# TA-TR-2026-14 v1.3: revision and primary-source audit

Date: 2026-09-22. Internal AI-assisted review and substantial revision by GPT-6 Astra Pro under Hongju Liu's direction. Not external peer review or a global-priority certificate.

## Version lineage and scope

The published v1.1 DOI 10.5281/zenodo.22871209 and v1.2 DOI 10.5281/zenodo.22885976 remain unchanged. Version 1.3 belongs to original concept 22871208 and remains paper 14, not an additional numbered paper. Its new results cannot be backdated to the first edition. The three Bitcoin Originals are outside the modification scope.

## Scientific changes

1. Replace the main constant-expenditure-share restriction by two increasing homothetic consumption-ratio schedules. Derive a target-indexed scalar resource equation and the associated disposable-income share. Prove monotonicity in the zero-labor proportional-ownership benchmark rather than assuming it.
2. Classify asymptotic support shares under power-law demand tails, including all exponent ties. The limit can be zero, an interior value, or one. Recover the old Cobb–Douglas limit as a special case; do not present its positive limit as a general consequence of fixed scarcity.
3. Separate a target-supporting tax from a minimum tax and an all-equilibria guarantee. Prove a sufficient finite-labor monotonicity condition, then give an exact complementary-demand counterexample with at least three equilibria at the same tax.
4. Preserve v1.2's population-quantifier, monetary-gap, residual-provision, and asymptotic-boundary corrections. Do not revive earlier overclaims.
5. Add an independent forward-demand root calculation, exact rational counterexample sign tests, all budget/resource checks, and optional symbolic derivative identities. These are internal checks, not empirical data, calibration, or independent replication.

## Novelty boundaries

The broad question is old: abundant aggregate output does not by itself identify access; ownership and entitlements matter; prices respond to demand and asset income. General-equilibrium multiplicity and restrictions that restore uniqueness are also classical. Theorem 1 uses standard household optimality and market clearing; its novelty must not be equated with the first proof of a welfare or uniqueness theorem. Proposition 3 is a sufficient condition, not claimed to be the weakest condition in the literature.

The residual candidate contribution is the specific inverse access calculation, the zero/interior/one demand-tail comparison, and the quantified implementation-versus-selection failure test assembled for this question. An earlier equivalent result would reduce originality to a reproducible synthesis and application. This review does not establish that future economic models must cite this paper, that its results are paradigm-changing, or that a particular journal would accept it.

## Primary sources consulted and access limits

- Song Zichen (2026), *Abundance Is an Expenditure Condition: AI Automation, Scarcity Rents, and Access*, arXiv:2609.13801v1, dated 12 September 2026. Full primary HTML: https://arxiv.org/html/2609.13801v1 . Read the model, equilibrium result, ownership decomposition, redistribution and institutional comparisons, and limitations. It already links automation expenditure, rents, heterogeneous ownership, and access in a closed three-good model. It separates consumption guarantees from welfare. These are direct constraints on broad originality, not independent support for this paper's particular frontier.
- Båge and Wilson (2026), *Pinning the Wage to Scarcity and Technology*, SSRN abstract 7226858: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7226858 . The indexed primary record lists posting on 10 August and revision on 4 September. The predecessor's preserved source audit documents full-text comparison with the expanded cross-post https://clawrxiv.io/abs/2609.02873 , including its subsistence and redistribution appendices. Retrieval of that cross-post was intermittent in this revision; a complete fresh reread is not asserted. Those inherited overlaps remain explicitly disclaimed rather than silently removed.
- Geanakoplos and Walsh (2016), *Uniqueness and Stability of Equilibrium in Economies with Two Goods*, Cowles Discussion Paper 2050, August 2016. Primary institutional PDF: https://cowles.yale.edu/sites/default/files/2022-08/d2050.pdf . Inspected the title page, introduction, assumptions, and relevant discussion of homotheticity, endowment effects and uniqueness. The paper studies a different two-period preference structure. General multiplicity and uniqueness cautions are inherited; our exact tax counterexample is not a first discovery that equilibrium can be nonunique.
- Comin, Lashkari and Mestieri (2021), *Structural Change With Long-Run Income and Price Effects*, Econometrica 89(1), 311–374, DOI 10.3982/ECTA16317. Primary publisher metadata and abstract: https://onlinelibrary.wiley.com/doi/abs/10.3982/ECTA16317 . Used to locate the nonhomothetic income-effect literature. A complete full-text proof audit of that paper was not performed. Version 1.3 remains homothetic and does not claim to solve those income effects.
- Korinek and Lockwood, *Public Finance in the Age of AI: A Primer*, NBER 34873, DOI 10.3386/w34873. Primary author listing: https://www.korinek.com/research . Accessible primary manuscript: https://www.brookings.edu/wp-content/uploads/2026/01/Korinek-Lockwood-FINAL-for-website.pdf . The Brookings file is a December 30, 2025 manuscript issued January 2026; it is not silently treated as identical to the February NBER edition. The relevant inherited issue is financing and the changing fiscal base, not the new frontier's proof.
- Jones (2026), *AI and Our Economic Future*, JEP 40(3), 3–22, DOI 10.1257/jep.20261505. The publisher supplies final bibliographic metadata. The retrieved author PDF at https://web.stanford.edu/~chadj/AIandEconomicFuture.pdf was marked May 18, 2026, version 2.0; exact identity with the final publication is not asserted. Used for inherited bottleneck and factor-price distinctions.
- Sen (1981), DOI 10.2307/1882681; Acemoglu and Restrepo (2018), DOI 10.1257/aer.20160696; Ray and Mookherjee (2022), DOI 10.1016/j.red.2021.09.003. Retained foundational references and bounded attributions from the predecessor's primary-source audit, not a claim that their complete proofs were re-audited here.
- Liu (2026), v1.2, DOI 10.5281/zenodo.22885976. Full predecessor manuscript and audit read from the repository and preserved build artifact. First-party prior edition, not independent corroboration.

Search used title/author and targeted homothetic-demand, redistribution, access-frontier and equilibrium-uniqueness queries, with primary publisher, institutional and author manuscript sources. Several broader queries returned irrelevant material; that is not evidence of absence of prior work. This is a targeted review, not a systematic-review protocol or exhaustive literature search.

## Critical review sequence

Round 1: verify actual v1.2 publication and merge, read the complete manuscript, and distinguish valid limited results from broad novelty.

Round 2: locate the additional close Song manuscript; narrow the claimed contribution. Derive the general inverse equation and compare substitution regimes.

Round 3: test a proposed finite-labor extension. Monotonicity failed under strong complementarity; replace the overbroad claim by a sufficient condition and preserve an explicit multiple-equilibrium counterexample. An initial numerical bracketing point was corrected from 0.1 to 0.05 after exact sign testing; the final proof and code use the verified bracket.

Round 4: verify derivatives symbolically and budgets through independent forward demands; run the complete deterministic audit. Exact results and counts are in the generated checks.json, not inferred from manuscript wording.

Round 5: inspect the exact DOI-bearing PDF and package before publication, record a hash-bound review gate, and verify anonymous public readback after release. This source note does not pre-assert those later gates; their receipts are separate.

## Current quality assessment

A self-contained, reproducible theoretical working paper with a candidate incremental contribution, not an established foundational theory. Compared with v1.2, the principal advance is removing one restrictive demand form and identifying where inverse implementation does not imply guaranteed access. The mathematical structure is deliberately small enough to be checked. Static endowments, two groups, two goods, homotheticity, idealized collection, and omitted investment/behavioral responses materially limit application. Neither numerous assertions nor DOI registration removes those limitations.
