The Three-Phase Crash Model
Every great crash follows: bubble → resolution → aftermath
Linda Yueh argues that all great financial crashes share a universal three-phase structure. Phase one is the bubble: irrational exuberance fuelled by genuine technological or economic progress, amplified when participants pile in using borrowed money. Phase two is resolution: the credibility and speed of policy response determines whether the crash deepens or stabilises. Phase three is the aftermath: the duration and severity of recovery is set almost entirely by the cause (how much debt was involved, whether banks were dragged in) and the quality of the policy response in phase two.
Yueh drew this framework from studying the 1929 crash, which she treats as the template because it contains all three phases in unusually clear form. The Roaring Twenties created the bubble, FDR's bank holiday and Fireside Chat represented the resolution, and the decade-long Great Depression — punctuated by the 1937 recession-within-a-depression caused by premature policy withdrawal — defined the aftermath. She applies the same three-phase lens to every crash she studies: 2008, the eurozone crisis, Japan's lost decades, and COVID.
The framework is diagnostic, not predictive. It does not tell you when a crash will happen, but it tells you what phase you are in, what decisions matter most at that phase, and which historical analogues are most relevant. The aftermath phase is where most policy mistakes occur — either withdrawing support too soon (as in 1937) or tolerating zombie firms too long (as in Japan). Knowing the phase reframes what question to ask: in phase one, ask how much leverage exists; in phase two, ask how credible is the response; in phase three, ask whether viable businesses are being supported or zombies are being propped up.
- Every great crash begins with a bubble driven by genuine innovation plus human FOMO — you cannot eliminate phase one.
- The crash becomes great when leverage is involved: debt turns a correction into a catastrophe.
- Resolution quality is determined by speed and credibility of the policy response, not by the size of the intervention.
- The aftermath duration is set in phase one: banking crashes produce decade-long recoveries; equity-only crashes produce short, shallow recessions.
- Premature withdrawal of policy support re-triggers the crisis — the 1937 recession-within-a-depression is the canonical warning.
- Identify the bubble and its fuelDetermine whether the rising asset class is driven by genuine fundamentals or pure euphoria, and — critically — whether participants are buying with borrowed money. Leverage is the key diagnostic: the dot-com crash was equity-funded and caused a shallow recession; the 2008 crash was debt-funded and caused a decade of pain.Pro tipAsk who is lending the money, not just who is buying the asset. Banks dragged into a bubble guarantee a banking crash, which is categorically worse.WarningGenuine technological progress and a bubble are not mutually exclusive. Dot-com was both — the technology delivered, the timing and companies did not.
- Assess policy credibility in the resolution windowThe first ~10 months of the resolution phase are critical. A credible leader backed by real policy changes confidence faster than any technical measure. FDR's 1933 bank holiday worked because it was backed by Deposit Insurance; Draghi's 'whatever it takes' worked because the ECB was backed by the political will of Eurozone leaders.Pro tipFinancial markets respond to the signal more than the policy. Draghi ad-libbed his speech and it stopped the acute phase of the Euro crisis before any actual policy was deployed.WarningCredibility borrowed without backing runs out fast. Hoover made optimistic statements for years without policy support and destroyed his credibility; his successor had to spend political capital rebuilding it.
- Support viable businesses, not zombiesDuring the aftermath, policy must distinguish between firms that are illiquid (viable but starved of credit) and firms that are insolvent (not viable even with credit). Japan's failure to make this distinction led to decades of zombie-firm support that prevented the banking sector from ever truly recovering.Pro tipCentral banks can provide targeted credit lines to sectors like SMEs at below-market rates, making it structurally attractive for banks to lend to viable firms rather than rolling over zombie loans.WarningPerfect application is impossible — some support will always reach unviable firms. The goal is to minimise zombies, not eliminate them.
- Monitor the aftermath for premature withdrawalThe most common policy error in phase three is removing support too early. The 1937 recession-within-a-depression happened because the Treasury and Fed believed recovery was complete. Watch real unemployment, credit availability, and business investment before tapering support.Pro tipHistory doesn't repeat but rhymes — the 2013 'taper tantrum', when the Fed signalled bond-buying cuts, triggered EM crises. Even the signal of withdrawal can be destabilising before the real economy has healed.
On taking office in 1933, FDR faced a month-long bank run. Over a weekend he closed all banks and the stock exchange, then gave his Fireside Chat radio broadcast telling Americans directly that only sound banks would reopen and it was safer to deposit money than keep it under the mattress. The policy was backed by Deposit Insurance legislation already in progress.
At the Global Investment Conference in London during the 2012 Olympics, ECB President Draghi ad-libbed a line — not in his prepared remarks — telling global investors: 'Believe me, we will do whatever it takes to safeguard the Euro.' Bond spreads between peripheral economies (Greece, Portugal) and safe-haven Germany began narrowing immediately. No policy had yet changed.
After the early-1990s property crash, Japan took eight years to address its insolvent banks. State support flowed to zombie firms — companies that were technically alive because of cheap credit but not viable businesses. The banking sector could not recover because its balance sheets were full of uncollectible loans.
The dot-com crash was funded largely by venture capital and retail equity investors not buying on margin. When it crashed, the NASDAQ fell 80% and didn't recover until 2015. But because banks were not the primary lenders, no credit crunch followed — the recession was short and shallow. The 2008 housing crash was different: banks held the mortgages, so when collateral fell, they pulled back lending, creating a credit crunch that starved the whole economy.
Yueh developed this framework across her book 'The Great Crashes: Lessons from Global Meltdowns and How to Prevent Them' (2024), writing about ten major crashes spanning a century. She traces the three-phase structure to the 1929 crash, which she opens the book with because it remains the foundational model that all subsequent crash management has been calibrated against. The Federal Reserve's error of withdrawing support prematurely in 1937 — triggering a recession inside the depression — was so consequential that it has shaped central bank doctrine ever since, making 1929–1941 the clearest available specimen of all three phases.