The Ongoing Banking Fiasco
SVB was just the beginning
As for the broader financial system, it’s unlikely we’ll see anything catastrophic.
- SoBasically, March 11 2023
That aged well. What started as a startup-friendly commercial bank mismanaging their assets rapidly escalated into a global banking crisis. As we follow up from our last article on Silicon Valley Bank’s collapse, we’re going to provide a comprehensive timeline of how one crash simply led to another and pushed this situation well beyond the United States.
Checking in with SVB
After the California Department of Financial Protection and Innovation (DFPI) seized SVB on March 10th, the Federal Deposit Insurance Corporation (FDIC) took over and established the Deposit Insurance National Bank of Santa Clara to, as the name suggests, insure customer deposits.
The problem was that 89% of SVB’s depositor funds were uninsured as they exceeded the FDIC insurance limit of $250,000. This meant that SVB depositors could only receive $250,000 from the FDIC out of the potentially millions of dollars they kept with the bank. If that much money simply vanished into thin air, then not only would the FDIC’s reputation be irrevocably destroyed, but the American banking system as a whole would suffer a huge blow in credibility.
To avoid such a situation, the FDIC received an exception from the Treasury to pay off the full $172 billion to SVB customers. When all was said and done, SVB was reopened as a bridge bank (a temporary, state-owned entity to act in replacement of a failed bank), Silicon Valley Bridge Bank, N.A.
Silvergate Bank
Originally founded as a savings and loan association (a financial institution specializing in mortgages) in 1988, Silvergate Bank gradually moved into cryptocurrency in the early 2010s, going as far to set up a cryptocurrency payment system called the "Silvergate Exchange Network" (SEN). This basically enabled users to trade cryptocurrencies for fiat money such as the USD or GBP. The crypto industry actually seemed to work wonders for the bank’s growth, with Silvergate assets doubling in 2017 after servicing crypto exchanges and firms that couldn’t secure funding from larger banks (similar to how SVB would finance JPMorgan-rejected tech startups).
Fast forward to late 2022. FTX just failed, crypto prices are in freefall, and now Silvergate is sitting with the prospect of a bank run. Not only was FTX one of Silvergate’s biggest customers, but fearful investors withdrew $8.1 billion, creating a $1 billion loss in Q4 2022. If that doesn’t sound bad enough, Silvergate had nearly $12 billion in deposits from digital asset customers in Q3 2022. By the end of the year, those funds tanked to less than $4 billion.
Some analysts say that Silvergate’s business model was fundamentally unsustainable to begin with. Unlike most banks, Silvergate didn’t make money primarily from loans (sometimes referred to as credit exposure, which is the potential loss that a lender can incur if the borrower defaults) but rather, through securities. According to its balance sheet from September 2022, Silvergate held $11.4 billion in bonds and Treasury securities but only $1.4 billion in loans.
This makes the bank more reliant on deposits. Why? Not only do securities’ values fluctuate on a daily basis, but their interest rates are usually much lower than the interest rate on loans. This effectively reduces the bank’s revenue and makes it more reliant on deposits to fund its operations. On the note of security value fluctuation, loans retain their value through interest rate changes unlike bonds. So by investing their deposits mostly into securities - making them the primary source of liquidity - the bank’s liquidity is also much more likely to fluctuate.
This risk exposure can be softened if the bank has enough deposits coming in to make up for any potential liquidity loss. As previously stated, however, Silvergate experienced the precise opposite of that. To pay for the withdrawals, they had to sell $5.2 billion in debt securities at a $718 million loss and borrow $4.3 billion from the Federal Home Loan Bank of San Francisco.
Unlike SVB, however, what really killed Silvergate was their declining leverage ratio and how the market reacted to that. This essentially measures a bank’s ability to repay its financial obligations with its tier 1 capital (which includes disclosed reserves and common stock), and so a higher leverage ratio suggests a safer financial situation for the bank. For a bank to be considered “well capitalized,” it should have a leverage ratio of 5%. Silvergate’s leverage ratio in September 2022 was 10.45%, but that dropped to 5.12% in December 2022.
Technically, from a risk management standpoint, Silvergate should be just about fine. Even if the leverage ratio dropped below 5% (which was likely gonna happen anyway as they had to sell more bonds to repay the FHLB loan), they’d still be considered “adequately capitalized” as long as it was above 4%. But when a bank’s leverage ratio is halved in a few months, markets sound the alarm. Investors and customers interpret that as a sinking ship that should be avoided. And at this point, one could argue that it was only a matter of time before Silvergate shut down. A handful of major crypto firms such as Coinbase, Galaxy Digital Holdings and Paxos soon cut off ties, and eventually, on March 8, Silvergate declared that it was going to liquidate.
Signature Bank
Like Silvergate, Signature Bank started out as a mainstream commercial bank with a niche focus, namely in the real estate and residential markets of NYC. Also like Silvergate, Signature took a sharp turn into the crypto markets in 2018, with cryptocurrency businesses representing a quarter of its deposits by Q4 2022. These customers included several big firms such as Celsius Network, Coinbase, Binance, and Circle Internet Financial - the issuer of the USDC stablecoin (which Signature also held billions of dollars in).
But Signature’s main involvement in crypto wasn’t its depositors; in 2019, they announced a blockchain-based payment network called Signet. This was a real-time gross settlement (basically one that allowed for instantaneous transactions between financial institutions) system with no transaction fees or third parties, making it an ideal platform for crypto firms. By the end of 2020, over 740 clients were using Signet.
Signature’s demise is rather straightforward from here. As cryptocurrency prices dropped following the FTX collapse, Signature customers withdrew billions and the bank’s deposits dropped from $106.1 billion in the beginning of the year to $88.6 billion. This certainly wasn’t good news, but probably wasn’t quite enough to induce a bank failure. But when SVB (who also had strong crypto ties) went under, depositors panicked and demanded their money back from Signature, culminating in a bank run. It’s likely that people simply grew wary of any financial institution with major exposure to the crypto markets, which Signature had plenty of.
On Friday March 10, Signature lost more than $10 billion in deposits. That Sunday, the New York State Department of Financial Services shut down Signature Bank and handed it over to the FDIC. They designated it as a “systemic risk” to the financial system and was therefore allowed to go beyond the $250,000 insurance limit. Depositors probably breathed a huge sigh of relief at that news, seeing as how 90% of Signature’s $89 billion in deposits exceed the maximum insured by the FDIC. Signature Bank was the 3rd-largest bank failure in US history.
Credit Suisse
By March 19, the American-borne financial instability had spread to Europe, with bank stocks across the world taking a huge toll. Credit Suisse, one of the world’s largest and most important banks, was hit especially hard. Given their involvement in numerous misconducts and difficulties in recent years, the Swiss bank’s vulnerability shouldn’t come as a huge surprise.
In 2021, they had to downgrade their investment banking services following the implosion of Archegos Capital Management and Greensill Capital, both of which were involved in financial scandals and also happened to be borrowing from Credit Suisse. After that, in early 2022, Swiss prosecutors accused the bank of money laundering for a Bulgarian cocaine trafficking operation between 2004 and 2008. The bank was found guilty of failing to prevent the crime from taking place and sentenced with a CHF 2 million fine.
Later that year, social media rumors of the bank’s imminent collapse spread like wildfire, creating considerable pressure on the Credit Suisse stock price (which lost 75% of its value in 2022). As it turns out, however, they reported a loss of nearly $8 billion at the end of 2022 – their biggest loss since the 2008 financial crisis. More recently, Credit Suisse’s biggest shareholder, the Saudi National Bank, said that they wouldn’t invest beyond their 10% stake in the bank. Although chairman Ammar Al Khudairy didn’t cite any financial reasons for this decision, that very statement was enough to terrify investors.
As Credit Suisse bond prices fell 10%, the bank had no choice but to repurchase $3 billion of that debt from investors. Not only does this reduce the amount of debt owed (which is generally what firms look to do in troublesome financial situations), but it also partially restores investor confidence. Looking at the 5 year credit default swap prices, it’s safe to say that the market wasn’t feeling too bullish on Credit Suisse as a borrower:
Think back to what a credit default swap is: essentially a way for investors to protect themselves against the risk of a borrower defaulting. A higher CDS price means that the cost of insuring against a default has increased, which in turn suggests that the market is fearful of the borrower failing to pay their financial obligations. Being a bank, Credit Suisse wanted to do everything they could to avoid being seen as uncreditworthy, and a bond buyback seemed to be the best way to do that.
To do so, however, they needed government aid. Although they intended to refrain from using public funds to save a bank after the 2008 crisis, the Swiss National Bank (SNB) agreed to help and provided a $55 billion emergency loan. Not only did this action outrage the public as the SNB effectively broke its promise, but it also didn’t induce the investor-soothing effects Credit Suisse hoped it would: withdrawals totalled nearly $11 billion that week.
Apparently, however, this was all part of the plan. The SNB already accepted that Credit Suisse was beyond saving, at least on its own merits; they only agreed to help them with the bond buyback just so the Swiss Financial Market Supervisory Authority (FINMA) would have time to find a buyer. Just as the SNB authorized the emergency line of credit, they also reportedly delivered the following message:
You will merge with UBS and announce Sunday evening before Asia opens. This is not optional.
On March 19, UBS agreed to acquire Credit Suisse for $3.2 billion.
Econ IRL
ChatGPT has taken the internet by storm and now several experts anticipate a complete revolution in our way of life not so far ahead of us, with large language models (LLMs) behind it. This week’s paper, put together by researchers at OpenAI (the company that built ChatGPT), OpenResearch, and the University of Pennsylvania examines the potential impact of these technologies on the US labor market.
The short and simple answer is that nearly half of American workers could have half of their tasks exposed to LLMs. Web designers, journalists, accountants, and data managers, among others, are thought of to have the highest exposure to LLMs (“exposure” being defined as whether access to a GPT would at least halve the time to complete the task). Curiously, the authors found that the occupations with no exposed tasks were more often than not ones that don’t require college degrees; short order cooks, motorcycle mechanics, pile driver operators, etc.
‘Till next time,
SoBasically