Skip to main content
Causeway
Track F · CrisesNode F2

Replay 2008

Layer 2 · Working Model

Replay 2008 — and two other episodes for the rhyme

~ 5 min
Step 01 · The six-stage template

Every modern financial crisis has six stages. Build-up, complacency, trigger, panic, contagion, intervention. Names change; structure doesn't.

The widget below applies the template to three episodes. Default view is 2008; click the buttons to swap in 1997 Asian Crisis or 1929 Great Depression and see the same shape.

Step 02 · Read the two columns

For every stage, the widget shows what was visible to a normal observer on the left and what was happening underneath on the right. The gap between the two is the alpha. Reading newspapers tells you the left column; reading the plumbing tells you the right.

Step 03 · Watch the tells

Each stage has a single "tell" — the one line in period reporting that, if you knew what it meant, told you everything. For 2008: "BNP Paribas freezes three funds because we can't price them" (Aug 2007). That sentence is the whole crisis in one line.

Step 04 · Scrub through, then compare

Use the stage buttons or the arrow keys to walk through 2008. When you're done, swap to 1997 and walk it. The underneath columns will feel uncannily similar. The 1929 episode is the deep cut — same structure, but the intervention is missing the modern toolkit, so the consequences are catastrophic in a way 2008 was not.

In your life

If you've heard 'this time is different,' you are likely in Phase 1.

The phases are recognisable in real time. Calling the exact turn is harder, but the structural read is doable. Pair this replay with F1's anatomy framework and F3's bubble- sizing tools, and you have the working toolkit for reading the next one as it builds.

Six-stage template · scrub three episodes
US subprime mortgages → global credit freeze. The reference modern crisis.
← / → to scrub · 2003–2006
What's visible to a normal observer

House prices keep climbing. Homeownership at all-time highs. Wall Street earnings records.

What's happening underneath

Mortgage credit underwriting deteriorates. Securitization pipeline produces AAA-rated debt from BBB-rated loans. Bank leverage at investment banks reaches 30:1.

Metrics at this stage
Bank leverage (×)22
Credit spread (bps)70
VIX12
Bank deposit growth (%)8
Fed B/S ($T)0.8
The tell ▸
AAA tranches yielding 80 bps over Treasuries while subprime loss models assume housing never falls nationally.
Same shape, different costumes. The tells live in the right column.
Step 04 · Now try a different policy path

The replay above is the official record. The widget below lets you run the alternative — what if the Fed had cut earlier, or deeper, or held ZIRP through 2014? The transmission lags come from the same rate-transmission model used in C5; we convolve your deviation path against the C5 kernel and surface the response in unemployment, inflation, and growth.

The point is not to predict. The point is to make the counterfactual legible. Small assumptions about lags and elasticities produce large differences in the headline numbers — and that uncertainty is the whole reason the debate around 2008 policy remains live.

Counterfactual · what if the Fed had chosen differently?
Baseline — the Fed's actual rate path. Everything moves to match the historical record.
Policy-rate deviation from history (percentage points)
200720082009201020112012
Unemployment-quarters
0.0
pp · qtrs (sum)
CPI deviation
0.0
pp (avg across horizon)
Growth gap
0.0
pp · qtrs (sum)
Unemployment (U-3)%
4.47.29.92007Q12010Q12012Q4
What happenedCounterfactual
Core CPI · YoY%
0.81.72.62007Q12010Q12012Q4
What happenedCounterfactual
Real GDP growth · QoQ annualised%
-8.5-1.94.72007Q12010Q12012Q4
What happenedCounterfactual
How this is computed ▸

The widget convolves your specified policy-rate deviation against the C5 rate-transmission model — eight channels, each with a documented lag and elasticity. Unemployment has a 14-month lag and a +0.10 elasticity per 100bp hike; core CPI lags 18 months with a −0.20 elasticity; the growth channel is approximated at an 8-month lag, −1.0 per 100bp.

Caveats. The C5 elasticities were calibrated for normal regime transmission — not the zero-lower-bound, balance-sheet crisis transmission of 2008-2010. The model will systematicallyundersell the effect of interventions during periods when the financial plumbing was actually broken (the QE channels, the swap lines, the BTFP-equivalent). Treat the counterfactual lines as directional, not as forecasts.

Baseline series come from FRED (FEDFUNDS, UNRATE, CPILFESL, GDPC1) for 2007Q1–2012Q4. Counterfactual policy rates are clamped at 0% to respect the actual zero lower bound.

Small assumptions, large counterfactual differences. The whole point of the widget.