View Point
In My Opinion

Alternate Claims Maturity Metric – It’s About Time

By Grover Edie
Posted on Business and Technology Opinion

It is about time we, the experts on property-casualty loss reserving, came up with some additional means of measuring the maturity of losses.

When I first got into this business, the passage of time was the sole dimension of claims maturity being used. That was long ago. The data we get from our company or clients is often only what is required to be reported on the NAIC Annual Statement. And even when we go beyond that data, we usually use whatever data is already available.

Arranging historical loss information by accident year and calendar age is generally acceptable. Where appropriate, however, our professional standards require us to modify the “standard” techniques by suitable adjustments; the goal is for our techniques to respond better to changing or unusual environments. I suggest that one of the parameters of historical loss data that could be improved is “age.” There could be a host of reasons why the “maturity” of data is not properly reflected by calendar age: e.g., an acceleration in claims reporting or changes in case adequacy because of a change in claims handlers. Be it suspicion or fact, proving or disproving that “something happened” to affect the maturity of loss data is often a difficult, if not impossible, task.

When I first got into this business, the passage of time was the sole dimension of claims maturity being used. That was long ago.

 

In order to measure true claims maturity, we need to develop metrics that would enable us to verify the following:

  • Material changes in the manner in which losses are reported — not just delay (time) between the incident and its reporting, but the level of detail provided. More early detail usually means the loss will be further along in the claims handling process than an identical claim with less information.
  • Changes in claims personnel that cause an improvement or deterioration in claims processing.
  • Changes in the legal environment requiring an attentive change in some component of claims handling, such as payment, first notice response, or other issues.
  • Changes in the underlying characteristics of risks, causing future frequencies or severities (or both) to be materially different from historical data.

Ideally, such metrics would be more responsive to changes in underlying conditions, alerting decision makers and managers to possible changes in claims handling, underwriting or pricing.

Some of these metrics could be based upon information the claims department is already compiling. Examples of such information are:

  • Caseload per adjuster — closed counts per unit of time, outstanding counts, outstanding amounts, count of claims in suit, average age of claims in inventory.
  • Average call times per adjusting unit (department or type of coverage).
  • Average call wait time per adjusting unit.
  • Delay between occurrence and reporting to company, both averages and extreme values.
  • Potential for salvage or subrogation. Important for physical damage lines, this could be an absolute or a scored value.

There is unlikely to be one metric that fits all. Each line of business or claim type is likely to require its own.

For liability lines, another factor would involve some sort of indicator of just how aggressive the attorney representing the plaintiff has been in the past. We score credit for underwriting and premiums, so why not score the plaintiff’s representative for purposes of loss reserving? Furthermore, an attorney engaged on a claim is often not discovered by the claims adjuster until a suit is filed. Can we get information predicting the likelihood of attorney involvement before the suit is filed? Additional data elements might also be helpful.

For workers compensation, the number of visits to a health care practitioner might be a measure of the claim maturity. The type of health care practitioner visited might also be an indicator.

For automotive liability, the Insurance Institute for Highway Safety’s occupant safety rating of the plaintiff’s vehicle might have a predictive value. Perhaps knowing whether the injured individual was wearing a seat belt would help predict “time to heal,” and thus future medical costs and the time until costs are finalized.

For personal injury protection and medical coverages, some sort of rating concerning the vehicle’s loss propensity would be valuable. If an insured drives a safer car, will it produce lower or higher medical costs? Will the velocity of the claims handling process be different with safer vehicles?

The answer to all of this is in predictive modeling and the insights it can provide into relationships buried within mounds of data.

 

The answer to all of this is in predictive modeling and the insights it can provide into relationships buried within mounds of data. We have a lot of very smart, talented and experienced people in our Society. If we put our heads together, we could come up with some additional measures of claims maturity that would benefit us and our companies and clients.

Many of us just finished doing yearend loss reserving for our companies and clients. What can we offer in terms of alternate maturity metrics?

Hopefully, the CAS’s current initiative to create a predictive analytics credential and promote research in data science will help answer some of these questions.

Grover Edie, FCAS, is the AR Editor in Chief.