From the 2008 Financial Crisis to the Re-pledging Market: Potential Crisis Under Insufficient Liquidity and Leverage Risks

Original Author: Possibility Result

Original Compilation: Deep Tide TechFlow

Introduction

As ETH staking yields drop to around 3%, investors are turning to a tokenized restaking pool known as Liquid Restaking Tokens (LRTs) to enhance ETH-denominated returns. Consequently, the value within LRTs has soared to $10 billion. The primary driver of this trend is approximately $2.3 billion used as collateral for leveraged operations. However, this strategy is not without risks. Each component of LRTs carries its unique risks that are difficult to model, and the on-chain liquidity is insufficient to support effective liquidation during large-scale deleveraging events.

As Ethereum (ETH) staking yields fall to about 3%, investors are beginning to turn to a tokenized pool called Liquid Restaking Tokens (LRTs) in search of higher Ethereum returns. As a result, the value within LRTs has surged to $100 billion. This trend is primarily driven by around $23 billion in collateral used for leveraged operations. However, this also comes with risks. The various positions within LRTs have unpredictable idiosyncratic risks, and the lack of on-chain liquidity makes effective liquidation during large-scale deleveraging events challenging.

The current situation with LRTs bears some resemblance to the conditions before the 2008 financial crisis. In 2003, the federal funds rate fell to its lowest point in 50 years at 1%. In pursuit of higher dollar returns, investors flocked to the U.S. real estate market. Due to the lack of liquidity in individual mortgages, financial engineers packaged them into mortgage-backed securities (MBSs). The core issue of the 2008 collapse was excessive leverage and the lack of liquidity in MBSs, which, like LRTs, contained unpredictable idiosyncratic risks. When poor mortgage practices led to increased defaults, the cascading liquidations, panic, and liquidity shortages triggered a severe global recession.

Given these similarities, we should reflect and attempt to answer: What can we learn from past lessons?

Brief History of 2008

(Note: There is much more that is not mentioned here, but to keep on topic, I have selected some of the most relevant points to our story.)

A simplified narrative leading to the 2008 economic recession is as follows:

Incentives of Loan Originators and Securitizers

The increased demand for mortgage-backed securities (MBSs) naturally incentivized an increase in mortgage supply. Thus, the "originate and distribute" model became increasingly popular. This allowed mortgage institutions (originators) to quickly transfer default risk to securitizers, who then transferred it to traders (distributors) seeking higher yields. By transferring risk, the process of originating loans became more scalable, as they could rapidly originate and sell mortgage debt without a massive balance sheet and effective risk management.

Here lies our first principal-agent problem: since mortgage originators do not need to bear the risk of the loans they issue, they have the means and motivation to issue more mortgages with almost no risk. The result of this incentive mechanism was the emergence of a particularly poor type of mortgage known as "loans designed to default."

Incentives of Rating Agencies

However, in addition to mortgage originators and securitizers, rating agencies also supported these seemingly stable sources of income. The involvement of rating agencies played a significant role. Depending on the structure of each specific mortgage-backed security (MBS), rating agencies were responsible for assessing which securities were high quality (AAA) and which were high risk (B-rated and below). The participation of rating agencies accelerated the onset of the financial crisis in two ways:

• The fees for rating agencies were paid by the institutions responsible for packaging and securitizing the mortgages. This conflict of interest led rating agencies to compete for more business by lowering rating standards. For example, the rating agency Fitch nearly lost all of its MBS rating business because it granted fewer AAA ratings.

• The risk models at the time were flawed, particularly in that they incorrectly assumed that the default risks of different mortgages were independent. As a result, securitizers could grade MBSs based on risk (the riskiest portions would bear X% of losses first in the event of default) to create collateralized debt obligations (CDOs). The lowest-risk portions were more likely to receive AAA ratings, while the highest-risk portions could be repackaged, re-rated, and rated again. The top portions of these new CDOs often received AAA ratings again (it is important to note that default probabilities are not independent).

Over-Leverage

In 1988, the Basel I Capital Accord was approved, which established capital requirements for internationally active banks. The so-called capital requirements refer to how much capital banks must reserve for every dollar of "risk-weighted" assets they hold. Simply put, this effectively capped the maximum leverage ratio for banks at 12.5:1. If you are familiar with cryptocurrency lending protocols, you can think of risk-weighted capital requirements as similar to loan-to-value ratios for different assets. However, "risk-weighted" is not always used to reduce risk; it is sometimes used as a tool to encourage banks to pursue other objectives.

To incentivize banks to finance housing mortgages, securities related to housing mortgages were set at half the risk of commercial loans (50%), meaning banks could use double the leverage (25:1). By 2007, Basel II further reduced the risk weight of AAA-rated mortgage-backed securities (MBS), allowing banks to increase their leverage ratio to 62.5:1 (note: lower-rated MBS had even lower leverage ratios) (Government Accountability Office report on mortgage-related assets).

Despite capital requirements, banks achieved "rating and regulatory arbitrage" through special investment vehicles (SIVs), circumventing further leverage restrictions. A SIV is an independent legal entity "sponsored" by a bank, but it has its own balance sheet. Although SIVs themselves had little credit history, they could still borrow at lower rates to purchase assets because the market generally believed that the "sponsoring" bank would provide support in the event of losses. In reality, banks and these off-balance-sheet SIVs were almost one and the same.

For a long time, banks did not need to meet any capital requirements for SIVs' debts. It wasn't until Enron hid its debts in carefully designed off-balance-sheet vehicles to support its stock price, ultimately leading to its collapse, that regulators began to re-examine this issue. However, despite this, there were no substantial regulatory changes—SIVs still only needed to comply with their sponsoring bank's 10% capital requirement. In terms of leverage ratios, banks could still use a leverage of 625:1 on AAA-rated mortgage-backed securities (MBS) through SIVs. (Note: this does not mean banks would necessarily maximize leverage or only hold MBS; it just means they had that capability).

Thus, SIVs quickly became the primary channel for funding mortgages in the global financial system (Tooze 60).

Complexity Leading to Opacity

From this, we can also learn an important lesson about complexity. Finance is not simple; at its core, some participants are better at assessing and bearing risks than others. Evaluating a government bond in isolation is relatively straightforward. A single mortgage is somewhat more complex but still within reasonable bounds. But what if faced with a pool of mortgages based on complex assumptions? Or a risk tranching based on even more assumptions? Or a pool of mortgages that has been repackaged and tranched multiple times? This is undoubtedly mind-boggling.

In these complex packaging and tranching processes, many people choose to delegate the risk assessment work to the "market," rather than conducting detailed due diligence on these derivatives.

The pursuit of complexity in the derivatives market is driven by significant incentives, often benefiting savvy investors while disadvantaging the inexperienced. When financial engineer and Goldman Sachs employee Fabrice "fabulous Fab" Tourre was asked who would buy their synthetic CDOs, his response was: "Belgian widows and orphans" (Blinder 78).

However, the narrative of "Wall Street greed!" oversimplifies the situation. In fact, the AAA-rated bonds issued from 2004 to 2007 (the peak of market exuberance) did not suffer severe losses—by 2011, cumulative losses were only 17 basis points—yet the global market experienced an unprecedented collapse. This indicates that excessive leverage and poor collateral may not be the only reasons.

In the book “The Credit Crisis”, Gorton and Ordonez propose that when the cost of disclosing collateral quality information is high, even routine market fluctuations can trigger an economic recession. The model shows that as the market goes a long time without significant shocks, lenders reduce the information costs used for ratings. Consequently, borrowers with low-quality and high-rating-cost collateral gradually enter the market (e.g., subprime mortgage-backed securities held in SIVs). As ratings decline, borrowing costs decrease, leading to increased market activity, as borrowers can obtain collateral at lower costs. However, when the value of certain high-risk collateral experiences a slight decline, creditors may reconsider paying the rating costs for assessment. As a result, lenders begin to avoid those high-rating-cost collaterals, even if their quality is not poor. This credit tightening could lead to a significant decline in market activity (Gorton and Ordonez).

Similarities Between MBS (Mortgage-Backed Securities) and LRT (Liquid Restaking Tokens)

The demand for secure ETH yields in the crypto market (especially Ethereum) is akin to the pursuit of secure dollar yields in traditional finance. Similar to the dollar yields on government bonds in 2003, the yields from ETH staking are also gradually compressing. With approximately 30% of the ETH supply staked, the current yield has dropped to around 3%.

Like the mortgage-backed securities (MBS) of 2008, the decline in staking yields has prompted the market to seek higher-risk investment opportunities for greater returns. This analogy is not new. Particularly in the article by Alex Evans and Tarun Chitra titled “What PoS and DeFi Can Learn from Mortgage-Backed Securities,” they compare liquid staking tokens (LSTs) to MBS. This article discusses how LSTs help stakers earn both staking yields that secure network safety and DeFi yields, avoiding competition between the two. Since then, LST holders have primarily leveraged their positions by using them as collateral for loans.

However, the relationship between MBS (mortgage-backed securities) and Liquid Restaking Tokens (LRTs) seems more complex.

While LSTs like stETH aggregate validators with relatively homogeneous risks (as they validate relatively stable protocols), the restaking market is entirely different. Restaking protocols facilitate staking across various active validation services (AVS) while aggregating. To incentivize user deposits, these AVS pay fees to stakers and operators. Compared to regular ETH staking, the opportunities for ETH restaking are limitless—yet this may also introduce unique risks (e.g., unique penalty conditions).

With higher yields, the risk-seeking crypto market has flocked to deposits, with a total locked value (TVL) of approximately $14 billion at the time of writing. In this growth, Liquid Restaking Tokens (LRTs) occupy a significant share (around $10 billion), tokenizing shares in the restaking position pool.

On one hand, regular ETH staking yields feel like "government-issued and supported." For instance, most stakers might assume that in the event of a significant consensus error leading to large-scale penalties, Ethereum will undergo a hard fork.

On the other hand, restaking yields can come from any source. They cannot rely on issuing ETH within the protocol to incentivize ongoing security. If there are flaws in the implementation of custom penalty conditions, Ethereum's hard fork could spark further controversy. If the situation is dire enough, we might see whether a hard fork resulting from the DAO hack leads to any moral hazard, akin to bank bailouts, where banks are deemed "too big to fail," otherwise posing systemic risks to the global financial system.

The incentives of LRT issuers and ETH restakers are similar to those of mortgage securitizers and banks seeking higher yields. Therefore, we may see that loans designed to default in the crypto space may not only emerge but could become commonplace. A specific type of default loan is known as a NINJA loan, as borrowers have No Income, No Job, and No Assets. In restaking, this phenomenon manifests as low-quality active validation services (AVS) obtaining large amounts of LRT collateral to gain short-term yields provided by token inflation. As we will discuss in subsequent sections, if this occurs on a large scale, there will be significant risks.

Actual Risks

The most significant financial risk is the occurrence of a slashing event, causing the value of LRTs to fall below the liquidation thresholds of various credit protocols. Such events will lead to the liquidation of LRTs and may significantly impact the prices of related assets, as the assets within LRTs will be unlocked and sold for more stable assets. If the initial liquidation event is large enough, it could trigger a cascading liquidation of other assets.

I can think of two possible scenarios that could make this situation a reality:

• Vulnerabilities in newly implemented penalty conditions. New protocols will have new penalty conditions, meaning new vulnerabilities affecting a large number of operators may arise. If "designed to default" active validation services (AVS) become very common, the likelihood of this outcome is high. That is to say, the scale of the penalty event is also very important. Currently, AAVE (at the time of writing, its LRT collateral is $2.2 billion) has a liquidation threshold of 95% for borrowing ETH collateral weETH (the most popular LRT)—this means that exploiting a vulnerability would need to lead to over 5% of the collateral being affected by a penalty event to initiate the first wave of liquidation.

• Social engineering attacks. Attackers (whether protocol or operators) can persuade various LRTs to invest capital with them. Afterward, they would establish a large short position on LRTs (possibly including ETH and other derivatives). Since this capital does not belong to them, they have little risk other than reputational risk. If the builders or operators do not care about their social reputation (perhaps because they are pseudonymous), and the profits from short positions and attack bounties are substantial enough, they should be able to make considerable profits.

Of course, all these scenarios are only possible if the penalty mechanism is enabled—which is not always the case. However, before the penalty mechanism is activated, the benefits of restaking for the protocol's economic security are minimal, so we should be prepared for the risks of penalties.

Avoiding Past Mistakes

So the biggest question remains… What can we learn from the past?

Incentive Mechanisms Matter

Currently, the competition among liquid restaking tokens primarily focuses on providing the highest ETH-denominated yields. Similar to the increased demand for high-risk mortgages, we will see a demand for high-risk active validation services (AVS)—I believe this is where most of the penalty (and liquidation) risks lie. Individual high-risk assets are not particularly concerning, but when they are used to take on excessive leverage without sufficient liquidity, they become problematic.

To limit excessive leverage, lending protocols set supply caps, which determine how much of a specific asset the protocol can accept as collateral. The supply cap largely depends on the available liquidity. If liquidity is scarce, it will be more challenging for liquidators to convert the liquidated collateral into stablecoins.

Similar to how banks take on excessive leverage to increase the nominal value of their portfolios, lending protocols may have significant incentives to violate best practices to support more leverage. While we hope the market can completely avoid this situation, history, such as events in 2008, tells us that when people face profit promises and the costs of information disclosure are high, they often tend to delegate (or completely ignore) due diligence.

Learning from past mistakes (e.g., the incentives of rating agencies) tells us that building an unbiased third party to help assess and coordinate the risks of different collateral types and lending protocols would be very helpful—especially for liquid restaking (LRT) and its secured protocols. They should utilize their risk assessments to propose safe, industry-wide liquidation thresholds and supply cap recommendations. The degree to which protocols deviate from these recommendations should be made public for monitoring. Ideally, this organization should not be funded by those who might benefit from high-risk parameters but rather by those who wish to make informed decisions. Perhaps this could be a crowdsourced initiative, a grant from the Ethereum Foundation, or a profitable "come for the tools, stay for the network" project serving individual lenders and borrowers.

With support from the Ethereum Foundation, L2 Beat has done well in managing similar initiatives for Layer 2. Therefore, I hold some hope that similar efforts could succeed in restaking—such as Gauntlet (funded by the Eigenlayer Foundation), which seems to have already started, although there is currently no information on leverage. However, even if such projects are established, the possibility of completely eliminating risk is slim, but at least it can reduce the cost of information acquisition for market participants.

This also leads to a second relevant point.

Model Inadequacies and Liquidity Shortages

We previously discussed how rating agencies and mortgage securitizers severely overestimated the independence of mortgage defaults. The lesson we learned is that a decline in housing prices in one region of the U.S. can significantly impact housing prices in other regions, not only elsewhere in the U.S. but also globally.

Why?

Because a small number of large participants provide most of the liquidity for global economic activity, and these participants also hold mortgage-backed securities (MBS). When poor mortgage practices lead to a decline in MBS prices, these large participants' ability to provide liquidity to the market also diminishes. As assets need to be sold in a less liquid market to repay loans, prices across the board (whether related to mortgages or not) also decline.

A similar overestimation of "shared" liquidity may inadvertently arise in the parameter settings of lending protocols. The setting of supply caps aims to ensure that collateral in the protocol can be liquidated without leading to bankruptcy. However, liquidity is a shared resource that every credit protocol relies on to ensure solvency during liquidation. If a protocol sets its supply cap based on liquidity at a particular moment, other protocols can make their supply cap decisions one by one, causing each previous assumption about available liquidity to lose accuracy. Therefore, lending protocols should avoid making independent decisions (unless they do not have priority access to liquidity).

Unfortunately, if liquidity is accessible to anyone at any time without permission, protocols will find it challenging to set parameters safely. However, if priority access to liquidity can be granted in certain situations, this uncertainty has a solution. For example, the spot market for assets used as collateral could set a hook that queries the lending protocol whenever a swap occurs to check if liquidation is possible. If liquidation is underway, the market should only allow asset sales triggered by messages from the lending protocol itself. This functionality could enable lending protocols to set supply caps with greater confidence by collaborating with exchanges.

Case Study:

We may already have a case study to observe the development of the LRT market.

AAVE offers over $2.2 billion in weETH collateral on-chain, but according to Gauntlet's dashboard, the on-chain liquidity for exit paths to wstETH, wETH, or rETH is only $37 million (this does not even account for slippage or USDC exits, making the actual liquidity worse).

As other lending protocols begin to accept weETH collateral (for example, Spark currently has over $150 million in total locked weETH), competition for limited liquidity will intensify.

The liquidation threshold for ETH loans collateralized by weETH is 95%, meaning that a liquidation event affecting over 5% of the LRT collateral should be sufficient to trigger the first wave of liquidation. Therefore, hundreds of millions to billions of dollars of selling pressure will flood the market. This will almost certainly lead to selling pressure on wstETH and ETH, as liquidators convert assets into USDC, thereby risking subsequent waves of liquidations on ETH and related assets. But as mentioned earlier, as long as no liquidation occurs, the risk is minimal. Therefore, deposits in AAVE and other credit protocols should currently be safe and not face liquidation risks.

Key Differences

It would be inappropriate to write an entire article about the similarities between LRTs and MBSs (as well as today's cryptocurrencies and the financial system before 2008) without discussing some key differences. While this article conveys some similarities between MBS and LRT, they clearly have differences.

One of the most important distinctions is the openness, over-collateralization, algorithm-driven nature, and transparency of on-chain leverage compared to the characteristics of leverage in banks and shadow banking. The inefficiency of over-collateralized capital brings some significant advantages. For example, if a borrower defaults (and there is sufficient liquidity), lenders should always expect to recover their funds—this is not the case for under-collateralized loans. Their open and algorithm-driven characteristics also allow for immediate liquidation of assets and enable anyone to participate in the liquidation. Therefore, untrustworthy custodians and nefarious counterparties cannot engage in harmful practices such as delaying liquidation, executing liquidations at prices below their value, or re-pledging collateral without consent.

Transparency is a significant advantage. On-chain information about protocol balances and collateral quality is verifiable by anyone. In the context of the previously discussed research by Gorton and Ordonez, we can say that DeFi operates in an environment where the cost of assessing collateral quality is lower. Therefore, the cost of revealing collateral quality information should be lower, allowing the market to adjust more frequently and at lower costs. In practice, this means that lending protocols and users have richer information resources to make decisions on critical parameter selections. However, it is worth noting that for restaking, there are still some subjective off-chain factors, such as code quality and team background, where the cost of acquiring this information is higher.

An anecdotal sign is that since the collapses of companies like BlockFi and Celsius, on-chain lending activity seems to have increased. Notably, we see significant growth in deposits in AAVE and Morpho, but there has been almost no occurrence of off-chain lending operations on a scale comparable to previous cycles. However, obtaining specific data on the current scale of the off-chain lending market is not easy—this suggests there may be significant but underreported growth. Unless there is a direct hack of a lending protocol, all else being equal, based on the reasons mentioned above, the implementation of on-chain leverage should be more resilient.

As the risks of LRTs being slashed increase, we may once again witness a prime opportunity to see the advantages and disadvantages of transparent, over-collateralized, open, and algorithm-driven lending in practice. Ultimately, perhaps the biggest difference is that if something unexpected occurs, we do not have a government to bail us out. For lenders, there is no government support or Keynesian token economics. Only code, its state, and changes in that state. Therefore, we should strive to avoid unnecessary mistakes.