SARS hunts for data science, AI-skilled talent
The South African Revenue Service (SARS) is slowly reclaiming its position as an employer of choice – attracting top skills, says commissioner Edward Kieswetter.
This comes after the tax authority experienced an exodus of highly-skilled personnel, in the aftermath of State Capture allegations.
Speaking yesterday on PSG’s Think Big webinar series, Kieswetter acknowledged the revenue service still has a long way to go in its rebuild process, but has made measurable progress.
He pointed to improvements in revenue collection, compliance and trade facilitation as indicators of a positive upward trajectory and signs of growth.
However, some constraints remain, notably the slow bureaucracy within the public sector as opposed to the private sector, as well as bringing people with technical specialisations into the organisation.
He explained: “I cannot compete with the financial and insurance sectors at the same level of remuneration, share and bonus allocation. I have to offer a different value proposition – a commitment to service, a commitment to a higher purpose.
“The skills we need at SARS [are] no different to the skills that Michael Jordaan would need in building a bank, or building a mobile communications company.
“Yet I cannot compete at the same level that they do, so it’s making it harder for us to attract the kind of skills; but you know what…we work at this. We have seen an improvement; we have seen the tide turning and slowly we will claw our way back.”
He revealed that SARS has, in the last year, brought onboard 400 to 500 additional skills in the technical area. In addition, it has just employed the first 50 of a cohort of 250 new service consultants.
“We’re doing multiple interventions to improve our attractiveness, in order to improve our efficacy as a revenue administration.
“Think about the level of data science technology and artificial intelligence we need…it is no different to any private sector organisation. Notwithstanding the challenges that we face, we see that as opportunities.”
As to timeline projections before SARS is where he’d like it to be, he stated: “I’ll be retired long before we get to there.”
According to Kieswetter, SARS’s approach to selecting cases of non-compliance is increasingly based on data-driven efforts and machine learning algorithms.
The revenue authority has been vocal about utilising emerging technologies and deriving insights from data for its tax collection efforts.
Yesterday, the commissioner explained SARS spends an inordinate amount of time on risk profiling. For example, last year, the tax authority collected just over 4.3 million value-added tax (VAT) returns, he revealed.
“It is humanly impossible to process every one of them with human effort,” he stated. “Through the risk selection and risk profiling artificial intelligence that we have built, we’re able to select only 10% of cases; that is, just over 450 000 instances that are selected for further audit.
“Of that, 92 out of every 100 VAT returns we don’t even touch because it’s not alerted for us. Eight out of every 10 is alerted for us. Of those that are alerted for us, we will review 71% and it is generally processed within the turnaround time. The 25-30% requires additional pouring of effort and asking for additional information. By just doing that work last year, we prevented the outflow of R42 billion impermissible fraudulent refund claims.”