Tax enforcement by algorithm
Tax compliance and enforcement appear on 118 recent papers, and 23 of them carry a digital, big-data or AI keyword, led by China's digitalized tax administration. Tax authorities now match invoices, bank data and platform records automatically. Which taxpayers change behaviour, and does automated enforcement crowd out voluntary compliance?
What prompted this
Since October 2025, 118 papers carry a tax-enforcement, tax-compliance, tax-administration or tax-audit keyword (53 in economics journals, 40 in finance, 15 in accounting): "tax compliance" on 49, "tax enforcement" on 34, "digitalization of tax administration" on 9, "Golden Tax Project" on 6. Twenty-three of the 118 also carry a digital, big-data, machine-learning or e-invoicing keyword.
Questions
- When electronic invoicing becomes mandatory, how much does reported revenue rise, and does the effect fade as firms learn the system's blind spots?
- Does algorithmic audit selection target the right taxpayers, and does it create perceptions of unfairness that reduce voluntary compliance?
- Do firms respond to data-driven enforcement by moving to cash, to informality or across borders?
Data that could help
Firm-level tax data around the phase-in of electronic invoicing (China's Golden Tax Phase IV, e-invoicing mandates in Italy, India and Latin America), audit selection records where available.
Next step
Use the staggered roll-out of mandatory e-invoicing across regions in one country to measure changes in reported revenue and VAT.