, or quadruple counting or...)
  • Asset quality (remember when $TURBO was worth almost $100mln?)
  • Huge liquid staking skew (Lido 👀)

But, this is not supposed to be a dismantling of TVL, so why does it matter in the context of this article? From personal experience I've seen it frequently used as a mental denominator with respect to Defi, be that liquidity, market risk, protocol risk, hack exposure, and so on. And that's a dangerous place to be.

"Did you see X got hacked for 2 mil? No big deal, it's TVL is 120mil - it'll be fine."

So as you digest this short article, get that $40bln TVL number out of your head. The true benchmark for actual market size and liquidity is multiples lower.

Systemic Risk

With that in mind, let's think about systemic risk. So, what is systemic risk? In short, it's the risk that seemingly isolated events have the potential to cause the entire system to collapse. Almost everything has some degree of systemic risk, but in the context of Defi, where composability is a key selling point, it's far more prominent than most.

In crypto, we love to throw around various advanced technical and economic terms to make ourselves sound smart, irrespective of whether we truly understand them. Basing the stability of our protocols purely on rational economic actors is a dangerous game. Unless you believe engineers make the best economists or vice versa, this should raise some concern. Whilst I admire the ambition, it undoubtedly leads to a false sense of security in some cases.

I mean, of all people, should those of us in crypto really be relying on rational economic actors? 95%+ of the activity in the space is wholly irrational, in the hope that others may act in an even more irrational manner. $SHIB? $ELON? $FLOKI? Cryptodickbutts. I could go on. Forever. Think of $GME and WallStreetBets which is a good proxy. In that case, the hedge funds were initially acting in a reasonably rational manner - and they got completely wrecked by the meme.

Markets can stay irrational longer than you can stay solvent." -John Maynard Keynes, 1936

Curve is a timely example of what i'm trying to illustrate here.

Note:- At the time of writing, some OTC sales of $CRV that have helped the repayment of loans (although a majority are still outstanding with high interest rates). I expect this will work itself out over the next couple of days with minimal drama, but its still a valuable case study.

A lot has been written about the hack, but to briefly summarise, at no fault of the Curve protocol, a bug was discovered in the compilers of certain versions of Vyper that failed to correctly protect against re-entrancy and allowed an attacker to drain a handful liquidity pools, including the $CRV/$ETH pool.

Meanwhile, the founder of Curve, and the largest holder of $CRV by a mile, Michael Egorov, is lending out almost half of the circulating supply of $CRV, borrowing stables against it.

So let's just think about this for a second, in simple terms.

  • A protocol has been hacked, causing the price of its token to fall sharply.
  • At the same time, a hacker has walked off with a bunch of the protocol's native token as part of the hack, but has yet to sell.
  • A large holder of the token is using a highly illiquid quantity of the token as collateral to borrow stable coins from a handful of protocols.

What could possibly go wrong? Well, everything...

The demand to lend against $CRV on money markets collapses, causing interest rates to spike. In order to either pay down those loans or service the interest, the holder liquidates $CRV causing further downward pressure on the price. As the price falls, the safety margin above the liquidation threshold decreases. And remember, this is one individual and that's a key point. There's no gradual drip feed of liquidations. On each protocol, it's one liquidation threshold. On Aave for example, that's $63mil. If we get anywhere near that, $CRV is broadly worthless, at least in the short-term. Now money markets are under-collateralised. And then what...?

So here we have a perfect example of systemic risk. You can see why the current structure of Defi can be seen as inherently dangerous. And we should remember that in the wider scheme of things, this was a relatively small, isolated hack. It could have been a lot worse.

Misaligned Incentives

"Decentralized finance represents a revolutionary shift in the financial landscape, giving power back to the individual."
-everyone, 2021

A further point less widely discussed is the total misalignment of incentives. A lot of shade is being thrown at Michael E right now, but isn't his behaviour somewhat rational? And that's what we're all about in crypto right? Rational actors. Right? Righttttt? In fact, you could argue that attempting to pay down any of the loans was his first irrational action.

We all know founders can't be seen to be selling their own token. If anything is to signal the near immediate death of a protocol, it's that. But using it as collateral to take out huge loans goes widely unnoticed. And when it is noticed, barely an eyebrow is raised. What would you do if you had the choice of $60mil in cash, or $120mil on paper. Are you telling me you wouldn't take anything off the table?

So what Aave and the other Defi protocols have done, is effectively sold individuals like Michael E a massively underpriced put option (I already hear people saying it isn't underpriced, there's a haircut, interest, etc. - but trust me, given the true liquidity of $CRV, its underpriced - especially when exiting the position at no loss ex-interest can be done at any point up until default).

The Future

I appreciate this sounds negative. There's so much potential here. But we need to start thinking about risk in crypto differently. The competition between protocols to attract customers has come with a disregard to the real risks that exist.

Take $CRV for example again. If I sell just $1mil of it today for stables on Curve or Uniswap, I get less than $600k. Thats 40% slippage. And yet i'm able to use this token as collateral to borrow over $100mil?

We need to stop trusting people just because they sound clever, and instead look at the bare facts. We're socialising risks without knowing it. You use Defi, you've indirectly underwritten Michael E's risk.

That said, there are reasons to be optimistic. The recent events have opened a lot of eyes and from here I expect there will be increased focus on reassessing risk. Furthermore, Defi growth has turned a corner, particularly in areas such as liquid staking. Restaking is an interesting area and may offer a positive path forward for improved risk capture without significant market consequences in negative scenarios (staked ETH restaked to underwrite protocol risk anyone?).

In addition, the L2 space is seeing continued differentiation on the protocol front where we're finally getting away from forks of forks. We're also seeing a revival in the alt-1 space, where Solana traction looks set to boom once again and Avalanche is exploring interesting areas in the tokenisation/securitisation front.

There is a lot to be optimistic about.

← Back to research

© 2024 Perfect Dimension Studios Ltd - All Rights Reserved