On Bank Runs

How SVB collapsed in a matter of days

Silicon Valley Bank, a four decade old institution that enabled growth of thousands of good technology businesses globally, succumbed to a bank run over the last weekend. More banks followed suit in the coming days.

Without getting into the pathos, the politics, or the financial intricacies of this tragedy — below is a lucid explanation of how simple motions in a complex system that is taken for granted, can play out to create an extinction level event, under certain conditions.

Note: The focus here is on explaining the mechanics of the economic system in an intuitive way. In order to do so, macro-economic technicalities have been left out. But, here are other sources if you are interested in it: oped, podcast.

The Dynamics

Instead of narrating what happened to SVB in the past tense, let’s work through how money and information flows through the banking system under certain conditions:

You can follow along this stock-and-flow diagram of the banking operations relevant to this discussion. Arrows represent the flow of money or information. Green/Red indicate an increase/decrease in volume at destination, with increase in volume at source.

1. Say, the Federal Reserve System (the Fed) reduces the Bank Rate — the rate at which banks borrow money from the Fed - that guides downstream lending and interest rates, for loans and savings respectively. There are reasons for the Fed to change rates, but we are concerned with the downstream effects of this action, and will focus only on that here.
2. The reduction in interest rate (and hence returns on savings) will incentivize investors to channel their monies to other investment streams, increasing the supply of money available for ventures.
3. These ventures (businesses) will deposit monies received in a bank increasing the total amount held by the bank. If businesses of a certain category (like technology or retail) deposit in a single bank due to preference (because they provide customized services for the category) or are asked by their investors (because of existing relationships), a single bank may end up with a lot of money of the category’s ecosystem. Note: SVB’s portfolio predominantly comprised technology and life-sciences startups. They claimed to have half of US VC backed startups of these categories as their customers.
4. Banks give out loans at a lending rate that is higher than the interest rate, with the intention to profit from the spread (i.e. the difference between the two rates). But, in a low-interest regime, customers (like the businesses mentioned above) can get cash from equity investors so there isn’t enough demand for loans. This could lead to a drop in potential interest income.
5. In this scenario, the bank will have plenty of money sitting idle. To make this money work, it will invest. The bank may chose to invest in safe instruments like government bonds that guarantee returns in the long-term. Note: SVB invested in long-term fixed interest bonds (i.e. the return rate on these are locked for a certain (long) period of time).
6. All is good so far. The withdrawal rate is predictably periodic or small and infrequent, and in the long-run the bank will make money.
7. Now, let’s say the Fed spikes the Bank Rate in a dramatic way in response to an external reason. The reason could be something like controlling inflation — but we will focus only on the downstream impact.
8. With Bank Rate increasing, the supply of easy money in the market will drain out (same dynamic as #2, but in reverse direction), leading to a drop in deposits.
9. On the other end of the flow, to fund their day-to-day operations, depositors (like the technology startups who were SVB’s clients) will rely on their bank balance, leading to an increase in withdrawals.
10. Additionally, with an increase in the Bank Rate, Interest Rate will have to increase, and the bank will now have to pay more on its deposits compared to the low-interest regime.
11. In order to maintain liquidity and pay their depositors, who are now dipping deeper and more frequently into their saving — the bank will have to sell some of their investments.
12. But, with the increase in Bank Rate, when there are higher-interest options available — the demand for securities that are fixed at a lower-interest rate will go down. This will drop the price of the bond. So, the banks will be selling their investment at a loss.
13. The word about the bank’s losses will likely get around leading to a drop in trust by the depositors on the bank’s ability to give them back their money on-demand.
14. If the clientele is highly connected and if there are a few powerful influencers who — even if inadvertently — instill fear, the news will become viral. Panic is sure to ensue. Depositors will try to get in front of the line to withdraw money before it runs out, each acting in their best interest.
15. Soon, the bank will run out of money. All depositors will lose sleep. Most will lose their money — unless there is an external intervention.

The Bank Rate spiked by 4%-pts within one year. One of the steepest ever!

Was this really a Black-Swan event?

After working through this crude model, the turn of events seem obvious. But, could this scenario have been forecasted? Maybe.

A drawing from 1915 appears differently to different people. Some immediately see a young woman, and others an old woman. Squint, and you see both. Once you see the pattern, it’s impossible to unsee it!

But, one might wonder if there are fractures in the system?

• Do banks get enough time to meaningfully respond to steep Bank Rate hikes, and manage their liquidity? See #5, #12.
• Should banks have a concentrated clientele? See #3, #8, #14.
• Should industries have concentrated banking partners? See #3, #8, #14.
• Should a bank’s key income sources be similarly correlated to a common event like a Bank Rate change? See #4, #12.

It’s a relief that further snowballing of fear was averted by US government’s intervention. Hopefully science & technology entrepreneurs can go back to working on climate, health, convenience etc. and not scramble for existence.