In increasing operational complexity within the calculation chain, it has become critical to assess the sensitivity of output results to individual assumptions and inputs involved. We provide the methodological basis of sensitivity analysis and demonstrate, the differentiated behavior of Sobol and Shapley indices, and the performance of the Quasi-Monte-Carlo estimator in that context. Topics in the paper include:

• Model risk management
• How sensitivity analysis can address model risk management
• A deep dive into sensitivity analysis
• Estimators for sensitivity indices
• Generation of random numbers