Singapore has the least complicated system for paying taxes in the world, according to a 2019 study by UNSW Sydney and consulting firm KPMG. How did the country achieve this?
The strategy is to design around taxpayer needs. “We believe that most taxpayers will voluntarily comply when we can design a tax system and processes that are easy to comply with,” says Cheng Hui Yi, Director of Insights and Solutions in the Compliance, Strategy and Insights Division in Singapore’s tax body, the Inland Revenue Authority of Singapore (IRAS).
Cheng’s team uses data science and behavioural insights to design simple processes and make paying taxes easier. She shares how this approach has helped IRAS encourage good behaviour and detect non-compliance more effectively.
Led by user experience
Data analytics can tell us a lot about what is happening, but it doesn’t tell us why. “Other than analysing the data, we need to understand why taxpayers behave in a certain way and what their behavioural motivations are,” Cheng says. “The combined use of data and a deeper understanding of the human (using design and BI) enable us to come up with solutions that are more effective.”
IRAS used this approach to upgrade the process for filing tax clearances. Employers in Singapore have to file tax clearances for foreign employees who have stopped working for them, or plan to leave Singapore for an extended period of time.
Cheng’s team used design thinking principles to understand the needs of the users and create an ideal experience of filing tax clearances. They conducted ethnographic research with some employers and employees to find out current pain points, co-created more than 50 ideas and tested these prototypes with users. One result of this is a simplified, pre-filled form so employers can amend details more quickly.
IRAS will also redesign its tax filing portal to simplify the flow and make it more intuitive, Cheng says. The new design will run from end-2020, and is expected to benefit more than 140,000 employers and employees. “We can expect higher filing compliance rates as more taxpayers would be able to fulfill their tax obligations easily,” she says.
Using AI to pre-empt queries
Some taxpayers fail to follow the rules only because they don’t understand them. IRAS plans to use AI to make paying taxes more straightforward, Cheng shares.
The algorithm will anticipate questions or issues that taxpayers might have. IRAS can then proactively reach out to confused taxpayers, or publish helpful guidelines on their website to pre-emptively resolve queries, she explains. AI chatbots can also help to answer taxpayers’ questions on the IRAS website.
On top of this, the tax agency will use AI to detect fraud, Cheng says. For instance, AI can process tax returns and automatically flag out suspicious cases. This will point investigators to potential fraud cases, some of which might not be picked up by humans, she says.
Building data science capabilities
Data forms the basis of a lot of IRAS’s services. “With the huge amount of data running on our systems, there are numerous opportunities for us to leverage the data available to better understand our taxpayers and improve organisational efficiency,” Cheng says.
IRAS recently launched a unified data platform that provides advanced analytics services. When tax officers receive complex, technical queries over email, the analytics engine recommends the most relevant response so frontline officers can reply quickly and accurately. The platform also allows IRAS to store, process and analyse more complex data formats, including textual data from emails.
The data platform improved decision-making and streamlined the internal reporting process. Data is fed directly into management dashboards, so senior management can make decisions based on up-to-date data. This has saved 120 man-days a year for the agency, since employees no longer have to manually report every detail of their work, Cheng says.
IRAS will continue to build data competencies amongst staff. It will focus on making data easily available, providing user-friendly data visualisation tools and data skills training, Cheng shares.
“The goal is to enable all staff to make data-driven decisions,” she adds.
Tech can help. The agency will explore more advanced AI techniques, such as using anomaly detection techniques to find unknown compliance risks, and using deep learning and natural language processing to draw insights from textual data, she says. Cloud-based analytics services and open source tools can help to “lower the barrier of entry to data science”, she notes.
IRAS is even restructuring its organisation to use data better. In addition to a central analytics department, each division will have an analytics team, so they can deploy new projects more rapidly with a good understanding of on-the-ground issues, says Cheng.
Paying taxes can be a complicated, frustrating and long-drawn process, but it doesn’t have to be. With a combination of data analytics and behavioural insights, Singapore’s tax agency will continue redesigning its system to create a simple, seamless process.