The K-A-T Framework: Macro-Financial Propagation across Funding Constraints, Input-Output Networks, and Policy Backstops

Version 2 (2026-06) — supersedes v1 (citable at its version DOI). Identification transparency: the panel LP-IV is now explicitly reported as a DESIGN-CALIBRATED ILLUSTRATION, not real-data identification. The CIP-basis regressor is a calibrated composite proxy of real FRED TED-spread/VIX inputs (Avdjiev-Du-Koch-Shin 2019 design), and the high-frequency monetary-surprise instruments are SYNTHETIC placeholders calibrated to the published moments of the EA-MPD (Altavilla et al. 2019) and Nakamura-Steinsson (2018) series (the actual intraday datasets were not accessible for this draft); the first-stage F-statistics (305-484) and peak IRF (0.212 s.d.) reflect the noise-free calibrated construction and are not evidence of strong identification. Substantive conclusions rest on the REAL n=11 structural half-life law (exact permutation p=0.009) and the REAL Feb-Apr 2026 private-credit gate-wave event study / difference-in-differences (excess-spread 0.7pp, wild-cluster-bootstrap p=0.006; 5-day CARs -2.9% vs -0.7%) on genuine equity returns.

We propose K-A-T as a unified framework for macro-financial propagation linking (i) funding and collateral constraints (K), (ii) network amplification (A) via the Leontief inverse on input-output and financial-network adjacency matrices, and (iii) policy backstop credibility (T) modeled as a put option on funding costs. The framework yields a closed-form half-life expression HL = ln 2 / (lambda_T + mu_i), with lambda_T proportional to T, and the central testable prediction that as T -> 0 the half-life remains finite, bounded above by the natural decay rate. Validation is provided by a hierarchical Bayesian random-effects meta-analysis of ten historical T-shock episodes, with grand-mean half-life compression of -64.9% (95% CI [-75.5%, -49.8%]). The February-April 2026 private credit gate cascade is held out as a posterior predictive check. The K-A-T monolith combines theory, LP-IV panel evidence, the disclosure-anchored DFAST extension, and the CPCRF policy proposal in a single piece. Companion papers separate the theory and empirical contributions.