Sahm Rule → Commodity Demand Destruction — Killed April 2026
Hypothesis: The Sahm Rule trigger (three-month average unemployment rate rises 0.5 percentage points above its 12-month low) signals the onset of recession and associated commodity demand destruction. Expected outcome: below-average commodity returns at 3–12 month horizons following a Sahm trigger at or above 0.5.
Battery result: 0 of 4 tests passed. Kill type: DIRECTIONAL INVERSION.
The finding: At Sahm Rule ≥ 1.5, commodity returns at the 6-month horizon average +20% — a substantial rally, not destruction. The 2008 and 2020 Sahm triggers, which reached the highest readings in the modern dataset, were both followed by significant commodity price increases in the 6-month window.
What went wrong mechanically: The Sahm Rule trigger coincides with central bank and fiscal responses that historically overwhelm the demand-destruction impulse. In 2008–2009, commodity prices crashed sharply — but recovered strongly within six months as stimulus flowed. In 2020, the same pattern occurred faster. At the 12-month horizon, both episodes showed positive commodity returns from the trigger date. The signal captures the beginning of the policy response cycle, not a sustained demand shortfall.
Test 1 — Monte Carlo null: Returns at Sahm trigger points are statistically positive in the null direction, failing the stated hypothesis at every threshold tested.
Test 2 — Blind era-split replication: The directional inversion is consistent across sub-periods. This is not a single-event anomaly; it is a structural pattern.
Test 3 — Specificity: The demand-destruction mechanism is not discriminated from a policy-response mechanism using commodity price data alone.
Failure mode: DIRECTIONAL INVERSION. The Sahm Rule is a valid recession-dating tool for its original purpose. Its extension to commodity demand destruction rests on a mechanism that is historically reversed by policy response within the observable investment horizon. It is killed as a commodity bear signal. Its value as a macro recession indicator is unaffected.
