A perfectly sound bank can be destroyed by a rumor in a single afternoon. In 2023 one was, by depositors tapping an app. This week you take a crisis apart and find the same skeleton every time.
Learning objectives. After this week you can (1) explain why a bank run can be self-fulfilling; (2) trace a fire sale into a liquidity spiral; and (3) name what 1907, the 1930s and 2008 had in common.
Remember this: a run needs no bad news, only the fear that others will run first.
1) Why runs feed themselves
No bank can pay every depositor at once (Week 2), and a solvent bank can still die of illiquidity (Week 9). Put those together with belief: if I think others will withdraw, my best move is to withdraw first. If everyone thinks that, the run happens whether or not the bank was sound. That is a self-fulfilling run: the bank pays whoever asks first.
2) Fire sales and the spiral
A bank facing a run must raise cash now, so it sells assets fast and cheap. That is a fire sale. Then the spiral turns:
- Falling prices cut the capital of other firms holding the same assets.
- They sell too, and prices fall further.
- Lenders pull back even from sound borrowers: a credit crunch.
- The damage spreads through funding, assets and fear: contagion.
Leverage speeds it up: a bank levered 10 to 1 is wiped out by a 10% fall in its assets. The circuit breaker is the lender of last resort. Walter Bagehot's 1873 rule: lend freely, against good collateral, at a penalty rate, so solvent banks never have to fire-sell.
3) Three crises, one skeleton
| Crisis | What ran | What came after |
|---|---|---|
| 1907 | Depositors at New York trust companies | The Federal Reserve, opened 1914 |
| 1930 to 33 | Waves of runs on thousands of U.S. banks | Federal deposit insurance, 1933 |
| 2008 | Lenders refusing to roll short-term repo loans to investment banks | Emergency lending and new rules |
Every time: short-term funding, long illiquid assets, leverage, and a sudden shift in beliefs.
4) Your team can have one too
Example. A team with a $60M bank spends $20M in each of its first two races and its Winnings are $0 both times. It now holds $20M with 4 races left, against $40M of planned spending at $10M a race. Thinner budgets can mean weaker results, weaker results mean less income, and less income means thinner budgets. There is no fire sale to try: the $9M second car can't be sold back. There is no lender of last resort: a team that can't fund a budget doesn't start (DNS), and a DNS earns no prize and no sponsor bonus. In a world with no backstop, your reserve is your central bank, and the time to build it is the good weeks.
Quick self-check
(a) Why can a run kill a solvent bank? (b) What are Bagehot's three conditions? (c) What did the 1930 to 33 runs lead to? (Answers: deposits are payable now but assets are long, and fire sales turn a cash shortage into real losses; lend freely, good collateral, penalty rate; federal deposit insurance in 1933.)
Common mistakes
- Thinking runs only hit bad banks. Illiquidity can sink a sound one.
- Trusting good-times funding. The easiest money to raise is the first to vanish.
- Mixing up the cures. Liquidity loans for the illiquid; capital or closure for the insolvent.
This week in the arena
Stress-test your team. Take your bank and this week's planned spend, and suppose every race from here pays nothing. How many races could you still fund before a DNS? That count is your crisis buffer.