The Forced Urgency Gap: Why loss aversion is not identified in the states that matter — and what it means for an AI labor transition

webpage: The Forced Urgency Gap: Why loss aversion is not identified in the states that matter — and what it means for an AI labor transition Publication date: 2026-07-25 Creators: Nogueira Grossi, Pablo — Affiliation: G6 LLC — ORCID: 0009-0000-6496-2186 Version: v1 Language: English License: Creative Commons Attribution 4.0 International (CC-BY-4.0) Reserved DOI: 10.5281/zenodo.21561819 Files to upload (2): Forced_Urgency_Gap_deposit_v1.pdf — the paper (CC BY 4.0) ForcedUrgency.lean — the machine-checked amplification proof (MIT) Description Working paper — not peer-reviewed; self-archived on Zenodo. Behavioral finance attributes household selling in downturns to loss aversion. This paper shows that attribution is not identified: when a household's liquidity state is unobserved, contractually forced sales — margin, redemption, and loan-to-value triggers — are observationally equivalent to preference-driven "panic," so the estimated loss-aversion coefficient is a mixture of preference and compulsion whose compulsion share peaks in exactly the crisis states where the coefficient is applied. This wedge is the Forced Urgency Gap. The gap is identified through tests where preference and compulsion make divergent predictions: a buffer-depletion lag; cross-sectional sorting on liquidity rather than loss magnitude; and two natural experiments — the 2007–09 global financial crisis versus the 2020 COVID shock — using FRED and Federal Reserve Z.1 data, 1971–2026. Forced supply is absorbed by unconstrained buyers with a harvest coefficient ρ, giving cumulative displacement A = 1/(1−ρ); concentration ratchets monotonically. AI labor displacement is shown to be an urgency shock of the same class, reducing "will the transition compress or explode the wealth distribution?" to two measurable coefficients — the contract share and ρ — rather than to sentiment. The amplification result (Section 4) is kernel-verified in Lean 4. The included file ForcedUrgency.lean, run with lake env lean against a built Mathlib (Lean v4.33.0-rc1), reports for all six theorems: #print axioms → [propext, Classical.choice, Quot.sound], with no sorryAx. The empirical coefficients — the value of ρ, the identification failure, the buffer lag — are deliberately NOT formalized and remain open pending estimation. Every claim is tagged: [DATA] (observed in FRED/Z.1), [MODEL] (derived within the framework), [OPEN] (not yet established), or [VALUE PREMISE] (an explicit normative choice). Author: Pablo Nogueira Grossi (G6 LLC), ORCID 0009-0000-6496-2186. Keywords forced urgency gap; loss aversion; identification; latent liquidity constraint; fire sales; harvest coefficient; wealth concentration; AI labor displacement; household finance; behavioral finance; bridge liquidity; macroprudential policy; formal verification; Lean 4; Mathlib Additional notes Accompanying code (ForcedUrgency.lean) is licensed under the MIT License. The Section 4 amplification result is machine-checked in Lean 4 / Mathlib (Lean v4.33.0-rc1); reproduce via lake env lean ForcedUrgency.lean against a built Mathlib. This is an independent working paper (G6 LLC); it has not been peer-reviewed. Related/alternate identifiers (optional) If you want to link your other Zenodo records (e.g. the Tribonacci substitution-chain paper, 10.5281/zenodo.20230642), add them under "Related identifiers" as "cites" or "isSupplementTo" — only if genuinely related; leave blank otherwise.