Step 6: Sensitivity Testing
Sensitivity testing is required to ascertain that the conclusions are not the product of a particular set of assumptions or the result of a particular set of random scenarios. This step in the DFA process requires the testing of key input factors such as renewal rates, inflation and interest rate sensitivity of future premiums and liability payments, changes in capital market equilibrium assumptions, and variability of loss ratios. Sensitivity testing highlights the major factors affecting each business segment and the degree to which those factors affect each segment. Each factor needs to be tested independently, and relevant factors should be tested in tandem. Sensitivity testing allows for the assessment of the individual as well as collective impact of modifying key factors by business segment.
Since the underlying framework for DFA is simulation, sensitivity testing should include research into the number of simulations required to assure that the results of the analysis are robust. The required number of simulations will depend on many factors such as whether the analysis is dealing with relative comparisons or absolute levels. The metrics used for the objectives and constraints will also impact the required number of simulations. For example, downside risk measures typically require more simulations than simple standard deviations. There is no magic number or formula that tells exactly how many simulations are required for a particular analysis, so the user is left to ascertain, through sensitivity testing, that the findings of the DFA study are robust and can be easily reproduced.
Step 7: Presentation of Findings
The importance of the presentation of DFA findings should not be underestimated. While the DFA professional has the benefit of months of analysis in developing understanding of the problems, issues and solutions, they must summarize and present their findings to the senior management or Board of the company briefly and succinctly. This is no easy undertaking. The presentation of the DFA study should do more than show the numbers and present the conclusions, rather the presentation should tell a story. The story should review the highlights of each step of the DFA process and lay out the logic that went into the analysis in such e way that the conclusions become evident before they are revealed. It is important to keep in mind that the value of DFA is not just in the answer but also in the increased understanding of the issues that lead to the answer and ultimate decision.