The risk committee has seen the numbers twice. Everyone agrees the exposure is growing. Nobody wants to be the one who acted too early on incomplete data. The item rolls to next month's agenda.
Nobody decided to wait. The calendar decided for them.
What the review found
On September 18, 2026, Michelle Bowman, the Federal Reserve's Vice Chair for Supervision, presented initial findings from an independent review of how Silicon Valley Bank was supervised. Staff believed it was "personally safer to take no action" unless they were certain the action was exactly right. Unclear decision rights reinforced that belief. Staff couldn't tell who could confirm that an action was correct, because responsibility, authority and accountability had been split apart. The review described the result as a risk-averse culture.1
Cultures are built from small, repeated choices, like that agenda item rolling forward.
The finding is about bank supervisors. The mechanism isn't. Any organization where acting wrongly is punished and waiting is not is set up to get more waiting.
Delay is a decision with a price
Waiting looks free because its cost arrives later and lands on nobody in particular. Make it land. When a decision is deferred, record the deferral as a decision in its own right:
Deferred decision: [decision]
What we're waiting for: [the specific evidence]
What waiting costs, per month: [exposure or opportunity, estimated]
Who owns that cost: [name]
Decide by: [date], whether or not the evidence arrives
Then ask a harder question. What evidence, if it arrived, would change the decision? If nothing could arrive by the decide-by date, the committee isn't gathering evidence. It's avoiding the decision with better minutes.
None of this argues for acting fast on thin evidence. Some decisions should wait. The difference is whether the waiting was chosen, priced and owned, or simply happened because the next meeting was a month away.
Fix the decision rights first
The review found that unclear decision rights made inaction the safe choice. The repair starts there. Write down who can decide, on what evidence and by when. A committee can advise, challenge and record dissent. One named person should be able to say yes, and be expected to. Put that in the committee's terms of reference, not in a memo that leaves with the chair.
That person also needs cover for acting on the best evidence available at the time. If the only defensible move is certainty, people will wait for certainty, and certainty can arrive with the loss.
In a startup, the bottleneck is obvious: one founder, one inbox, one decision waiting. In a large firm the same bottleneck hides in a meeting cycle, where it looks like diligence.
Waiting is a decision. Give it an owner and a price.
Related Reading
Read The Indicators Were Green. The Situation Was Not. for the opposite failure: acting on signals that looked certain and weren't.
