Keywords = Fiscal Accountability
Sustainable urban infrastructure

A design science simulation of algorithmic verification for human capital augmentation in green fiscal governance

Articles in Press, Accepted Manuscript, Available Online from 09 September 2026

B. N. Ivansyach, A. W. Johannes

Abstract BACKGROUND AND OBJECTIVES: Green public financial management in decentralised fiscal systems is frequently constrained by limited administrative capacity to integrate fiscal, ecological, and digital expertise in expenditure verification. This study aimed to design and evaluate an Automated Ecological Verification framework intended to support fiscal-ecological verification capacity by embedding rule-based verification functions within budget information system architecture rather than relying exclusively on specialised personnel.
METHODS: A Design Science Research Methodology comprising six sequential development steps was applied to design a Semantic Translation Layer enabling real-time, bidirectional interoperability between fiscal expenditure classification codes and an ecological outcome registry. The artefact was evaluated through simulation using a synthetic regional government budget dataset of 240 transactions distributed across five programme categories. Ground-truth labels were assigned independently of the engine's classification output, and projected outcomes were assessed against a manual verification baseline and a deterministic environmental quality index model. No expert practitioner validation, user acceptance testing, field piloting, or real APBD transaction data were used at this stage.
FINDINGS: The AEV validation engine achieved an average classification accuracy of 97.1%, ranging from 95.8% to 97.9% across programme categories, and correctly identified all simulated instances in which environmentally suggestive expenditure labels concealed technical specifications that violated predefined ecological thresholds. Twenty-two of the 240 transactions (9.2%) required structured human adjudication. Under the Full Compliance scenario, the model produced an upper-bound projected environmental quality index improvement of 8.5 points. These findings represent simulated classification performance and scenario-based projections rather than observed administrative, fiscal, or environmental outcomes.
CONCLUSION: The findings demonstrate the technical feasibility of the proposed AEV architecture under controlled simulated conditions. The framework may reduce reliance on routine manual verification while preserving specialised human judgement for context-dependent cases; however, the present study does not establish improvements in staff workload, capability, decision quality, organisational learning, or operational governance effectiveness. These outcomes require validation through real APBD data, expert assessment, user testing, and field implementation.