Data-driven insights help SARS yield gains for fiscus

Simnikiwe Mzekandaba
By Simnikiwe Mzekandaba, IT in government editor
Johannesburg, 23 Oct 2023
A total of 89.4% taxpayers and traders used digital platforms to interact with SARS in 2022/23.
A total of 89.4% taxpayers and traders used digital platforms to interact with SARS in 2022/23.

The South African Revenue Service’s (SARS’s) data-driven and automated risk engines that detect non-compliance among taxpayers and traders are yielding results, says commissioner Edward Kieswetter.

This, as the tax authority attempts to become a smart, modern organisation with “unquestionable integrity, trusted and admired”.

Kieswetter last week presented SARS’s 2022/23 annual report to Parliament’s Standing Committee on Finance (SCOF), saying 96% of the compliance risks detected by the SARS automated risk engines are correctly identified.

“In the last year, SARS detected more than two million instances of potential non-compliance/fraud,” said Kieswetter, which resulted in excess of R79 billion added to the national fiscus. “These funds would not have been in the fiscus without the ability to detect risk in an automated way.”

The tax authority collects over 90% of the revenue required by government to serve the South African population.

Kieswetter stated SARS’s performance to date is an accumulation of a four-year journey that began in 2019.

Addressing SCOF members, the commissioner noted an increase in compliance through the automated detection of risk.

As a result, SARS is continuously building and improving its risk detection methodology and technology – data and artificial intelligence (AI), he said. “This is an important area that will continually receive attention, but will also require significant investment going forward.

“In the past, we had a lot of people employed to capture data and do manual assessments. Increasingly, more and more of that needs to be done by third-party data using AI [artificial intelligence] to do the work that traditionally people did and to augment the work of the more complex assessments that people still have to do.”

SARS commissioner Edward Kieswetter.
SARS commissioner Edward Kieswetter.

The revenue service has, over the years, introduced numerous digital channels to interact with taxpayers, including the MobiApp, Lwazi Chatbot, USSD channel, SMS service and auto-assessments.

AI chatbot Lwazi provided one million positive responses to taxpayers, said Kieswetter, noting 89.4% of taxpayers and traders used digital platforms to interact with SARS during the period under review.

Kieswetter described auto-assessments as an important area of ease of business, on which SARS has placed a huge focus over the last three years, utilising third-party data at its disposal and machine learning algorithms to produce auto-assessment outcomes.

For the year under review, SARS set itself a target of 90% auto-assessments, but achieved 94.7% (2.90 million people). During the 2021/22 period, SARS achieved a target of 91.03% for standard taxpayers’ auto-assessments.

“This is a huge testament to the integrity and usefulness of data science in our work going forward.”

SARS is looking to expand its datasets and increase the products produced from this data, according to Kieswetter. SARS’s suite of advanced analytics products now includes 20 machine learning models.

“Data is gold, but ultimately the value is what we extract from the gold. We have a long way to go, but we compare well with our international colleagues in terms of increasing our data science discipline.

“Increasingly building machine learning models, and improving our risk assessment capabilities, compliance and indexing capability are all work that feeds off data science. This also empowers taxpayers to help themselves and empowers our employees to help our taxpayers better.”

Personnel progression

According to Kieswetter, SARS’s capabilities encompass the technology and smart widgets deployed, as well as the people invited to be part of the SARS institution.

He pointed out that people are SARS’s most important asset. Kieswetter highlighted that the revenue service historically under-invested in people, but this has changed.

“We’ve appointed a chief learning development officer. We have reconceptualised the career formulation framework for our employees, using a five-year progression model and placing significant focus on investing in our own people.

“Slowly but surely, SARS is being seen as an employer of choice again. We hope we can continue to build our capacity by bringing in a healthy balance of provided opportunities for our own staff, but also refresh the so-called gene pool.

“In an organisation that has to manage in a complex and changing world, it’s important that we balance the institutional knowledge and experience of long-standing SARS members with perspectives that we can enrich ourselves with from the private sector.

“The world is changing, therefore our staffing model and skill profile have to be futureproof. The world of work is not going to be what we have lived through in the last decade or two; it’s exponentially changing and we have to be ready for that.”

The world is data-driven and increasingly augmented by AI, and technology platforms are the way to interface and engage with the public and stakeholders, he said.