Alternative model specifications can affect the estimated structure and dynamics of the business cycle, which has important implications for policy analysis. This paper evaluates the consistency of model estimations in the extant literature and tests the sensitivity of alternative models tailored to the South African economy. We find that both parameter estimates and model dynamics are sensitive to model specification. Our findings suggest that a three-equation New-Keynesian model and a traditional open economy model provide qualitatively and quantitatively similar results to the benchmark medium-scale New-Keynesian model with sticky prices and wages, habit formation, and investment adjustment costs. However, significant differences from the benchmark New-Keynesian specification are revealed once financial frictions are included. In addition, the types of exogenous shocks included in the model are key determinants for the variation of results.