Many companies don’t believe that their retained liabilities are big enough or complex enough to warrant hiring an actuary. Therefore, internal calculations are often completed using spreadsheets or various software tools. In some situations, these calculations may give a reasonable range of retained liabilities. However, the magnitude of the losses, the number of years involved, the complexity of the risk, the reporting lag, and other nuances may lead internal calculations to produce indicated results that are significantly different from those provided by an actuary.
These differences may be highlighted in several specific instances when an internal analysis comes under actuarial, audit, or third-party scrutiny. Examples include:
- A new auditor for a firm requiring an external opinion from a qualified actuary
- An existing auditor deciding that the internal calculations do not produce indications within a reasonable range or do not use reasonable methods
- The individual in charge of the internal analysis leaving the company, causing them to switch to an independent actuary
- The company being part of a merger or acquisition, causing the internal calculations to be compared to external independent calculations in the due diligence process
- The company hiring a new CFO who wants validation of internal calculations and methodology
Internal calculations can cover a broad spectrum, ranging from sophisticated methods or software tools to “back-of-napkin” calculations. In some instances, companies hire consultants to review their internal calculations and offer advice on methodology. Other times, the calculations use only one method with little consideration given to common actuarial methods or approaches. Regardless of the type of methods used, the results may differ from an actuary’s indications. The areas that typically account for the biggest differences are detailed below.
One of the most common problems found in internal analyses occurs in the estimation of ultimate losses for the current year. The current year is normally defined as the year that is in progress and has not reached policy expiration (normally 12 months of development). These are some common problem areas:
- Use of loss development factors on a period that has less than 6 months of development, as development methods are generally unreliable for periods at that level of maturity
- Inappropriate interpolation of loss development factors prior to 12 months of development
- Development of a claim at retention occurring very early in the policy period
- Not considering the current period at all when estimating the outstanding liabilities (required reserves)
- Reserves should be estimated as of an accounting date, which normally includes a provision for the partial current period. This could exclude several months of exposure if not considered.
Another common issue is for an analysis to be completed without exposure information. For example, if no exposure information is available, the implicit assumption is that each historical year has an exposure which remains constant in size. Loss rates by exposure are important in the projection of future losses, especially for workers compensation if payroll can be provided by class code. Therefore, the implicit assumption of constancy can lead to loss projections that are not credible.
The area of loss development factors generally creates the biggest issues. In internal calculations, standard development methods using loss development factors should be considered for most risks. There are several potentially problematic areas in the selection of these factors:
- Use of industry loss development factors when data is available to compile unique factors
- Use of unlimited data instead of data limited to the appropriate retention
- Use of countrywide benchmark data for workers compensation when a state-specific mix is more appropriate
- Giving too much weight to unique factors when historical patterns are not yet fully credible
- Use of general rules of thumb instead of actual loss development factors (e.g. adding a fixed amount of estimated IBNR, or using a “rough” IBNR-to-case ratio)
- While this might not sound like a loss development factor issue, the determination of estimated ultimate losses by rough rules of thumb produce implied loss development factors.
As the retained risk grows in terms of claims volume and retention, so does the risk of internal calculations not being comparable or close to actuarial indications. Another SIGMA blog by Enoch Starnes, entitled “Transitioning from Loss Forecaster to a SIGMA Analysis”, provides information for clients using our own software tools that may need or want to transition into a full actuarial analysis. The information provided in that blog can be considered for a company looking to move from any internal process to a formal actuarial engagement. If you have any questions, feel free to contact us at support@SIGMAactuary.com.