Frequency Factor Development
Similar to the severity thresholds, frequency factors/benchmarks must also be determined. You will have to decide on how to evaluate your data set (number of observations) in relation to the larger company or business unit setting. For instance, if a spill occurred at one of 10 operating locations, the frequency of spills could be assigned a value of 10% for the company or business unit. On the other hand, because an actual event took place within the period of the review, it could also be assigned a value of 100%.
Coordinate with your risk management department to align your frequency factors with those used by the company generally. If it is necessary to establish your own frequency factors, do so using a thoughtful and credible approach that is defensible and repeatable. Avoid simply defaulting to quartiles (0%–25%, 26%–50%, 51%–75%, 76%–100%). The ranges should reflect company-specific risk benchmarks.
The numerical values are necessary for the calculations that follow this step.
- Rare: The situation does not currently exist; no conditions were observed that are conducive to the risk developing and there was no identified trend related to the risk. This could be set at 1%–10%.
- Unlikely: The situation does not currently exist, but minor conditions were observed or documented that are potentially conducive to the risk developing and/or initial indications of a trend related to the risk appear to be developing. Examples include information indicating that regulatory enforcement or community activism activity may be developing. This could be set at 11%–30%.
- Likely: While the specific situation was not observed at the time of the site visits, conditions were observed that are conducive to the risk developing. An example would be a situation that is in compliance with legal operating requirements, but where situations are developing that create a potential for third party claims. This could be set at 31%–74%.
- Very Likely: The situation was observed as existing at the time of the site visit, or appeared imminent. An example would be a regulatory non-compliance or contamination observed during site visits. This could be set at 75%–100%.
As with severity thresholds, where ranges are used in frequency thresholds, it will be necessary to convert those to a single numerical value for purposes of the ROI formula. The midpoint of the range is appropriate. For instance, if you establish a range of 11%–30% for “unlikely,” 20.5% should be used.