Two-Layer Transactions as Commitment Technology: From Price Wars to Verifiable Positive-Sum Quality Competition

Platforms compete through intertemporal promises backed by residual claims (such as revenue shares, redeemable rights, or equity-like user claims). When settlement of these claims is discretionary and cannot be independently verified, such promises become cheap talk and competition tends to collapse into price or subsidy wars, with predictable quality degradation and fragility. This paper introduces Two-Layer Transactions (TLT), where each transaction produces both a standard price receipt and a claim receipt, and a claim becomes asset-like only if it is admitted into a verifiable finality outcome by a preset cutoff. The paper formalizes a minimal notion of settlement for the claim layer, called LTL, and provides an operational acceptance framework, AX-GATE, built around a strict “fail-closed” principle: if verification fails at the relevant window, no claim is finalized. When issuance caps can bind, the paper shows that deterministic trimming can create a selective-disclosure problem, and that preventing multiple publicly consistent “final” outputs requires a completeness-binding commitment of the candidate set before finalization. In repeated interaction, verifiable finality supports compliance incentives and generates an endogenous credibility state that disciplines claim valuation. Using a standard spatial competition benchmark, the paper shows that higher credibility shifts rivalry away from pure price competition and toward promise quality and product quality, and it characterizes when equilibrium quality rises with credibility. These comparative statics are shown to be robust to outside options and to settings with partial or asymmetric market coverage. Welfare implications are stated through an explicit decomposition across key stakeholders and penalty terms, yielding conditions under which clean admissibility becomes endogenous. Finally, the paper emphasizes that, given identical canonical verifiable inputs at the cutoff, finality outcomes are implementation-independent.