Silvergate and Signature Failed Because Their Depositors Were Correlated
Two banks serving crypto clients failed in 2023 alongside Silicon Valley Bank. All three shared a deposit base concentrated in one industry.
The Common Feature
Three American banks failed in close succession in early 2023. Silicon Valley Bank served venture backed technology companies. Silvergate and Signature had built substantial businesses serving cryptocurrency firms.
Each had specific problems, and they shared a structural characteristic that mattered more than the differences. Their deposit bases were concentrated in a single industry.
Why Liability Concentration Is Underweighted
Bank risk management focuses heavily on the asset side. Loan concentration limits restrict exposure to any single borrower, sector, or geography, because a downturn affecting many borrowers simultaneously produces correlated losses.
Deposits receive far less attention, because they are liabilities rather than exposures. That framing misses that a deposit base can be concentrated in exactly the same way, and correlated withdrawal is a faster failure mechanism than correlated default.
Correlated borrowers produce losses over quarters. Correlated depositors produce a failure over days.
How the Correlation Worked
A bank serving one industry has depositors who read the same news, use the same advisors, know each other, and respond to the same conditions.
For the crypto focused banks, the deposit base contracted as the sector experienced substantial stress through 2022, and confidence deteriorated further as high profile failures occurred. Deposits were also largely uninsured, since business balances typically exceed the insurance threshold, giving depositors a genuine reason to move quickly.
At Silicon Valley Bank the mechanism was similar, with venture investors reportedly advising portfolio companies to withdraw, which produced coordinated action among depositors sharing the same advisors.
The Asset Side Interaction
Liability concentration became fatal through interaction with asset side problems. Deposits gathered during a period of sector expansion had been invested in longer dated securities purchased when rates were low.
When rates rose, those securities were worth less than carrying value. That was tolerable while deposits were stable and disastrous once withdrawals forced sales, which converted unrealised losses into realised ones and destroyed capital.
Neither factor alone would necessarily have caused failure. Together they were sufficient.
What Should Be Measured
The practical implications follow directly. Deposit concentration by industry, the proportion of deposits above the insurance threshold, and the degree to which depositors are connected to each other are all measurable and were not prominent in standard analysis before 2023.
The speed of modern withdrawal compounds all of it. Deposits move by application in seconds, and information spreads through group messaging faster than any institution can respond.
The Bottom Line
These banks failed because their depositors were effectively one depositor. Concentration analysis belongs on both sides of the balance sheet, and the liability side fails faster.