#357: The Demise Of Silvergate Capital, Silicon Valley Bank, And Signature Bank Should Send A Strong Message To The Fed, & More

Last Wednesday, during a historic week for banking and financial markets, Silvergate Capital announced that it had entered Federal Deposit Insurance Corporation (FDIC) receivership and would liquidate assets and wind down operations. Silvergate catered to institutional investors looking to transfer government issued (fiat) currency to crypto exchanges. Taking fiat deposits and routing them through its Silvergate Exchange Network (SEN), Silvergate grew crypto-related deposits more than six-fold in two years, from less than $2 billion in early 2020 to ~$13 billion in 2022. After the collapse of FTX last November, its customers began to draw down deposits, which Silvergate honored initially by liquidating assets on its balance sheet and tapping the Federal Home Loan Bank of San Francisco for a credit line. In early March, however, citing audit procedures and regulatory inquiries, the company delayed its 10-K filing, and one week later wound up in FDIC receivership. Prior to its demise, several crypto firms, including Coinbase, moved their banking from Silvergate to other crypto-friendly banks like Signature Bank.

On Friday, in response to a bank run, the FDIC took control of Silicon Valley Bank (SVB). Deeply intertwined with the startup and venture capital communities globally, SVB not only has underwritten the majority of venture-backed companies in the US, but also has extended commercial and personal loans to startups and their employees and served as the banking partner for many venture capital funds. Shares of SVB fell 60% on Thursday after the company announced on Wednesday that it intended to raise $2.25 billion in equity to cover a $1.8 billion loss caused by liquidating its available-for-sale securities portfolio. Several prominent venture funds urged their portfolio companies to withdraw funds following the abrupt capital raise and portfolio liquidation. On Friday morning, the U.S. Securities and Exchange Commission (SEC) halted the trading of SVB, and by noon the FDIC had taken control.

Like Silvergate, SVB grew its deposit base aggressively, from $60 billion in 2020 to $190 billion in 2022, fed by the boom in venture funding. Betting on continued low interest rates and deposit growth, it invested most of the windfall into a $120 billion portfolio of long-term mortgage-back securities and municipal bonds that yielded less than 2% on average. The majority of these investments landed in a category called “held-to-maturity”: they would not face mark-to-market losses unless sold. As interest rates increased and venture funding dried up, however, deposit outflows gained momentum, raising the probability that SVB’s held-to-maturity securities would mark to market, potentially recognizing $15 billion in unrealized losses that could have bankrupted the bank.

Although the FDIC has taken decisive action to guarantee full access to deposits—insured and uninsured—for both SVB and Signature Bank, which was closed by regulators as well, consumers and businesses may be reticent to engage with any banks other than the highly regulated SIBs (systemically important banks), also known as money center banks. Regional banks may have trouble regaining traction in deposit growth. A string of recent strategic defaults in the commercial real estate space could add insult to injury, as regional banks are much more exposed than larger banks to commercial real estate loans.

For thoughts on how these financial sector events could change the conversation at the U.S. Federal Reserve (Fed), we invite readers to listen to the March 10th episode of In The Know With Cathie Wood: Is The Fed Listening? Hopefully, the Fed is beginning to understand its role in this unfolding crisis: focused on lagging indicators, it raised interest rates an unprecedented 19-fold in less than a year. The bond market has been warning the Fed for roughly nine months that trouble has been brewing. As measured by the difference between 10-year and 2-year U.S. Treasury yields, the yield curve inverted last July. Last week, the inversion hit ~100 basis points (~1%), an extreme not seen since the Volcker-led Fed in 1981. Adjusting for the level of interest rates, however, the inversion is unprecedented in post WWII experience: 100 basis points (or 1%) on 4% long bond yields is suggesting something much more serious about financial conditions than 100 basis points (1%) on 15% in 1981. In our view, the Fed will have no choice but to respond with lower interest rates.

Monoclonal antibodies (mAbs)—drugs like Herceptin, Humira, Keytruda, and Opdivo—are a powerful class of therapeutics that generate ~$200 billion in revenues per year.¹ Recently, scientists expanded the scope of antibody-based medicines by creating new modalities like antibody-drug conjugates (ADCs) and bispecific antibodies.

In 2019, the FDA approved the first VHH antibody medicine (Cablivi) for the treatment of rare blood clotting disorders. Produced by alpacas, VHH antibodies are smaller, more nimble molecules compared to their human counterparts. Recently, Twist’s (TWST) biopharma division created and launched a high-throughput VHH antibody discovery service that should lower the barriers to discovery and development of powerful VHH-based medicines.

Once researchers select a disease target, or antigen, they can use several methods to discover a viable antibody drug against the target. They can immunize animals like mice, rabbits, or alpacas and harvest potentially therapeutic antibodies from blood samples. They also can expose antigen-bound viruses to different combinations of potential antibodies in a process called phage display. More recently, researchers have leveraged single B-cell isolation methods using high-throughput technology platforms like the Berkeley Lights (BLI) Beacon. Already compatible with human and mouse B-cells, Twist’s new workflow will enable Beacon to work with alpaca B-cells.

In the company’s recent pre-print, the authors show how alpacas immunized with prostate specific membrane antigen (PSMA), a known drug target, could have their B-cells isolated, analyzed, and cloned on the Beacon system. The researchers first tested the resulting VHH antibody candidates, demonstrating that several bind tightly to the target antigen, and then leveraged Twist’s machine learning algorithms to “humanize” the alpaca-derived antibodies, reformulating them to decrease the likelihood of unwanted immune reactions. Twist plans to offer this workflow as a service alongside its other in vitro, in silico, and other in vivo antibody discovery techniques.

[1] Lu, R. et al. 2020. “Development of therapeutic antibodies for the treatment of diseases.” Journal Of Biomedical Science, vol. 27. https://jbiomedsci.biomedcentral.com/articles/10.1186/s12929-019-0592-z

During its Investor Day, for the first time Tesla disclosed the accident rate of its Full Self-Driving (FSD) system: one accident per 3.2 million miles, which is more than six times safer than the national average of one per ~500,000 miles. That said, until its recent update, v11.3, the FSD beta software stack applied only to surface, or city streets. Surface streets have a higher accident rate than highways, as they represent ~30% of miles driven but 75% of crashes.  Further, the national average includes older cars that do not have active safety features, like emergency braking (AEB), that now are standard on all new US cars.

To evaluate the incremental safety of FSD, an apples-to-apples comparison would be against a manually driven Tesla. Adjusting to surface street specific accident rates, compared to the “non-Autopilot” accident rate in Tesla’s safety report, FSD-equipped cars appear to be ~5X safer than manually driven Tesla cars and 18X safer than the average car in the US, as shown below.¹

Source: ARK Investment Management LLC, 2023;  https://www.tesla.com/VehicleSafetyReport; https://tesla-cdn.thron.com/static/AA7YQM_Investor_Day_2023_Keynote_W9DARX.pdf?xseo=&response-content-disposition=inline%3Bfilename%3D%22Investor-Day-2023-Keynote.pdf%22; https://cdan.nhtsa.gov/tsftables/tsfar.htm; https://www.bts.gov/content/roadway-vehicle-miles-traveled-vmt-and-vmt-lane-mile-functional-class. For informational purposes only and should not be considered investment advice or a recommendation to buy, sell, or hold any particular security.

[1] The Non-Autopilot accident rate, labeled “Manually Driven Tesla” in the graph, as well as the national average, were both adjusted to surface-street only accident rates using the ratio of percent of accidents to miles driven on surface streets, including an adjustment for auto-pilot miles, which ARK estimates are ~10% of miles driven in Teslas. Note that this a global statistic and the US average are likely higher. Chart FSD data as of March 1, 2023. Five important caveats to consider: (1) a driver might disengage FSD in the most difficult driving moments, which could boost miles per accident; (2) FSD will not work in extreme weather; (3) while passenger cars make up the majority of vehicles in crashes and total miles, the national statistics are not confined to passenger cars; (4) our comparison includes all crashes, fatal and non-fatal; (5) The national average includes vehicles that do not have active safety features such as auto emergency braking, which are now standard on all new cars, including Teslas.