Sichuan's tax authority has pushed big data into its internal supervision work under a four-step approach: find issues in the data, trace them to people, trace people to institutional gaps, then lock down authority through rules. A provincial team combining discipline inspection, audit, risk management, inspection and tax policy staff runs the programme on a 1+7+N structure, building indicator models around invoices, funds, authority and responsibility and scanning anomalies from the Golden Tax Phase IV system, the e-tax bureau and internal control platforms. More than 70 models have been built, 75 are running in production, and since 2023 the work has produced 269 leads, 119 disciplinary actions and eight detentions. A second-round review keeps grassroots workload down: of more than 5,000 leads sent from the national platform, only 726 were forwarded after filtering, and the rest were graded by high, medium and low risk. Findings are also fed back into the rules, which are written directly into business systems so that abnormal operations are blocked.
News Detail
Sichuan Tax Authorities Use a Four-Step Method to Align Smart Supervision with Smart Taxation
Sichuan's provincial tax bureau has built big-data indicator models into its internal supervision work, following a four-step method that runs from spotting anomalies in data to tracing individuals, institutions and power controls. A joint team drawn from discipline inspection, audit, risk management, inspection and policy departments runs the programme under a 1+7+N structure, modelling risks across invoices, funds, authority and responsibility. Since 2023 the province has built more than 70 supervision models, 75 of them running in production, generating 269 leads and 119 disciplinary actions, with eight people placed under detention. A second-round review filters false positives, cutting 5,000-plus leads from the national platform down to 726 pushed to grassroots units.
2026-10-30