Improving captive analytics through claim lag metrics

The COVID-19 pandemic has caused worldwide business operations to slow in recent months. The need for accurate, credible information, however, has not reduced. On the contrary, captives facing an uncertain future have likely found that such a need has grown exponentially.

Decisions regarding renewal negotiation and loss accrual are becoming increasingly frequent as the year goes on. As a result, loss projections are a common topic of discussion. These estimates of ultimate loss are associated with a specific risk and normally cover an upcoming 12-month accident or report period (depending on the policy structure). For captives, loss projections are used to determine captive premiums by risk and may also be used to negotiate excess pricing.

Since the underlying data being used to produce loss projections is currently in a state of flux, the resulting estimates could be discredited. Many captives in this situation may feel that they have no recourse, largely because they are unaware of analytic methods used to retain credibility. To alleviate this issue, we will be discussing one such method, which is to monitor emerging claim lag data and relay the insights gained into loss projection or accrual adjustments.

The insurance industry is overwhelmed with general ideas on how the pandemic’s impact might play out. While some underlying assumptions may differ, one point has reached consensus: pandemic-related data is still very immature. Immature data hampers analytic methods that rely on consistent, credible data, but it does not mean analytics cannot still provide valuable insights.

Understanding early trends in claims data can help organisations stay on top of their risk-related responsibilities.

The remainder of this article will focus on workers’ compensation, but these concepts can apply to other high frequency coverages as well.

Claim lag definitions

First, let’s explore the concept of “claim lag”. Many types of claim lag can exist, but the following are some of the most commonly examined:

1. Report lag: the difference between a claim’s occurrence date and the report date.
2. Closure lag: the difference between a claim’s report date and the close or settlement date.
3. Treatment lag:The difference between a claim’s report date and the first treatment date; or
   The difference between the actual treatment date and the date treatment was deemed necessary.
   

Regardless of the lag type being analysed, the most important aspect of a claim lag analysis is consistency. As long as data fields can be compared credibly to historical data, the insights provided will be similarly credible. Once a consistent basis of data is established, we can dig deeper into coverage-specific issues.

In workers’ compensation workplace injury claims may be reported quickly, but repetitive-use claims or latent disease claims may have much longer “report lag”. Similar types of claim lags may help in the early analysis of data in a rapidly changing environment.

Creating a comparison

To create a lag analysis based on emerging data, it may be necessary to create smaller accident periods to compare with historical information. Many companies track their workers’ compensation experience on an annual accident period basis matching policy inception.

However, it might be helpful first to compare data from the March to May 2020 period, or the April to June 2020 period, to a corresponding accident quarter at the same maturity. As an example, if the March to May 2020 period is being reviewed, a chart using the following format could be constructed to examine claim closure (Table 1).

The benefit of organising data in this way is that historical occurrence periods already have the appropriate quarterly data points. Once it is created, updating the chart becomes a relatively simple task of adding quarterly data for the current period. It is also worth noting that, even though Table 1 is examining claim closure, multiple data points can and should be examined in conjunction. Suggestions on additional data points are provided in the following section.

Above all, creating, updating, and reviewing these quarterly comparisons should lead to increased involvement from management in decision-making. Each quarter, this type of quantitative analysis can be supplemented with qualitative reviews to determine loss accrual adjustments. Management could decide that an initial 10 percent increase to the captive’s loss accrual is appropriate based on current analytics and qualitative items, but this amount may change over time as quarterly data points emerge and trends are identified.

Analysis and insights

Comparisons may be relatively simple to create, but without knowing potential trends and issues to identify, valuable information could prove elusive. As such, it is important to work closely with the claim management team during this process to ensure their expertise is fully used. One potential trend lies in the lag averages.

If these were to improve or decline significantly in the quarterly comparison, the next step would be to gain an understanding of the directional impact on loss projections and the impact’s potential magnitude. For example, a treatment lag increase could point toward an increase in the overall loss projection, as claims without early treatment intervention may be more costly.

Another example could be the improvement of claim settlement lags, which would likely indicate a possible decrease in the overall loss projection. Historical data could also be reviewed to help determine the magnitude of savings related to accelerated claim closure. Note, however, that this type of review may require more time to obtain meaningful insights, as settlements on immature claims will be low.

During this time, it may also be useful to analyse previously unavailable claim information. Claims specifically coded as COVID-19 or other emerging pandemic risks could be examined in bulk and compared with historical information in the manner outlined above. Doing so would help drive a better understanding of the ways in which pandemic-related claims currently behave and what programme-wide effects this may have in the future.

Captives with high-frequency/low-severity coverages may wish to track and analyse this information internally, but others may find themselves without the time, expertise, or resources to take on such a task. In that case, prompt discussions with an actuary may be necessary. Many actuarial firms are able to track this information from loss runs and provide their own insights and analytics in their actuarial reporting.

On the other hand, low-frequency/high-severity risks, such as those relating to events occurring every five to 10 years, may not lend themselves to quantitative claim lag analytics. For captives covering these types of risks, qualitative reviews could still prove useful in determining whether the effects of the pandemic impact overall frequency assumptions.

Irrespective of the type of information being gathered, it is important that it ultimately reaches the actuary, as it can play a significant role in decision-making regarding projected losses.

Conclusions

Since loss projections determine the premium set for each coverage within the captive, allowing the actuary to make fully informed decisions will prove vital to the captive’s long-term financial health. Should the loss projections be inaccurate or uninformed, actual losses may deviate substantially from expectations, meaning the captive could require additional capital.

It is also worth noting that this information could impact historical losses in a similar manner. Today’s discussions and suggestions are mostly concentrated on the future, but existing open claims (or claims with the potential to re-open) may undergo a great deal of variance in their medical or indemnity costs due to both treatment and closure lag. Monitoring information on an emerging and a historical basis makes a captive well-prepared to adjust as impacts from the pandemic become clear.

Michelle Bradley is a consulting actuary at Signa. She can be contacted at:  mb@sigmaactuary.com

Enoch Starnes is an actuarial analyst at Sigma. He can be contacted at:  enoch@sigmaactuary.com

Al Rhodes is president and senior actuary at Sigma. He can be contacted at:  al@sigmaactuary.com