This paper aims to examine whether the WTO Agreement on Subsidies and Countervailing Measures (SCM Agreement) and the Anti-Dumping Agreement (ADA) can meaningfully discipline a new class of trade-distortive practice: AI-powered algorithmic pricing supported, financed or directed by States through national AI strategies, data-access regimes and compute subsidies. This study proposes a new conceptual framework for treating AI-enabling inputs as countervailable financial contributions.
This paper adopts a doctrinal and conceptual methodology, combining black-letter analysis of WTO case law on subsidy, public body, benefit, specificity, normal value and constructed value with comparative review of three national AI policy ecosystems (China, the European Union and Türkiye) and economic literature on algorithmic collusion. It tests existing doctrines against the technological architecture of contemporary algorithmic pricing systems. The doctrinal analysis is anchored in an explicit theoretical framework that combines institutional theory, the structure–conduct–performance (SCP) paradigm of industrial organisation as applied to multi-institutional legal settings by Ghosh and Mukhopadhyay (2020) and regime-complex and polycentric-governance.
The SCM Agreement’s closed list of financial contributions, its goods/services dichotomy, its market-based benefit benchmarks and its public body doctrine are structurally ill-suited to capture state support for AI-enabling inputs (compute, data, foundation models and energy). The ADA’s normal-value methodology, its “particular market situation” doctrine and the constructed-value rules following EU – Biodiesel produce arbitrary or under-inclusive results when applied to algorithmically priced exports. Without doctrinal renewal, AI-mediated subsidisation will fall through the gaps.
To the best of the author’s knowledge, this paper offers the first integrated SCM/ADA analysis of algorithmic pricing as a state-supported phenomenon and articulates a five-part reform agenda including a new SCM sub-category for AI-enabling inputs, an ADA Article 2.2 amendment, a plurilateral Joint Initiative, enhanced notification disciplines and capacity-building for developing Members.
