Public-sector decision support for digital-agriculture rollout: Feasibility-constrained provincial sequencing in Vietnam

Abstract

Digital-agriculture programmes are increasingly central to public-sector modernization, but governments cannot usually roll them out everywhere at once. This article examines Vietnam as a territorially heterogeneous planning case and asks how public agencies can stage provincial rollout when productivity opportunity, equity need, low-greenhouse-gas (low-GHG) baseline advantage, and implementability do not peak in the same places. The study develops an ex-ante decision-support framework that links baseline forecasting, multi-criteria decision analysis (MCDA), exact 0–1 integer-programming portfolio selection, and regret diagnostics under weight uncertainty. The framework is not an ex-post causal evaluation of concrete policy packages; rather, it provides a transparent sequencing architecture for bounded public allocation. The results show that a neutral Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) baseline produces a territorially mixed priority set rather than a single readiness block, while feasibility-constrained optimization expands the selected frontier from 23 provinces under the narrow B0 envelope to 37 under B1 and 48 under B2. Robustness checks based on Dirichlet weight draws, rank-reversal tests, alternative feasibility proxies, and alternative MCDA methods show that the broad sequencing logic is not driven by one arbitrary weight vector or one isolated readiness proxy. The contribution is a practical public-sector planning framework for converting spatially misaligned objectives into defensible rollout tiers under administrative constraints.