US Interest Burden × CAMS Stress Architecture

Does CAMS structural stress anticipate fiscal over-commitment? A first empirical test.

Kari Freyr McKern · July 2026 · 30-year panel (1996–2025) · FRED (OMB) × CAMS5 ensemble

1. What Was Tested

The core claim of the US oligarchic-capture analysis is that a merchant-financier coalition has progressively locked in fiscal structures that favour asset-holders, increasing background systemic stress even when headline GDP growth looks acceptable. Interest payments as a share of federal revenue (IPR) is one of the cleanest external proxies for fiscal pre-commitment — it measures how much of the state's capacity is pre-allocated to servicing past obligations before a single new coordination function can be funded.

The test was three-part: (1) contemporaneous correlations between CAMS variables and IPR across 1996–2025; (2) lead-lag analysis — does CAMS Stress rise before IPR ramps, or only alongside/after?; (3) AR(1) baseline models with and without CAMS augmentation, including a walk-forward out-of-sample test (train ≤2020, test 2021–2025).

Fig 1. US Interest Payments as % of Federal Revenue, 1996–2025 (FRED/OMB)

2. IPR: What the Series Actually Shows

The IPR series has a distinctive double-valley shape. It peaked at 16.6% in 1996, declined through the Clinton surplus era to a trough near 8.5–9% in 2003–2015, then re-accelerated sharply — 14.8% (2023), 17.9% (2024), 18.5% (2025). The 2025 figure now exceeds the 1996 opening level of the CAMS panel. The trajectory from 2022–2025 is the steepest three-year rise since the 1980s high-rate environment.

3. Contemporaneous Correlations

CAMS Variable	r (with IPR)	p	Reading
Mean Bond Strength	+0.372	0.043 *	Higher institutional coupling → higher IPR (surprising)
Shield Node Value	+0.380	0.038 *	Military-security dominance correlates with fiscal pre-commitment
Hands Node Value	+0.317	0.087	Labour capacity → positive association (ambiguous)
Mean Node Value	+0.277	0.139	Moderate positive; not significant
Stewards Node Value	+0.280	0.134	Governance capacity → positive (fiscal regime coherence?)
Mean Capacity	+0.308	0.098	Near-significant; mirrors Clinton-era high-capacity high-IPR
Mean Stress	−0.303	0.104	Higher stress → lower IPR (inverse; fiscal stress ≠ CAMS stress here)
Archive Stress	−0.178	0.346	Weak inverse; not significant
Archive Node Value	+0.128	0.502	Near-zero; Archive not driving contemporaneous IPR
Stress Dispersion (SD)	−0.111	0.559	Not significant
Reactivity Ratio	−0.026	0.891	Essentially zero

Key Finding — Contemporaneous

Mean Stress correlates negatively with IPR (r = −0.303). This is the opposite of what a naive "stress predicts fiscal burden" story would predict. The explanation lies in the series shape: IPR was highest in 1996 when CAMS Stress was relatively low (the Clinton-era stability period), then IPR fell as stress began rising post-2001, and only now in 2022–2025 are both moving together. CAMS is not narrating what the debt office already knows — it appears to be measuring something orthogonal to the fiscal-burden cycle in the contemporaneous window.

4. Lead-Lag Analysis

Fig 2. Lead-Lag Correlations: Mean Stress & Mean Bond Strength vs IPR (lag in years)

Variable	lag −3	lag −2	lag −1	lag 0	lag +1	lag +2	lag +3
Mean Stress	−0.503*	−0.346	−0.326	−0.303	−0.338	−0.144	+0.190
Mean Bond Strength	+0.705*	+0.475*	+0.422*	+0.372*	+0.379*	+0.181	−0.118
Mean Node Value	+0.625*	+0.397*	+0.321	+0.277	+0.311	+0.117	−0.207
Archive Node Value	+0.693*	+0.332	+0.192	+0.128	+0.197	+0.099	−0.176
Stewards Node Value	+0.330	+0.290	+0.284	+0.280	+0.282	+0.115	−0.129

* p<0.05. Lag = CAMS year minus IPR year; lag −3 means IPR leads CAMS by 3 years (fiscal stress precedes CAMS response); lag +1 means CAMS leads IPR by 1 year.

Lead-Lag Finding — The Direction of Causality

The strongest correlations are consistently at negative lags (fiscal leads CAMS, not the reverse). Mean Bond Strength at lag −3 is r = +0.705 (p<0.05); Archive Node Value at lag −3 is r = +0.693 (p<0.05). This means high IPR years predict high CAMS Node Values and Bond Strength 3 years later — the causal direction appears to run from fiscal conditions to CAMS structural readings, not the other way around.

This is the clearest finding in the study: CAMS stress variables do not lead the interest-burden cycle — they follow it. Periods of low fiscal burden (low IPR) in the late 1990s–2010s coincided with or preceded CAMS deterioration; periods of high fiscal burden (high IPR, 1996 and 2023–2025) coincided with or preceded higher CAMS Node Values and Bond Strength.

Interpretation

High federal interest burdens appear in CAMS periods of relatively high institutional coherence (Clinton-era, and now 2023–2025 where Stewards and Archive remain moderate). The fiscal pre-commitment story is not being captured by a simple CAMS Stress → IPR mechanism. Instead, the pattern is consistent with a different reading: the capacity to carry high interest burdens (measured by CAMS variables like Bond Strength and Node Value) precedes periods where that capacity is fiscally exploited. This is structurally interesting — it suggests CAMS may be measuring the conditions that enable debt accumulation to persist, rather than the stress it generates.

5. Predictive Models

In-Sample R² Improvement over AR(1)

Out-of-Sample RMSE (train 1996–2020, test 2021–2025)

Model	In-sample R²	ΔR²	RMSE	β(CAMS)
AR(1) baseline	0.764	—	1.431	—
AR(1) + Mean_BondStrength	0.783	+0.020	1.370	−0.098
AR(1) + Archive_V	0.779	+0.015	1.385	−0.122
AR(1) + Mean_Stress	0.774	+0.010	1.400	+0.267
AR(1) + Stewards_V	0.767	+0.003	1.421	−0.053
AR(1) + Reactivity_Ratio	0.764	0.000	1.430	+0.359

Model	OOS RMSE (2021–2025)	vs AR(1)
AR(1) baseline	3.416	—
AR(1) + Archive_V	3.347	−0.069 ✓
AR(1) + Mean_BondStrength	3.385	−0.030 ✓
AR(1) + Mean_Stress	3.447	+0.031 ✗
AR(1) + Stewards_V	3.525	+0.110 ✗

Predictive Finding

Archive Node Value and Mean Bond Strength provide modest but consistent out-of-sample improvement over AR(1) (RMSE reductions of 0.069 and 0.030 respectively, test period 2021–2025). Mean Stress and Stewards slightly worsen OOS performance. The signals that carry independent information are structural coherence variables (Bond Strength, Archive), not stress variables. The in-sample ΔR² gains are small (≤0.020) — CAMS adds marginal signal on top of the AR(1) baseline, not transformative predictive power.

6. The Dual Trajectory (Visual)

Fig 3. IPR vs Mean Stress vs Mean Bond Strength — Dual Y-Axis, 1996–2025

7. Jobs Worth Doing Next

Debt-to-GDP and primary deficit. IPR is only one fiscal test. Debt-to-GDP has a different trajectory (monotonic rise since 2001) and may correlate more cleanly with CAMS stress. Run the same panel against GFDEGDQ188S (FRED debt-to-GDP quarterly, annualised).

Node-specific deep dive: Stewards × Archive interaction. The oligarchic-capture hypothesis predicts that Archive (legal/administrative encoding) and Stewards (governance) should diverge — Archive remaining formally coherent while Stewards degrades — precisely in periods of high IPR. The contemporaneous correlations don't yet capture this interaction term. Worth testing Stewards_V × Archive_V as a predictor.

Regime-conditional analysis. The lead-lag structure may differ by JUNO regime. Split the panel: Stable Adaptive years vs Strained years and run separate correlations. This would test whether CAMS only picks up fiscal deterioration signal during structurally vulnerable periods.

Cross-national comparison. Run the same IPR × CAMS analysis for Germany and Australia (both have long CAMS panels). If the lag structure is reversed in those cases (CAMS leads fiscal), it would suggest the US-specific pattern reflects the oligarchic-capture mechanism — not just a universal CAMS lag artefact.

Legitimacy and conflict indicators. Next frontier: Gallup institutional trust, electoral volatility, and Varieties of Democracy indicators. These are the downstream effects the oligarchic-capture argument predicts; CAMS Stress should lead those more directly than it leads fiscal variables.

Verdict

CAMS does not lead the interest-burden cycle. The fiscal series leads CAMS by up to 3 years — high IPR is followed by higher CAMS Node Values and Bond Strength, not higher stress. This is not a null result; it is a structurally interesting finding. It suggests CAMS is measuring the institutional capacity that enables debt accumulation to be sustained (and politically managed), not the point at which fiscal over-commitment generates visible systemic stress.

The predictive tests show Archive and Bond Strength add marginal OOS signal (RMSE −0.069 and −0.030 over AR(1)), confirming these variables carry some independent information about the IPR trajectory. But the gains are small and the stress-based variables do not outperform.

The strongest falsification condition for the oligarchic-capture thesis now becomes: CAMS stress should lead legitimacy and conflict indicators more clearly than it leads fiscal ones. That test should be run next.