Thursday, April 6, 2023

Recovering From Silicon Valley Bank's Fall



 

The solutions lie with the Fed, VCs, in-house management, and Silicon Valley itself.


Shyam Ponappa  April 6, 2023


After Lehman Brothers in 2008, Silicon Valley Bank’s (SVB’s) fall is significant because of its role in an area that is highly innovative and productive, an innovator for the world at large. The fallout gives rise to the question of whether constructive approaches can preempt a recurrence.
 
Simplistically, banks rely on customers ordinarily withdrawing only a portion of their deposits. Reality is more complicated, with customers seeking higher returns from mutual funds, or in direct investment in stocks, bonds, and other assets. Still, banking works on trust, with people believing they can get their money when they want. 
 
The SVB crisis unsettles this notion. The bank’s fall was triggered by a near-panic in Silicon Valley social media networks of tech founders and leading Venture Capital (VC) fund managers.1  SVB did have an unsustainable business model with serious liquidity issues. It was highly concentrated as the go-to bank for Silicon Valley VCs and their clients (nearly half of all technology and biotechnology start-ups). Its investments in bonds were in long-term mortgage-backed securities. A large increase in deposits in 2021 from high-tech profits led to $80 billion invested in low-yielding bonds in 2022, taking SVB’s total holdings to over $90 billion.
 
This worked during the decade when interest rates were low, but became unsustainable once rates rose.  Fed rates escalated by 4.5 per cent in one year from April 2022. Fed warnings to SVB in July 2022 were ignored.  Meanwhile, the technology downturn forced client withdrawals, while the interest-sensitive bonds dropped in value. SVB ignored the losses because securities were held to maturity, until Moody’s warned of a coming downgrade, galvanising SVB to act. It planned to sell $21 billion of bonds at a loss to reinvest in assets with higher returns, and raise $1.75 billion of equity to make up the loss. The bonds were sold booking a loss of $1.8 billion.  Moody’s downgrade was milder because of this and the proposed stock sale, but the preparatory work for the stock sale was incomplete when markets opened on March 9. SVB’s announcement of the loss and proposed share sale came together with bad news about another bank’s closure, Silvergate Capital, linked to bankrupt bitcoin exchange FTX. SVB’s stock plummeted, and a group of VCs and their clients tried to withdraw as much as $42 billion that day. California banking regulators closed SVB, appointing the Federal Deposit Insurance Corporation as the receiver. The ecosystem of SVB, VCs and start-ups was destroyed by these insiders themselves.
 
The market may take years to revive, if at all. Whether that level of facile agility can be achieved remains to be seen. Meanwhile, start-ups everywhere have much harder conditions after SVB’s fall.
 
Unless…

An effort by a group of VCs holds out hope.  Nearly 700 VC funds, including some of the most prominent, such as Sequoia Capital, Accel Partners, New Enterprise Ventures, Khosla Ventures, Kleiner Perkins, Bessemer, and Light Speed, coordinated by General Catalyst’s Hemant Taneja, have expressed support for a revival of SVB or its successor.2 If they succeed in working out a solution with regulators and government, presumably with better liquidity and risk management, although sectoral concentration may be unavoidable, there could be a recovery of Silicon Valley’s inimitable capacity for converting ideas into products and services. The Fed investigation of what went wrong is expected to be made public by May 1. 


These events emphasise the need to balance beliefs, e.g., in free markets, with the reality of reliance on regulations, protocols, governments and state-sponsored capacity in crises. This is true globally for whatever reason, whether for commercial, political, disasters, riots, or wars. They also question assumptions that banking and financial services can be run without the regulatory encumbrances that underlie established practices, because technological developments enable disintermediation driven solely by profits devoid of societal considerations. Slackened regulations and increased forbearance under the Trump administration provide examples. 
 
Regulatory Needs
 
US and Indian banking regulations need various changes. The objective needs to be resolution once boundary conditions are breached, not forbearance, with moral hazard reduced through stiff penalties.  The requirement is to institute controls as needed.  Existing protocols, for example, the Fed’s stress test of a recession, may not identify problems created by rapid rate increases. The aim has to be to avert crises.  For SVB, this could have been timely bridge financing and management change/induction with aligned VC backers/investors, initiated by regulators.
 
This conflicts with America’s forbearance and hands-off approach, such as allowing choice in classifying securities as held to maturity or available for sale, changeable at will. This means interest rate risks could remain undetected even with tighter controls. Instead, pragmatic, rule-based, tracking software of variables such as interest rate mismatches for all securities, asset growth or depletion and corresponding risk management factors, could flag issues through exception reports.  Management and regulators could address these, instead of having to search and find problems through diligent scrutiny. Normative action, such as marking securities to market to highlight mismatches, asset bubbles, and the like, could be flagged with automated, rule-based measures on securities and loans.  Real estate loans could require escalating equity participation for larger and multiple loans.
 
One problem in India is that the level of digital automation is incomplete. For example, published material from the government or the Reserve Bank of India is often images of text, PDF, or numeric documents, instead of searchable language and figures in databases. Periodic numeric reports are often not compiled, requiring scraping routines to extract sequential data. 
 
Another is that we may have to ease into rule-based systems, as the preference for discretionary control tends to increase higher up.

It is possible to institute controls, take initiatives and design software alerts to flag and avoid situations like the collapse of SVB.