Vault
reference

The Failure Cascade Index

Created

The Failure Cascade Index

Builds-on: how-inflation-dies-the-empty-reservoir Related: two-economy-gauge (matched pair — this doc watches the failures, that one watches the fuel), normalization-vs-destruction-test, cyclical-20-and-the-ai-capex-mask, demand-side-audit-may-2026 Informs: regime-check rotation

Purpose

Make stage 4 of the demand-destruction sequence (the input-cost failure cascade) visible in US data before it becomes a headline. The UK had a clean insolvency series that showed its 2022-23 cascade in real time; the US doesn't publish one, so this composites the pieces. The empty-reservoir doc's core claim — businesses are options that fail discretely, so the cascade always looks "not happening" until it is — makes this the gauge most likely to be read too late.

Design principle inherited from two-economy-gauge: the US economy is tiered. Aggregate and large-firm series will mask the cascade the same way Case-Shiller masks 77 falling metros. Every component here is chosen to read the small-firm tier specifically, because that's where input-cost pass-through failure kills first (thin margins, no hedging, no pricing power, bank-line dependent).

The leading/lagging trap

Each component is tagged with its timing relative to the failure wave itself — not relative to the business cycle generally. This distinction is where gauge-reading goes wrong: credit tightening leads failures by 2-3 quarters (the fuse), filings are coincident confirmation (the explosion), and establishment-death statistics lag by a year (the coroner's report). A dashboard that averages all three into one number destroys the information. Read them as a sequence, not a composite score.

Components

# Series Source / access Freq Timing vs. failure wave Reading (July 2026) Calibration
1 Absorption wedge: PPI final demand YoY minus CPI YoY BLS, both monthly Monthly Leading (2-4 quarters) — measures costs firms are eating; the wider and longer the wedge, the more option-decay is accumulating ~+1.8-2.2pts (April: PPI 6.0% vs CPI 3.8%; May CPI 4.2%) — wedge open and wide 2021-22 wedge peaked ~+4pts and was closed by pass-through (consumers paid). The danger reading is a wide wedge that persists while volumes fall — pass-through no longer available
2 NFIB price-plans vs. compensation-plans divergence NFIB SBET monthly Monthly Leading (1-3 quarters) — firms planning price hikes while cutting comp plans = squeezing both ends before failing Net 36% raising prices (highest since Mar 2023), 34% planning increases (highest since Jul 2022) — while comp-raising fell to 28%, lowest of the year The 2026 configuration (price plans UP, comp plans DOWN, optimism below historic average) is the desperation signature — last-ditch pass-through attempts. Contrast 2021: both rose together (confidence)
3 SLOOS: net % of banks tightening C&I standards, small firms FRED DRTSCIS, quarterly Quarterly Leading (2-3 quarters) — small firms die when the line gets pulled, not when losses accrue. This is the fuse length 9% net tightening (Q4 2025), continued tightening Q1 2026; banks expect small-firm loan quality to deteriorate through 2026 Mild so far: 2008 peaked ~70%+, 2020 ~40%, 2023 regional-bank scare ~49%. A move above ~25-30% with the wedge still open = cascade acceleration ahead
4 Subchapter V elections (small-business Chapter 11 track) Epiq AACER monthly, ~2-week lag Monthly Coincident — the explosion itself, cleanest small-business failure count in US data Q1 2026 +67% YoY (833 vs 499); H1 +50% (1,663 vs 1,107); June +28% This is the series that answers "is the cascade happening": as of H1 2026, yes, it is — see reading below
5 Commercial Chapter 11 filings, total Epiq AACER Monthly Coincident (larger firms — confirms breadth beyond the small tier) Q1 +37% YoY (2,422); April +42%; H1 +28% (4,589); all commercial filings +13% (17,285) Broadening beyond small firms = stage 4 climbing the size ladder
6 Commercial loan delinquencies, banks outside top 100 FRED (delinquency rates by bank size); Equifax SBFE small-biz delinquency Quarterly Coincident-to-lagging — damage already booked Consumer analog already at 6.43% (cards); pull the C&I small-bank series on first full run Tiering check: if small-bank C&I delinquency diverges from large-bank the way consumer did, the mask is confirmed
7 Restaurant canary panel: OpenTable seated diners, Toast same-store sales, accommodation/food-services openings (JOLTS) OpenTable public data, Toast quarterly, BLS Weekly-monthly Leading sectorally — hospitality fails first everywhere (UK 2023: +37-45% insolvencies led the wave) Not yet pulled — first-run task UK sequence: hospitality insolvencies led total insolvencies by ~2 quarters
8 BED establishment deaths BLS Business Employment Dynamics Quarterly, ~9-month lag Deeply lagging — calibration and backdating only, never for recognition n/a Use to date the cascade after the fact, NBER-style

Current reading: stage 4 is firing (tier one)

The July 2026 initial reading produced a material correction to how-inflation-dies-the-empty-reservoir, which had placed the US at "stage 2 crossing into stage 3, stages 4-5 not fired." At the small-business tier, stage 4 fired quarters ago: Subchapter V +50-67% YoY is not an early warning, it's a cascade in progress — masked in public perception because (a) absolute filing counts started from post-COVID lows, (b) the aggregate commercial number (+13%) blends the small-firm surge with a calmer large-firm tier, and (c) small-business failures don't make headlines the way a large Chapter 11 does. Epiq's own attribution: "higher borrowing costs and softening demand" — input costs meeting the pass-through ceiling, which is this framework's stage 4 definition.

The sequence read as of July 2026:

What would mark stage 4 → stage 5: the tripwires already live in cyclical-20-and-the-ai-capex-mask — U-3 above 4.6-4.8%, Sahm above 0.30, continuing claims above 1.95M — plus, from this gauge, SLOOS small-firm tightening crossing ~25-30% (the fuse shortening while explosions are already audible).

Reading protocol

Quarterly, as part of the regime-check rotation:

  1. Update the wedge (monthly PPI/CPI), NFIB spread (monthly), SubV/Ch11 counts (monthly), SLOOS (quarterly).
  2. Report as a sequence position, not a score: fuse state → explosion state → breadth → transmission.
  3. Never average leading and lagging components. The trap this gauge exists to avoid: declaring "no cascade" because lagging series are quiet while leading series scream, or "cascade over" because leading series calm while the lagging wave is still landing.

First-run tasks

Sources

Backlinks