Author: Jay Jiyuan

The Takeaways:

1. Bank runs are still possible today because although we have learned to tighten centralized financial regulations, we are still weak in financial risk management that proactively predict risk and take preemptive steps — even for highly predictable and familiar risks.
2. We have seen textbook examples of (1) how interest risk grows to a bank run through the well-known inverse relationship between bond price and (2) how liquidity risk from duration mismatch exacerbated by dramatic change in interest rate has created enough momentum to kill a solvent bank.

Silicon Valley Bank: The Good Beginning Story

Silicon Valley Bank (SVB, NASDAQ symbol SIVB) is not exactly a household name even in northern California where the bank is located (in Santa Clara County, one of the nine counties in the San Francisco Bay Area.) Frankly, I heard the 6 season comedy TV Series “Silicon Valley” since 2014 but don’t recall the bank’s name on top of my head, even though it is the 16th-largest lender in America, with about $200 billion in assets.

According to this article by Seeking Alpha, “ from 2019 to late 2022, SIVB total deposits more than tripled, growing from $61.7 billion in 2019 to $173 billion as of December 2022.”

A great story, right? Sure it is or was. But then comes the bad one. It all started from what the bank did with the deposited money. Normally this is not a problem, because most of the time most banks will lend the money out to individuals and businesses who can make a better use of the money than the depositors can, and earn a higher interest than they pay to the depositors.

This is what most banks do for living, among other activities.

For example, say SVB received a total deposit of $1 million from a startup firm called NuLife in Santa Clara County. SVB pays NuLife $20,000 (2% of the $1 million) in interest for putting their money in the bank, while keeps the rest of $980,000 in its account. The bank will not let the money sit there collecting dust but will lend it to another business called OldBuz at 5% interest, which is $980,000 x 0.05 = $49,000. In so doing SVB will earn a net amount of $49,000 – $20,000 = $29,000.

Earning a higher interest rate from the loan recipients than the interest rate banks pays to the depositors, this is the basic business model and SVB is not different from others. The only difference is that it did better than many others by attracting more depositors, especially tech startups and venture capital firms.

The Pandemic Shocks & the Changed Course of Government

Then the pandemic changed everything. I agree with this article of Business Insider that the SVB fallout “was a byproduct of the Federal Reserve’s hiking of interest rates by 1,700% in less than a year.”

But to fully understand the impact of the quick change of course by Fed, we must understand how much Fed had done during the pandemic, or how hard the Fed worked to make sure all lenders and borrowers can have easy access to money.

This paper of the Brookings Institute summarized the key changes by the Fed during the pandemic months, which has been credited with staving off an economic crisis and bolstering financial markets at a time when there was a “sharp contraction and deep uncertainty about the course of the virus and economy sparked a “dash for cash” — a desire to hold deposits and only the most liquid assets — that disrupted financial markets and threatened to make a dire situation much worse.”

First of all, the Fed purchased large quantities of government bonds and other securities, famously known as Quantitative Easing or QE, to make it easier for individuals and businesses to access credit, to stabilize financial markets and to support economic activity.

The Full Package of Stimulus from the Pandemic Era

The pandemic stimulus is not a single step but a full package of multiple programs, spanning not just monetary but fiscal steps. Let’s begin with the four (new and renewed) programs to promote financial liquidity for banks:

1. Money Market Mutual Fund Liquidity Facility (MMLF), a program introduced in March 2020 to provide liquidity to money market mutual funds (MMFs), which are investment in short-term, low-risk securities like commercial paper, certificates of deposit, and Treasury bills. Fed lent money to eligible MMFs at a low interest rate in exchange for collateral in the form of high-quality assets, such as Treasury securities and agency debt.
2. The Primary Dealer Credit Facility (PDCF), a lending program introduced in March 2020 to provide short-term funding to primary dealers, which are firms that have a trading relationship with the Federal Reserve and participate in the buying and selling of government securities, such as Treasury bonds and bills. These firms, which include SVB since 2015, are considered essential to the functioning of the financial system. Under PDCF, primary dealers can borrow funds from the Fed for a period of up to 90 days, using a variety of eligible collateral such as Treasury securities, agency debt, and mortgage-backed securities. The interest rate charged on these loans is set by the Fed and is typically lower than market rates.
3. Direct lending to banks with a lowered rate by 2 percentage points (from 2.25% to 0.25%). It’s said that eight big banks agreed to borrow from the discount window in March 2020, just so that other banks won’t feel bad and fear that markets and others will think they are in trouble.
4. Temporarily relaxing regulatory requirements to encourage banks to use their regulatory capital and liquidity buffers to increase lending during the pandemic.

The above are not the only game in town, as the Fed had other things in mind. Turned out that the liquidity it added to the shocked economy covered corporation (through the Primary Market Corporate Credit Facility or PMCCF, Commercial Paper Funding Facility or CPFF, Supporting loans to small- and mid-sized businesses, Supporting loans to non-profit institutions), households and consumers (through Term Asset-Backed Securities Loan Facility or TALF) and state and municipal borrowing (through Direct lending to state and municipal governments, Supporting municipal bond liquidity).

The following answer from Perplexity.ai, which integrates current web search results into the GPT (Generative Pre-training Transformer) process, tells us more about the fiscal stimulus on top of the monetary policy changes:

“The US government implemented both fiscal and monetary stimulus measures to mitigate the economic impact of the COVID-19 pandemic. Fiscal stimulus measures included direct payments to individuals, paycheck protection, student loan forbearance, eviction and foreclosure moratoriums. The economic impact of the pandemic sent the US economy into a recession in February 2020, with unemployment rates rising as high as 14.7% in April 2020.The direct payments to individuals were referred to as “economic impact payment” checks amounting to up to $1,200 per eligible adult.

“There were three rounds of such checks, including additional payments of up to $600 and $1,400 per person in 2021. The size and scope of these direct checks was a new experiment for the US government.

“The Federal Reserve Bank estimated that US fiscal stimulus during the pandemic contributed to an increase in inflation. However, economists largely agree that the money helped local governments shoulder significant pandemic-related costs and many governments avoided deep budget cuts. Many states have even reported surpluses.”

Here we have it: a teamwork of government branches toward the same goal of avoiding pandemic induced recession.

The Insightful Warning

Installing stimulus is one thing, foreseeing its full consequences is another. It is the latter that is the key for risk management. For that we must thank the former U.S. Treasury Secretary Lawrence H. Summers, who was the first to point out the danger of inflation following fiscal and monetary stimulus during the pandemic. The following answer from Perplexity.ai tells us that “In February 2021, he warned that additional government stimulus efforts to combat a pandemic slowdown raised the risk of inflation. He has since sent several warnings to Washington urging them to tap the brakes on stimulus or risk unleashing a serious burst of inflation.” Not only that, but Summers disagrees with the common belief that inflation is transitional as points out by this article of Barrons.com. “He was right: Consumer prices rose 8.6% year over year in May, the fastest pace in 40 years.”

Coordinated Stimuli Require Coordinated Risk Management

I agree with this article by Business Insider that cites Lundy Wright, partner at Weiss Multi-Strategy Advisers, that “(w)hen you raise interest rates quickly, after 15 years of overstimulating the economy with near-zero rates, to not imagine that there’s not leverage in every pocket of society that will be stressed is a naive imagining.”

Wright used a “double negative” sentence that is not the easiest to understand. His message is simply that we should expect some passengers to fall off of the bus when the driver made a sharp turn without reducing the speed.

Once again, much attention has been given to Fed’s monetary policy but not enough to the fiscal side, at a time when coordinated risk management is called for.

The aforementioned article of Business Insider points out that Fed’s “prolonged period of low interest rates created many financial dislocations that are now flaring up.”

In the SVB case, the “dislocations” came from two familiar places. The first is interest risk showing as the inverse relationship between interest rate and bond prices; the second is liquidity risk as maturity mismatch between long term asset holdings and short term liability demands. The SVB management has done a lousy job in handling both risks. These when combined with the startups’ high liquidity demand, led to the collapse of SVB.

The First Dislocation After QE: Interest Risk

Let’s begin from the inverse relation between interest rate and bond price. I’ll simply call it interest risk because what drives the inverse relation is the changed interest rate.

Long story short: The pandemic QE made it hard for banks to earn high interest income from writing loans to businesses and individuals. This is because when money is everywhere available, charging a high interest for bank loans is a mission impossible, it only drive customers to your competitors.  

So here is the problem: SVB was sitting on a big pile of deposited money and must find a place to keep it safely and profitably. SVC chose to park its money in treasury bonds, which is known as “risk free” because they are backed up by federal government tax revenues so there is no need to worry about default, meaning no need to worry about the bond issuers (the Treasury Department) to go bankrupted without paying off their debts to bond buyers or investors.

Of course, hardly anything under the sun is entirely risk free. Although the Treasury Bond (or “T-Bond” as it is often called, or simply “Treasuries”) itself is very safe, it does have a time related problem, where T-Bonds fight with each other to turn it into a risky game.

To illustrate, let’s think of a car dealership selling both used cars and new cars. The Treasury department is a new car dealership and only sells newly issued T-bond to investors. Bond buyers however don’t have to keep their bonds until maturity, just like car buyers don’t have to keep the same cars until they are no longer functioning. Bond buyers can sell their bonds before maturity just like a car buyers can sell his used car before it is dead — in a secondary market.

But here is where a risk free product becomes risky: In the secondary market where bond owners trade among themselves like used car owners do to their second or third hand cars, the price is not guaranteed. Some “used” T-bonds can sell a high price while others low, just like “hot” and not so hot brands of used cars.

One thing different in the used car market is that the price is almost entirely determined by the age of the car. A 1998 Civic will be sold cheaper than a 2011 Civic, other things equal. Another thing that is almost certainty is that nobody will ask a price higher than his original purchasing price — unless the car belongs to some rare classic models.

Unlike used cars, the old T-Bond can be resold in the secondary market either above or below the original purchase price. The age of bond has little impact on the resale price, only the difference between new and old interest rates.

Yes, in the bond game interest is the king, almost nothing else matters as much. Most people expect that the principal will be returned at the end of the maturity date, so This is so because people buy bond for the one reason of receiving interest payments, much like people buy car for driving. Buying bond is lending your cash to the bond issuer, who must pay interest to entice bond buyers.

Imagine someone wants to borrow $1,000 from you. Your first response is “Why would I lend you the money? I don’t even know you.” Well, the shortest — but convincing — answer from the stranger will be “I’ll pay you interest every month before I return your $1,000 in two years.” For most people, that’s a reason good enough.

Back to the old or secondary bond market, the key question buyers ask sellers is not how old the bond is, like in the used car market, but “what interest did you (or do I) get?” The reason is simple, if the seller gets 2% interest, that’s the rate the buyer in the secondary market will get after buying.

Here is the question: If the newly issued bond is paying 3% interest, why would anyone buy the “used bond” that only pays 2%? The only reason for a rational buyer is when the used bond (paying 2% interest) is offered at a lower price than the seller paid before. This is where an inverse relation between current interest rate and bond price comes into play.

This is similar to selling your old, gas inefficient car today at a lower price because buyers have more gas efficient new cars to buy from someone else in the new cars market.

SVB got itself into such a troubled situation as the bond seller: The bank needed to sell bonds for quick cash to pay depositors, but the only way for anyone to buy the bond is when SVB lowers the bond selling price. Every transaction when the bond changed hands means a loss for SVB.

The Second Dislocation After QE: Liquidity Risk

Liquidity risk, in its simplest term, means you are out of cash when you need them the most. It differs from poverty, which means no money anywhere in any form, people facing liquidity problem have money but in the wrong form or wrong places other than cash.  

SVB got itself into a liquidity trouble because it invested heavily into mortgage backed securities (or MBS for short), in addition to Treasuries.

For those not familiar with MBS, starting from mortgage loans would help. Strictly speaking, a mortgage is a loan, so you don’t have to say, “mortgage loan,” just “mortgage” is fine. Mortgage is a loan specifically for buying a home or property. Of course, lenders are not charities and then offer mortgage because they will receive interest payment from homebuyers.

When you take out a mortgage, you agree to pay back the money you’ve borrowed (called “principal”), plus interest, over a set period of time like 15-30 years, in addition to taxes and insurance. If the borrower fails to make payments on their mortgage loan, lenders have the legal right to take the home (or commercial property) back and put them in the market for sale through a foreclosure auction.

Generally speaking, mortgages that last anywhere at or above 10 years are long term loans. Most mortgages are therefore long term loans. The other feature is that individual mortgages are not considered securities because they have little risk — lenders can always take back the properties from borrowers, called collateral, for failure to make loan payments. Finally, a mortgage cannot be traded in the market because it is not an investment vehicle but a loan involving two parties: borrower & lender.

The story with MBS is different. First, it is created when banks issue mortgages to homebuyers and then pack or bundle up many mortgages and sell the package to a group of investors. As such, MBS is always an investment product bought and sold through a broker by investors, which include individual investors, corporations, and institutional investors on a secondary market.  

MBS is designed to free up the capital of the original mortgage lenders, often banks, credit unions and other financial institutions, so they can lend to more potential homeowners by leveraging investors who want to have a low risk investment at a discounted price.  

Secondly, most mortgages in the US are securitized, meaning they exist in the form of MBS that is traded in the markets for profit. Thirdly, most MBSs are issued by government-sponsored enterprises such as Fannie Mae, Freddie Mac, and Ginnie Mae that buy mortgage loans. Fourth, they are considered relatively low-risk investments especially if an MBS is guaranteed by the federal government, investors do not have to absorb the costs of a borrower’s default.

That said, an MBS is only as safe as the mortgages that back it up. During the subprime mortgage meltdown of 2007-2008, many MBSs were vastly overvalued due to non-payments. More generally, interest rate risk always exists with MBSs, as its price can drop when interest rates rise — just like the Treasuries (remember the discussion we had earlier?) In many ways MBSs are like bonds, both are fixed-income securities that pay a set amount of interest over time.

Interest risk explains “reinvestment risk” because when interest rate is low, borrowers want to refinance to take advantage of the lower interest rate. Let me illustrate with a hypothetic example.

Imagine you are currently paying off a fixed-rate mortgage with a 30-year loan term at 6% interest rate. Now, say the current interest rate is only 4%. You decide to refinance, which means to take out a new loan to pay off an existing mortgage. 

To make it easier to understand, say you have two investor friends, Fred and Sam. You met Fred first when the prevailing interest rate is 5%. Fred offered you $2,000 at 5% interest a year for two years, you thought you did not have choice, so you agreed. One year later you met Sam when the prevailing interest rate is 3%. After hearing your story with Fred, Sam says he’d be happy to lend you $2,000 for two years at only 3% interest.

Guess what you will do? You will borrow $2,000 from Sam and give it to Fred right away, telling him the deal is over as you find a lower interest to pay — assuming you’ve already paid $100 interest (5% of $2,000) to Fred for the last year. This is refinance and you saved yourself $80 because instead of paying $200 in two years to Fred, you only pay $120 to Sam in two years (3% of $2,000).

Fred now has a “reinvestment” problem because his money (a loan) was prepaid by you so he must find another borrower to lend the money to. Knowing Sam’s is willing to go as low as 3%, Fred will have to go down with the lower interest rate.

In addition to interest risk, credit and default risk is also associated with MBSs. This is straightforward: investors will experience losses if borrowers fail to make their interest and principal payments. Importantly, MBS investors are not the owner of the mortgage, so if a borrower defaults, not only the investor’s income will be interrupted but they do not claim any proceeds from foreclosure sales.

The Macroeconomic Risk of MBS: Negative Convexity

Interest risk, reinvestment risk and credit and default risk are all real but in the case of SVB, the macroeconomic risk associated with MBS is negative convexity.

First of all, convexity means curving outward—like the shape of the outside of a contact lens. The opposite is concavity, which means curving inward—like the shape of the inside of a contact lens. Put differently, a concave shape can be “filled,” while a convex shape creates a dome.

In our context, to understand convexity we need to understand duration first, which measures how sensitive bond price is to changes in interest rates. For example, if a bond duration is 3, it means when interest rate increases by 1%, bond price will decrease by 3%.

There are two contributing factors: time to maturity and coupon rate. The longer time before bond or MBS maturity, the higher the duration. This is easy to understand: If you lent money to someone and the borrower will pay you back tomorrow, you don’t care much about interest rate change because the loan has the “time to maturity” of just one day. Now, if you loan money to someone for 20 years, then interest change matters much more to you because your interest risk is higher over a longer period.

Similarly, the larger the coupon rate, the lower the duration, because a part of money has been paid back through coupons, which is just another name for “annual interest.” In an extreme case, we have bonds that are zero coupon bonds, meaning do not pay any coupon or annual interest at all until it’s maturity date, then duration is equal to time to maturity.

Now, convexity is measuring the rate of duration changes. Turns out duration is an approximation of the change in bond price in response to interest rate changes. For small changes in interest rate, it is accurate but not for larger ones as it always overestimates the price change if interest rates rise a lot, like the situation we are seeing today. Convexity helps correct this overestimation and provide a more accurate estimate of how much a bond’s price will change given a certain change in interest rate (or “yield” as commonly called).

With negative convexity, when the interest rate increases (like we are seeing today), the price of a negatively convex bond will fall by a greater rate. This does not hold for the opposite case when interest rates decrease. In other words, bond or MBS price is more sensitive to a rate increase than a rate decrease. A rate increase (like we see today with inflation) poses a bigger risk on bond price than a rate decrease. This makes negative convexity a bigger issue for SVB this time.

Callable bonds and mortgage-backed securities are examples of negatively convex bonds.