Structural Outlook
Where validated leading indicators point — based on historical patterns, not forecasts.
On average, its inputs sit higher than in 45% of their own quarterly histories since 2005.
Active Projections (1)
1 of 6 validated leading indicators currently show elevated readings. Historical patterns suggest downstream effects within the lag windows shown below.
Delinquency Rate on Credit Card Loans is currently elevated (+1.2σ). Historically, Charge-Off Rate on All Loans follows within 3 quarters with strong correlation (r = +0.76).
Based on historical patterns, not a forecast. See methodology.
When Initial Unemployment Claims is elevated, Continued Unemployment Claims typically follows within 1 quarter with very strong correlation (r = +0.95). Not currently triggered.
Based on historical patterns, not a forecast. See methodology.
When Initial Unemployment Claims is elevated, Unemployment Rate typically follows within 1 quarter with strong correlation (r = +0.79). Not currently triggered.
Based on historical patterns, not a forecast. See methodology.
When CPI Inflation Rate is elevated, Motor Vehicle Insurance CPI typically follows within 3 quarters with strong correlation (r = +0.77). Not currently triggered.
Based on historical patterns, not a forecast. See methodology.
When Energy CPI is elevated, Lower-Income Wage Growth vs. Inflation Gap typically follows within 1 quarter with strong correlation (r = −0.73). Not currently triggered.
Based on historical patterns, not a forecast. See methodology.
When Initial Unemployment Claims is elevated, U-6 Underemployment Rate typically follows within 1 quarter with strong correlation (r = +0.72). Not currently triggered.
Based on historical patterns, not a forecast. See methodology.
When Initial Unemployment Claims is elevated, Part-Time for Economic Reasons typically follows within 1 quarter with moderate correlation (r = +0.56). Not currently triggered.
Based on historical patterns, not a forecast. See methodology.
Cascade Timeline
Distress signals don't move in isolation — they propagate. This chart shows one of the strongest validated cascade sequences: how a signal in one part of the economy triggers downstream effects with measurable time lags.
Financial Conditions → Charge-Offs
Source: American Default Research leading indicator scanner. Chart axis uses a scanner-local z-score against a 2015-2024 baseline to place the chained indicators on one scale; this is a display normalization, not the American Distress Index method. Shaded regions: GFC (2007-2009), COVID (2020).
Historical Track Record
Each relationship was validated against up to three historical crises. The out-of-sample correlation (OOS r) tests whether the relationship held in data the model never saw during calibration.
| Leader → Follower | Lag | r | GFC | COVID | 2001 | OOS r |
|---|---|---|---|---|---|---|
| Initial Unemployment Claims → Continued Unemployment Claims | 1q | +0.95 | ✓ | ✓ | ✓ | +0.94 |
| Initial Unemployment Claims → Unemployment Rate | 1q | +0.79 | ✓ | ✓ | — | +0.79 |
| CPI Inflation Rate → Motor Vehicle Insurance CPI | 3q | +0.77 | ✓ | ✓ | — | +0.87 |
| Delinquency Rate on Credit Card Loans → Charge-Off Rate on All Loans | 3q | +0.76 | ✓ | — | ✓ | +0.83 |
| Energy CPI → Lower-Income Wage Growth vs. Inflation Gap | 1q | -0.73 | ✓ | ✓ | — | -0.81 |
| Initial Unemployment Claims → U-6 Underemployment Rate | 1q | +0.72 | ✓ | ✓ | — | +0.70 |
| Initial Unemployment Claims → Part-Time for Economic Reasons | 1q | +0.56 | ✓ | ✓ | — | +0.54 |
Lag: quarters between leader movement and follower response.
r: Pearson correlation at optimal lag. OOS r: Out-of-sample validation (calibrated 2000–2012, tested 2013–2025).
GFC: 2007–2009 financial crisis. COVID: 2020. 2001: Dot-com recession.
How We Got Here
Survival rate: 0.0108%. Read the full methodology →
Frequently Asked Questions
What is a structural projection?
A structural projection identifies a statistically validated historical pattern where one economic indicator consistently moves before another with a measurable time lag. The American Distress Index tests pairwise combinations of its indicators through a five-stage validation pipeline. Only 6 out of 57,541 tested pairs survive all five filters: cross-correlation, first-differencing, multi-crisis validation, Granger causality, and out-of-sample testing. These are not forecasts — they are observed historical patterns that may or may not repeat.
How are leading indicator relationships validated?
Each relationship must pass five independent statistical filters: (1) FDR-corrected cross-correlation on raw levels, (2) cross-correlation on first-differenced series to eliminate spurious trend correlations, (3) validation during at least two of three historical crises (2001 recession, 2007-2009 GFC, 2020 COVID), (4) Granger causality test confirming the leader statistically precedes the follower, and (5) out-of-sample validation where the model is calibrated on 2000-2012 data and tested on 2013-2025 data it never saw. The survival rate is 0.01%.
What does it mean when a projection is 'active' vs 'dormant'?
A projection is 'active' when the leading indicator currently sits far enough from its recent norm to match the precondition that has, in past cycles, preceded a move in the following indicator within the validated lag window. A 'dormant' projection means that precondition is not currently present, so no downstream effect is flagged. Active status is not a prediction — it reports that a historically validated precondition is present right now.
Get the numbers when they move.
New data drops, indicator updates, and ADI score changes — delivered when it matters. No spam.
or Create an Account for full access