The paradox of unemployment benefits refers to the unintended consequence where government-provided financial assistance for the unemployed may contribute to persistent high unemployment rates by reducing incentives to seek employment. This study investigates this phenomenon in South Africa by analyzing historical unemployment trends from 1991 to 2022 using an autoregressive integrated moving average (ARIMA) modeling approach. Utilizing time-series data from the World Bank, unemployment (% of the total labour force) (modeled ILO estimate), serves as the dependent variable, while autoregressive (AR) and moving average (MA) components function as independent variables. Parameter estimation, conducted via conditional least squares (CLS), reveals that the MA(5) coefficient (0.857637) is positive and statistically significant, indicating that approximately 86% of unemployment variations can be attributed to past errors or economic shocks. The estimated ARIMA(1,1,5) model is found to be covariance stationary and invertible, confirming its robustness for forecasting unemployment trends. Projections indicate a persistent rise in unemployment, with rates expected to increase from 29.7% in 2024 to approximately 38.8% by 2043. These findings highlight the need for a policy re-evaluation to balance social welfare support with effective labour market incentives. We recommend reforms in unemployment benefits, enhancing labour market policies, and implementation of skills development programs.