The South African Reserve Bank published a Working Paper analysing South Africa’s Phillips curve over 2000–2024 and finding it is state dependent, with the inflation–slack relationship becoming materially steeper when inflation is high. The authors argue this non-linearity has implications for the inflation target, recommending a target low enough that routine supply shocks do not push inflation into a “red zone” where the public becomes more attentive and inflation expectations become harder to bring down. Using a self-exciting threshold autoregressive model, the paper estimates the attention threshold as a band between 4.28% and 9.29% (mean 5.55%). Below the threshold, the output-gap coefficient is statistically insignificant, while above it the Phillips curve slope is significant and larger than in linear estimates. A Markov-switching robustness check indicates low-inflation regimes are more persistent than high-inflation regimes (probability of remaining low 0.8811 versus remaining high 0.7883). In regime-specific shock analysis, an oil price shock has more persistent and broader inflation effects in the high-inflation regime, with a cumulative impact on targeted inflation of around 0.99 percentage points, while effects in the low-inflation regime fade after about one quarter. Based on typical oil price shocks, the authors estimate an inflation target of 3.37% would provide a buffer against entering the threshold range and state that their results support targeting 3%, while noting the paper’s findings are preliminary research and do not necessarily represent South African Reserve Bank policy.