Public finances around the world are under pressure, forcing governments to look for innovative ways to preserve tax revenues and make services as user-friendly as possible.
On a recent webinarThe Global Government Forum, in collaboration with its knowledge partner SAS, convened a panel of four experts to discuss how governments are addressing the challenge of reforming tax systems, including through the use of data and AI.
Angela Render, chief content strategist for online services at the U.S. Internal Revenue Service (IRS), said she and her team spend a lot of time thinking about how to use data to make good decisions and deliver a better customer experience.
She said technologies such as AI offer “exciting opportunities” but added: “We need to treat AI and technology as a tool, not as some kind of magic wand that’s going to fix everything.”
That said, Render sees improving the search functionality of the IRS website as one of the biggest opportunities.
“The public expects search on the site to be as good as the big search engines, which is not the case right now,” she said.
“We’ve heard from many people that their journey on irs.gov begins with a commercial search engine that takes them to a specific page. If they need more information, they’ll try the site search, and if the results aren’t helpful, they’ll return to the commercial engine.”
AI can also help improve IRS service through voice bots and chatbots, which, although already in use, have not yet harnessed the potential of generative AI. Large language models (LLMs) such as ChatGPT have the potential to provide information to users on demand.
However, Render stressed that accuracy is imperative: “There may be industries where 95 percent accuracy is acceptable. For the IRS, it’s not.”
Render also highlighted the dual challenge of training AI and “ensuring that the source material is usable by both people and AI.”
“In the research I’ve done, both what I’ve read from others and what I’ve seen in my own testing, the source material matters,” she said. Fortunately, she explained, the IRS has learned that if it formats content in a way that’s “useful to people … structured and labeled for accessibility, and … written in clear, understandable language,” it’s easy for an AI to ingest and pass on to people.
“As we train AI, AI will also train us,” she said.
AI can also help identify actionable customer feedback, she concluded.
Collect revenues “fairly and efficiently”
Joseph Caruana, Commissioner of Malta Revenue and Customs, explained how the Government has set out a clear vision to collect revenue “fairly and efficiently by implementing best practices… and becoming an employer of choice in order to protect Malta’s borders and contribute to its economic and social well-being”.
Caruana stressed that taxes constitute “over 80% of government revenue”. This underlines the importance of collecting all taxes through voluntary compliance and ensuring that customs controls are applied effectively to “protect Maltese citizens, the economy and the environment”, in line with the modernisation model adopted by the government.
Malta is in the process of implementing a new system that will allow customs and tax to interact with each other. This system, combined with all the initiatives currently underway as part of the government’s digital transformation strategy, aims to make the government “more agile” in the use of data and more proactive in using technology to modernise its tax system.
But Caruana also touched on one of the key themes of the discussion: investing in people, not just technology. Without that, he said, “you’re not going to get the results you want.” Motivating and retaining the current workforce, as well as “investing in new blood” to transform services, remains a priority, he added.
Secret weapon
Shaun Barry, global head of fraud risk and compliance at SAS, said data analytics can be considered a “secret weapon” when it comes to transforming tax administration.
He elaborated by describing analytics as the secret weapon to which data provides the lifeblood or “oxygen.” Examples of where this approach is deployed include revenue forecasting, taxpayer services, cybersecurity and microsimulation, where governments use analytics to model changes in tax policy or legislation.
Sharing data is “not a magic bullet,” Barry said, but he stressed that using the data you have “is a powerful way to generate insights that can help tax administrators move forward.”
Adriano Subira, tax advisor to Brazil’s National Congress, explained how his country adopted widespread use of electronic payment methods, before explaining how the government is making life easier for taxpayers.
“The core of this tax reform system is assisted tax assessment and the massive use of intelligent electronic payments. Every tax code, every law, is an algorithm. We translate the law written in natural language into machine code,” he said.
Currently, every electronic invoice is authorized by the tax authority, but Subira said the government wants to give the taxpayer more independence to “register the sale, fill out the invoice and let the tax authority do the rest.”
Brazil is made up of what Subira calls three “levels or spheres”: the Union, 27 states and 5,500 municipalities. “We are like the European Union,” he joked. The country’s tax system reflects this complexity, with five different taxes “divided and distributed among the federated institutions – the Union, the states and the municipalities.”
The five taxes will be replaced by a dual VAT system over a 10-year transition period, ending in 2033. The reform will also shift the tax base from origin-based to destination-based taxation, meaning taxes will be applied where goods and services are consumed rather than where they are produced.
The distribution of revenues between the federal, state and municipal levels will be gradually adjusted over a 50-year period to ensure a fair distribution of funds under the new system. This transition is expected to reduce tax evasion, lower business costs and promote economic growth.
Subira said Brazil’s vision is to simplify tax administration, automate tax returns and refund deadlines, and improve compliance through a “government-to-platform” approach. This will reduce costs for the government and provide a better experience for citizens.
Data analytics has also been used extensively during the tax reform process to ensure the system remains simple and effective, with simulations carried out for proposed changes.
New skills
Speakers agreed that tax modernization is changing skills needs.
Barry said: “In recent years, tax administrators have had to understand technology and how it can be applied. They don’t need to be programmers, they don’t need to know every bit and byte, but they do need to be able to understand the concepts of using computers to automate.”
“The same thing is true today for analytics and it’s happening at a faster pace,” he said. “You need people coming into tax administration who aren’t necessarily mathematicians or programmers or who aren’t able to write code to create algorithms, but who need to understand the business concepts and how they apply to tax administration so they can guide and understand the strategy behind it.”
Caruana agreed: “I believe that with data-driven technology and compliance by design, we need people who are more focused on the taxpayer experience.”
The use of automation and data analytics will lead to a shift from process-oriented roles to more skilled analytical professionals who can make decisions based on insights from data analysis, he said.
Rewatch the webinar – Transforming Taxation: Creating a Better System for Governments and Citizens – here.
