Can we optimize the design of the "liquidity mechanism" to avoid the snowball effect caused by bank runs?

This article is authorized to be reprinted from pupupupuisland, opinions do not represent a position

In early November 2022, the founder of the largest cryptocurrency exchange Binance, CZ, raised security concerns about the second-largest exchange FTX, leading a massive sell-off of FTX's platform token. This move successfully caused market panic, resulting in FTX users withdrawing $6 billion within three days, triggering a liquidity crisis, withdrawal suspension, and ultimately leading to FTX declaring bankruptcy.

Handling such a large volume of withdrawals in a short period would be challenging for any bank or exchange. These days, social media is filled with panic and criticism; at this point, we cannot change what has already happened, but perhaps we can think about how to prevent such issues in the future.

Advertisement - Continue scrolling for more content

If a mechanism can be found to improve the operation of the blockchain economy to be safer and more efficient while minimizing the impact of runs, perhaps in the future, experimental implementation could be conducted through community and shareholder voting. Isn't this also a significant value among those "tokens"?

Water = Money, Water Tank = Exchange, Water Flow = Liquidity

If we want to encourage more public participation and discussion, the first thing we need to do is to simplify and popularize everything. Therefore, we will try to explain the concepts and operational principles using plain language and illustrations: Water is like money, the water tank is like a bank or exchange, and the flow of water in and out of the water inlet represents liquidity.

I. Why Do We Need a Water Tank?

Before we begin, let's first consider why banks or exchanges exist. If each of us uses our own wallet, and every transaction requires finding both buyers and sellers, and each function requires different platforms, it would be quite inconvenient and costly, posing many risks.

It's like everyone carrying their own cup of water, having to go to different places, find different people, and trade water back and forth with spoons. An exchange acts as a large water tank where everyone can pour their water in, and the water tank manager keeps a record in the ledger of who has how much money, who transferred money to whom, and who is willing to buy how much water with how much money, making it easier for everyone to engage in various investment and trading activities.

II. Building a Water Tank Also Requires Money

However, building a water tank also incurs costs and risks. Otherwise, who would bother to build a water tank, manage water for everyone, store water, and return it to users with interest?

The fact that needs to be clear is that all banks or exchanges cannot solely rely on low fees to operate and profit without utilizing user assets. In the traditional financial system, whether it's the government, banks, or insurance companies, they all use "a portion" of everyone's money for various types of investments to maintain a balance between assets and liabilities. Keeping water locked in a completely sealed water tank will not magically generate more water, and building, operating, staffing, managing, marketing, and upgrading the water tank all require money.

III. Separate "Custodial Water Tanks" and "Trading Water Tanks"

The water in each person's tank will also be divided into different purposes. Some may just want a safe place for custody, while others may want convenience for various transactions. We can simplify this into two categories: "Custodial Water Tanks" and "Trading Water Tanks":

The "Custodial Water Tank" only provides custody services and cannot perform any other transactions. Therefore, users need to pay a custody fee in advance when storing water, and they can safely withdraw it at any time. This water will not be used for any investments or transactions, so theoretically, it should not generate any interest.

The functionality of the "Trading Water Tank" will be more diverse, allowing temporary storage, trading, investment in other places, etc., with relatively higher risks. The main source of income for exchanges is through service fees from the trading water tank, so they may offer higher interest rates to encourage users to use this facility.

The operation of these two types of water tanks and the water inside them should be completely separate and not interchangeable.

IV. A More Structured "Trading Water Tank"

In addition to the basic categorization of the two water tanks, handling such complex inflows and outflows and water allocation in the "Trading Water Tank" should not be a "one-tank-fits-all" model. It may require different-purpose water tanks, some for convenient trading access, some for investment elsewhere, some as reserves, etc.

Having a more structured water tank compared to mixing all the water in one tank will lead to better management and risk control, allowing users to feel more confident in depositing their water. The issues faced by FTX this time were due to mixing user reserve funds, investment funds, and company assets, causing complications.

V. The "Transparency" of Reserve Tanks

With a structured reserve tank in place, the next consideration is how to store reserve water and ensure the safety of the reserve water. Historically, traditional financial institutions such as governments and banks could secretly alter their accounts or print more currency without most people knowing. In contrast, blockchain technology allows rules to be directly written into smart contracts, making alterations visible on the chain. The "user reserve funds" in the reserve tank should also be stored on a completely transparent and public blockchain network.

Therefore, users should only "see" three things: the total water volume, the proportions of various-purpose tanks, and the water in the reserve tank. Other business operations should, of course, maintain a certain level of confidentiality.

If not actual "transparency" but rather outsourcing the "guarantee" work to a third-party insurance company or legal team, it could also lead to resignations or bankruptcy. Even if legal procedures are eventually pursued, it could take several years of lengthy litigation, involving many complex issues such as liquidation, claims, cross-border matters, collective lawsuits, etc.

After the recent run on, many small exchanges and financial service providers hurriedly issued press releases, claiming that their reserves were "fine," but without any actual proof on the chain, such practices are quite questionable.

VI. The Safety of Reserves Depends on "Proportion" Rather Than "Amount"

With a transparent reserve tank, users can see the actual reserve water, but what amount of reserves is considered "safe"? After the FTX incident, Binance quickly released proof of reserve assets, but users could only see a "large sum," without knowing what proportion it actually occupies in the assets users have deposited. The safety of POR reserve proof should be based on "proportion" rather than "amount"; 1 liter of reserve water is 100% safe for a water tank of 1 liter, but only 10% safe for a tank of 10 liters.

If the reserve assets are denominated in the "market value" of a certain unstable coin, when a large amount needs to be sold for redemption into other currencies in the near future, it will inevitably become "cheaper" as more is sold, causing the actual market value of those amounts to shrink. Several other points to consider are:

1. Tokens issued by the exchange should not be included as reserve assets
2. Reserve ratios for various coins should be calculated separately
3. Downside risk should also be factored in
4. Avoiding excessive exposure to single or related ecosystem coins
5. The time required for liquidating reserve assets
6. Whether additional costs are incurred during large liquidations
7. Who controls the withdrawal of reserve assets
8. Risks such as private key loss or hacking of these assets

VII. How Should the "Safe Ratio" of Circulating and Reserve Water be Allocated?

What do you think the respective proportions of safe reserve water that cannot be moved, circulating water for trading, and water for investments elsewhere should be within a large exchange's water tank?

Each exchange and user will have different perspectives and answers. With transparency in place, the choices made by users will help find a reasonable answer (market price).

The following is a completely speculative discussion with no actual reference basis. Each value, unit, and statistical method would need to be handled by professionals in the field. Here, we provide a framework and logic for consideration, hoping to inspire more innovative ideas.

The "internal" trading volume of an exchange, no matter how large, is just a numerical change in the internal ledger. Therefore, the inflow and outflow of the entire water tank should be calculated as "circulating volume." For example, if a user's total assets in an exchange are 10 billion, and the average daily circulating volume is 1 billion, it means that in a 100-liter water tank, the water level could drop to a minimum of 90 liters and rise to a maximum of 110 liters.

Furthermore, since the 10 billion may consist of various stable and unstable coins, different proportions of "stock levels" for different coins and different "price volatility ranges" need to be "weighted" to calculate the fluctuation range. For instance, after weighting the statistics, if the median or average is 4 times, the original normal 10 billion circulating volume might require an additional 3 times in circulation, increasing the maximum circulating volume to 40 billion.

The highest fluctuation volume needs to be allocated in both the trading water tank and the reserve water tank to act as a buffer during a run. Therefore, I might allocate the ratio of trading volume to reserve volume as 1+3:3. This means that in a total water tank of 100 liters, 40 liters would be placed in the trading water tank, 30 liters in the reserve water tank, and the remaining 30 liters used for operations or other investments.

VIII. Real-Time Updating of Water Tanks and Pipes

However, these values mentioned above will be continuously updated in real-time, so the results will not be static after one calculation. Therefore, the logic might resemble concepts from machine learning: designing a basic formula with a fixed structure and task goal that adjusts its values and weighted inputs automatically based on current and past data.

Also, applying the concept of quantitative trading, analyzing fluctuations from past data, when the computer detects potential deviations beyond the expected range, it will automatically adjust some values in the formula, correcting them to proportions that can function healthily in real-time. This way, complex statistical issues like trading volume, circulating volume, reserve volume, withdrawal volume, fund allocation, etc., will be preemptively managed by the computer to prevent issues rather than rectifying them manually after they occur.

IX. Who Has Control Over the Withdrawal Valve?

When a bank or exchange anticipates signs of a significant run, they can address the issue authoritatively by either temporarily suspending services or delaying withdrawal speeds. For example, cryptocurrency lending company Celsius had once suspended withdrawals and transfers citing "extreme market conditions" and "protecting and preserving customer assets"; or Binance also halted users from withdrawing their bitcoins due to a "stuck transaction."

But every time a problem arises, exchanges seem to unilaterally breach contracts, making arbitrary decisions on whether to retain (seize) user assets in the exchange and without any publicly disclosed rules or breach penalties. So, is it really a good idea to forcibly enforce closing withdrawals using a human-centric approach whenever there are significant withdrawal issues? How do we know whether claims of "protecting users" are genuinely safeguarding user assets or protecting the exchange's liquidity?

Is there a mechanism that can prevent unilateral control over the withdrawal valve by exchanges, protect user rights, avoid market panic snowball effects, and safeguard exchange liquidity?

X. How Can the Withdrawal Valve be Optimized?

When a water tank uses a fixed-sized outlet, the water flows out at the same rate regardless of the water level inside. However, by using a large outlet for high water levels, a medium outlet for medium levels, and a small outlet for low levels in the "Trading Water Tank," the outflow speed can be adjusted according to the water level.

This means that during times of high circulating and reserve volumes, faster withdrawals can be allowed to increase market efficiency. Conversely, if there is market panic with gradually increasing outflows and decreasing reserves, the withdrawal time can be lengthened. The purpose of the extended time is to provide a cooling-off period for short-term panic emotions and to allow the exchange more time to liquidate assets elsewhere to replenish the "Trading Water Tank" to maintain a safe level.

In the illustration, we simplified the concept by using three water levels and outlet widths to represent reserve levels and withdrawal speeds. However, in practice, it would be more like an automatically adjustable pipeline with linear and real-time changes in width.

Additionally, a "water level sensor valve" could be set between normal circulation, volatile circulation, and liquidation circulation. When the water level drops below normal circulation and enters volatile circulation, the assets in the "investment tank" should automatically begin to sell, redeem, or liquidate, gradually replenishing the water tank to the safe level of normal circulation as the water level in the trading tank decreases. Instead of waiting until the money is gone to start looking for funds elsewhere, when the replenishment speed is not fast enough, and the water level in the trading tank gradually drops from volatile circulation to liquidation circulation, assets from other places should be liquidated automatically, and the reserve tank should be prepared for scheduling and withdrawal processes.

Conclusion: "Liquidity Mechanisms" Need to be Programmed to Reduce "Human" Factors

For this mechanism to function and gain trust, the calculation formula must be transparently placed on the chain, audited and certified by a third party, and executed automatically through smart contracts. This means entrusting the "computation, execution, and constraints" to computers that operate without emotions. While a slower withdrawal speed may reduce individual users' willingness, a clear mechanism can bring more trust to the collective users. Sometimes, an overly efficient market can lead to market panic and snowball effects.

The disasters that occur are not due to code vulnerabilities or issues with blockchain technology but rather "human" misappropriation of assets, "human" creation of panic, and "human" panic-induced runs. Perhaps we should reconsider whether we should transfer some of the asset management power from authoritarian and greedy exchange bosses to a fairer and more just code.

There is No Perfect Model in the Market; We Must Continuously Improve the Water Tank

Although the examples discussed above seem simple and ideal, the actual operation of the market is quite complex, and there is no perfect model that always works. We may not have the ability to implement this system practically, or these concepts may have been tried and failed before. However, during these times, we need to inspire more disruptive innovative concepts.

Some key points for discussion in "mechanism design" are:

1. Asset Transparency: Users need to see that "user reserve funds" are securely stored on the chain and cannot be misappropriated

2. Proportion of Reserve Assets: Finding a balance between market operations and risk management

3. Withdrawal Control Rights: Withdrawal mechanisms and controls should be executed by code, not human intervention

4. Preventing Runs: More efficient and automated fund allocation methods

5. Mitigating Run Impacts: Should withdrawal speeds be adjusted based on liquidity?

We must reiterate that the author of this article has no relevant economic background, and the numerical examples are speculative. The purpose of this article is to invite more people to present different ideas. If any terminology, metaphors, or information in this article is incorrect or can be narrated better, please feel free to provide feedback, and we will promptly make corrections. The content of this article is entirely open-source and can be freely reproduced, modified, used for personal or commercial purposes without attribution.

If you were to design a water tank and liquidity mechanism, how would you approach it?

In the long run, perhaps the FTX incident will lead us to take a more serious look at the risks that centralized exchanges have always harbored and strengthen independent thinking and media literacy. The crypto world never lacks armchair quarterbacks and those quick to criticize after the fact, so instead of continuing to vent on social media, why not brainstorm on how to prevent such problems from recurring in the future.

Our art team's operating funds and the virtual currency assets of three members were also lost in FTX and are currently being reorganized. Looking back at the efforts of the past two years, although all assets have evaporated, at least the accumulated experience, knowledge, skills, and works will not disappear. We hope