Global switch to indirect tax set for boost from Artificial Intelligence propensity to thrive on big, defined data models
The 30 year+ trend of adopting VAT, GST and other indirect taxes over direct taxes (corporate income tax; payroll taxes etc) has long been justified by the economic argument. Don’t tax or disincentivise vital income generation; instead tax consumption. (Although watch out wealth taxes!).
Source: OCED Tax Revenue Statistics 2023
This trend has a new ally: Artificial Intelligence – both machine learning and generative AI. Early successes with tax authorities adopting AI and machine learning have shown that indirect tax collections can be vastly boosted by AI’s ability to exploit well-defined and curated data. Since VAT is made up of millions of daily transactions governed by largely similar rules – much harmonised by the EU Directive and the OECD guidelines – AI has a deep reservoir of transactions to learn from and monitor for trends, errors or fraud. AI-driven algorithms and machine learning techniques come into their own in analysing, interpreting, and deriving actionable insights for tax authorities from large and complex VAT transaction datasets. And, as global e-invoicing and EU ViDA extend this pool of insights down into every single transaction, the power of the VAT and AI alliance will only grow.
This compares to direct taxes, particularly corporate income taxes, which remain deeply heterogeneous and subject to constant tinkering by governments. And AI itself will erode taxes on payrolls as it replaces many basic jobs and ballooning welfare support.
Indeed, tax authorities are way ahead of in-house VAT departments’ use of AI
You can see how we at VATCalc are already using it to generate AI VAT advice and item classification or tax codes to get the right taxing rates, country, exemptions and return reporting.
AI VAT case judge and jury
But what about the complexity of VAT case law – all those ECJ cases with shades of rules and interpretation? Well AI is cutting through this too. Projects such as Prodigit and Moonlit are able to instantly analyse VAT and similar tax case law against live transactions to give likely interpretations and rulings – including % chance of winning a court challenge.
Total VAT?
So will VAT’s conquest of direct taxes be complete? No, unlikely. Not least because taxes are not just about raising government revenues. They are about policy – social, economic and more. Governments use them to influence – reward and curb – behaviours of taxpayers, companies, populations. And that goes international, too. For example, cutting corporate tax rates to attract job-creating inward investment. Or lowering salary taxes on certain categories to promote employment objectives.
So, for government policy reasons, direct taxes still have a few years shelf life at least.
AI’s impact on jobs and inequality leaves wealth taxes in the mix
One exception to the dash from direct to indirect taxes will also be capital on wealth and capital. VAT is unlikely to fill the entire fiscal gap, particularly because AI will eliminate many basic jobs thus requiring an expanded social welfare. Also, governments are going to have to fund massive education programmes to prepare people for new skills and jobs.
So tax authorities will have to strengthen taxes on capital and wealth, particular Capital Gains Tax and pension earnings.
How tax authorities are using AI in VAT
Key areas where AI can assist tax authorities with VAT administration
- Risk Assessments: analysing huge volumes of sales and purchase transactions of taxpayers based on various factors such as historical data, industry benchmarks, and risk indicators. This helps tax authorities prioritize their resources and focus on high-risk areas that require closer scrutiny. In terms of the VAT Gap, this can include sudden surges of cross-border, intra-community transactions from new businesses, or abnormally large VAT deduction claims. This helps prioritise VAT audits and enforcement actions.
- VAT fraud exposure: unearthing unusual or suspicious transactions, especially fraud-prone EU intra-community supplies. By analyzing vast amounts of data, AI algorithms can flag irregularities that may indicate fraudulent activities, such as false claims, tax evasion, or identity theft.
- Large data analysis: AI, based on cloud-computing, can live process massive data loads to uncover potential fraudulent transactions. By adopting ML algorithms, tax authorities can detect patterns and anomalies in VAT returns, transactions, Intrastat, EC Sales Lists and similar, helping them identify potential areas of concern for further investigation.
- VAT payer guidance: latest natural language tools can provide AI VAT advice to taxpayers without the need for large teams of VAT-expert. chatbots or virtual assistants can be used by tax authorities to provide automated responses to common VAT queries. This can include:
- VAT registration rules;
- Returns obligations;
- Deadlines;
- Rights to deductibility; and
- International VAT rules
