When analyzing insurance-related data, actuaries must always be cognizant of the nuances within the underlying policies, particularly when it comes to their impact on Loss Development Factors (LDFs). The complexities they present may grow more pronounced when the policy type changes over time or has relatively unique parameters. Here, we address two critical questions which might be encountered and provide examples of possible solutions.

1: Transitioning Between Policy Types

How do you develop LDFs for a given risk when the underlying policy has changed from claims-made to occurrence, or vice versa?

The challenge presented by such a switch is indeed difficult but, thankfully, not insurmountable. Below are two potential approaches an actuary might take when preparing loss development triangles in this situation.

• Option 1: Segmentation by Policy Type

One option is to segment the policy periods in the client's loss history based on the type of policy written. Doing so would result in two separate sets of loss development triangles, each producing LDFs applicable to their respective policy type. This approach provides an accurate reflection of each policy by tailoring the resulting factors to their unique development patterns. However, an actuary taking this approach must keep in mind that each triangle would consequently have fewer data points to analyze, potentially reducing their overall credibility.

Example: Consider a company with claims-made policies from 2015 to 2019 and occurrence-based policies starting in 2020. In this case, an actuary analyzing their historical loss experience might create one set of loss development triangles for the 2015-2019 claims-made policies and another set for the 2020 and subsequent occurrence-based policies. Each set would then be reviewed separately to select their respective LDFs.

• Option 2: Restating Historical Data

Another approach involves restating historical data points in the loss development triangles to match the current type of policy. For example, if the current policy is written on an occurrence basis, prior loss run evaluations with policies written on a claims-made basis could be restated so that all claims are allocated to policy periods based on the loss/accident date instead of the report date.

Example: Let’s again consider the scenario referenced above. For a company that switched from claims-made to occurrence-based policies in 2020, an actuary might restate the 2015-2019 claims from prior evaluations as if they were under occurrence-based policies. This process would involve reallocating claims to the period the incident occurred in rather than when it was reported.

The approach taken by an actuary will vary depending on the situation and preference of the client. For example, SIGMA has worked with clients in the past who have voiced a strong preference for a specific approach, but we've worked with others who do not feel strongly about either option. Both scenarios provide important feedback for how we determine an effective, long-term strategy that incorporates client needs and objectives.

2: Retroactive Dates for Claims-made Policies

How are LDFs for claims-made policies impacted by the retroactive date?

Claims-made policies can add another layer of complexity to an actuary’s loss development analysis, due in part to the inclusion (or exclusion) of a retroactive date. While many policies of this type limit the timeframe on how far back reported claims can occur, some offer coverage on all reported claims, regardless of the event’s timing. The latter option presents an interesting possibility, as it opens the door to claims being reported from a much older time period. From a loss development perspective, however, this extended reporting lag does not necessarily impact future development patterns. Below is one way an actuary might determine this.

• Segmentation by Reporting Lag

They could, for example, segment the claims in a company's loss history based on reporting lag and then analyze the development patterns from each segment. If there is a significant difference, an actuary might make an adjustment to reflect the unique development pattern of older claims present in specific policy periods.

Example: Consider a company with a claims-made policy offering full prior acts coverage. This company has claims spanning several decades, and some of these are reported many years after they occurred. An actuary might choose to analyze the claims reported within five years of their occurrence date separately from those reported after five years. Despite very old claims being reported, if the development pattern of these older claims aligns with more recent ones, the overall impact on LDFs may be minimal. However, if there is a significant deviation, such as older claims having prolonged development, this must be accounted for in the analysis.

In Conclusion...

As you can see, actuarial analyses often involve decisions which extend beyond their purely quantitative aspects. These require an actuary to rely on their expertise of the risk being analyzed, their understanding of the overall objective, and their relationship with the analysis’ end user. Even for relatively straightforward concepts such as loss development, there can be several layers of complexity which an actuary must consider. Whatever approach they choose will likely have an impact on the projected losses, estimates of required reserves, and even the potential tail liability, so these considerations shouldn’t be taken lightly.

Because of this, it’s very important that those working with an actuary offer as much relevant information as possible when providing the data to be analyzed. Doing so can help strengthen the resulting analysis and ensure its usefulness as part of future risk management strategies.

If you need more information on this topic, please contact us at support@SIGMAactuary.com or schedule a call with one of our consulting actuaries.

© SIGMA Actuarial Consulting Group, Inc.