Outcome chronology
At year-end 2022 SVB reported USD 211.8 billion of consolidated assets, USD 120.1 billion of investment securities, USD 173.1 billion of deposits and USD 16.0 billion of stockholders' equity. It separately estimated USD 151.5 billion of uninsured U.S.-office deposits and reported fixed-income portfolio duration of 5.7 years. The latent funding and rate exposures therefore predated the final run. claim.svb.outcome.year-end-fragility table.svb.outcome.2022-position source · svb.outcome.2022-10k
On March 8, SVB disclosed that it had sold approximately USD 21 billion of securities, expected an approximately USD 1.8 billion after-tax loss and intended to raise approximately USD 2.25 billion. The capital raise was an intended transaction, not completed capital. claim.svb.outcome.march8-actions table.svb.outcome.march-sequence source · svb.outcome.march8-actions
DFPI reported that depositors initiated USD 42 billion of withdrawals on March 9 and that the bank ended the day with an approximately USD 958 million negative cash balance. California closed the bank on March 10 and the FDIC became receiver. claim.svb.outcome.run claim.svb.outcome.closure source · svb.outcome.dfpi-possession-order source · svb.outcome.bank-closure-8k source · svb.outcome.fdic-closure
The March 8 actions were a confidence trigger and communicated stress, but one securities sale is not a sufficient explanation. The duration and funding concentration existed before the announcement; the run explains the acute timing; deficient liquidity planning helps explain why withdrawals could not be met. claim.svb.outcome.trigger-not-sole-cause hypothesis.svb.outcome.multifactor source · svb.outcome.2022-10k source · svb.outcome.march8-actions source · svb.outcome.dfpi-possession-order source · svb.outcome.fed-review
Resolution perimeter
The joint government response protected depositors but did not rescue shareholders or certain unsecured debtholders. SVBFG later entered Chapter 11 separately from the bank receivership. First Citizens acquired the bridge bank's deposits and loans and about USD 72 billion of assets at a stated USD 16.5 billion discount, while roughly USD 90 billion remained in receivership. Those terms include a defined perimeter and cannot be converted into a simple enterprise value. claim.svb.outcome.systemic-response claim.svb.outcome.holding-company claim.svb.outcome.franchise-value table.svb.outcome.first-citizens source · svb.outcome.systemic-risk-statement source · svb.outcome.holding-company-ch11 source · svb.outcome.first-citizens-sale
Bank closure, holding-company distress and common-equity destruction were catastrophic. They do not prove that every loan, client relationship or operating capability had zero value. claim.svb.outcome.failure-perimeter source · svb.outcome.bank-closure-8k source · svb.outcome.holding-company-ch11 source · svb.outcome.first-citizens-sale
Causal assessment and rivals
The best-supported account is layered: concentrated and largely uninsured commercial funding plus lengthening securities duration created fragility; higher rates and weaker technology fundraising pressured asset values and client liquidity; management's hedge and liquidity decisions increased exposure; the March 8 loss and proposed raise crystallized concern; a networked run overwhelmed cash and collateral; and slow supervisory escalation reduced the time for remediation. The official Fed review is important primary postmortem evidence, but it is a self-assessment with a defined remit, and GAO called its own findings preliminary. claim.svb.outcome.fed-causal-findings evidence.svb.outcome.gao-limit source · svb.outcome.fed-review source · svb.outcome.gao-review source · svb.outcome.reuters-explainer
The “rates and technology downturn alone” rival cannot explain SVB-specific funding and duration choices. The “bond sale alone” rival mistakes a trigger for the entire vulnerability. The “run alone” rival explains speed but not the inability to withstand it. The “management alone” and “supervision alone” rivals each omit the other layer. No single source proves precise causal weights. hypothesis.svb.outcome.macro-only hypothesis.svb.outcome.sale-only source · svb.outcome.2022-10k source · svb.outcome.march8-actions source · svb.outcome.dfpi-possession-order source · svb.outcome.fed-review
Ex-ante signals and counterfactual
Cutoff-valid signals included USD 166.0 billion of estimated uninsured U.S.-office deposits, a sector-linked commercial funding model, a 5.2-year HTM duration, a 30.4 percent modeled EVE decline under +200 basis points, and contingent borrowing capacity that had not been established as operationally drawable in this public packet. Each has false positives: uninsured does not mean immediately runnable; duration can be economically held; static EVE models are assumption-sensitive; nominal facilities may be drawable. The appropriate response was validation and staged resilience, not a prediction of imminent failure. source · svb.cutoff.2021-10k source · svb.cutoff.q1-2022-10q
The feasible counterfactual is the Part A program: preserve controlled execution while hedging duration, testing borrowing channels, diversifying funding and measuring network-correlated deposit cohorts. The sources do not support a deterministic estimate of how much loss it would have prevented or whether it would certainly have stopped the run. counterfactual.svb.outcome.staged-remediation source · svb.cutoff.finance-charter source · svb.cutoff.q1-2022-10q source · svb.outcome.fed-review
Candidate agent rules
- Treat uninsured deposit share as a funding-risk input, never as an assumed one-day runoff; add controller and network correlation. rule.svb.deposit-correlation
- Reconstruct time-to-cash after collateral haircuts and operational constraints; nominal liquidity is not drawable liquidity. rule.svb.operational-liquidity
- Separate trigger, structural vulnerability and amplification in causal analysis; do not say one bond sale alone caused a bank failure. rule.svb.layered-causality
These are candidate rules from one case. They require cross-case testing, counterexamples and false-positive calibration before promotion. source · svb.cutoff.2021-10k source · svb.cutoff.q1-2022-10q source · svb.outcome.fed-review
Boundary conditions
Do not infer that every specialized bank is unstable, that all uninsured deposits will run, or that every HTM fair-value decline is a realized loss. Do not equate AOCI, HTM marks, tangible equity and regulatory capital. Do not infer that depositor protection was an equity rescue. Do not infer enterprise value by subtracting a stated transaction discount from acquired assets. And do not let March 2023 run speed erase the duration and funding decisions observable before it. source · svb.cutoff.2021-10k source · svb.outcome.systemic-risk-statement source · svb.outcome.first-citizens-sale