Issues in Applying Credibility to Group Long-term Disability Insurance Flashcards
Practical Applications of Credibility Concepts in Group LTD
Valuation
1. The 2012 GLTD valuation standard is a principle-based approach to reserving
2. The standard requires LTD insurers to fully or partially reflect company-specific claim termination experience in their valuation assumptions, provided the experience is credible
▪ Large LTD carriers with fully credible claim termination experience must calculate claim reserves using termination assumptions based entirely on company experience
▪ Mid-sized carriers, whose termination experience does not meet full credibility standards, can partially reflect their own experience when the experience is at least partially credible
▪ Insurers whose experience is not credible are expected to use 100% of the 2012 GLTD valuation table for calculating reserves
3. The weight of company-specific experience is determined from a limited fluctuation credibility model
▪ One of the underlying assumptions of this theory is that the random variables (i.e. claim terminations) are independent
▪ In reality, LTD claim terminations are not independent. Influenced by many factors like changes in claims management practices or the economy
▪ The non-independence of LTD claim terminations increases volatility and therefore reduce the credibility of claims experience
* “Selected variance factors” have been included in the limited fluctuation credibility model, to adjust credibility for non-independence
▪ The factors decrease with increasing claim duration because LTD terminations tend to be more volatile in early durations
* This is because terminations in early durations are dominated by recoveries highly correlated with cause of disability
* Also, benefits from other sources and the change in the definition of disability usually occur in early durations, and these are believed to impact claim terminations
▪ LTD claim terminations in the tail tend to be more stable, dominated by deaths as opposed to recoveries
▪ The structure of the “selected variance factors” ensures that fewer terminations are required to be considered fully credible in later durations of claim
4. Full credibility in the 2012 GLTD standard (i.e. the requirement to use 100% of company-specific experience) is determined from the number of expected terminations that yields an 85% probability that observed terminations are within 5% of expected terminations, taking differences in volatility by duration into account
Practical Applications of Credibility Concepts in Group LTD
Pricing of LTD Insurance Products
1. Experience Rating
▪ The weight given to historical experience is based on the credibility
▪ Most common formula for experience rating LTD products is: Premium Rate = (Credibility Factor) x (Experience Rate) + (1 – Credibility Factor) x (Manual Rate)
* When the experience is considered fully credible, the premium rate is based entirely on the experience rate
* When the experience is not credible at all, the premium rate is based entirely on the manual rate
* When the experience is considered partially credible, the premium rate is a weighted average of the experience and manual rates
▪ Credibility formulas used for experience rating LTD vary significantly across the industry
* Some are empirical, derived from experience studies using company-specific data,
* Others are based on traditional approaches such as limited fluctuation credibility
* The decision to use an empirical model or a more traditional model may be influenced by
- Limited claims experience
- Regulatoryrequirements
- Many factors that complicate the application of traditional credibility models in LTD (SEE “Challenges in Applying Credibility in LTD” section)
▪ Most credibility formulas used in experience rating consider the number of life years of exposure in the historical experience period
* The required number of life years for full credibility varies from insurer to insurer
* There are also differences in the credibility curves used by different insurers
* Some credibility formulas used in experience rating are based on claim volume as opposed to (or in addition to) number of life years, and some use expected claims whereas others use actual claims
* Some carriers use multiple credibility formulas to differentiate between voluntary and employer paid products, or to differentiate by geographic location
* Some insurers have different formulas depending on whether the historical experience was favorable or unfavorable
2. Manual Ratemaking
▪ Actuaries analyze LTD experience by gender, age, industry, and other variables to calculate pricing manuals. The choice of variables and how finely the experience should be divided is a subjective choice
▪ Actuaries use credibility to determine if experience is suitable for calculating manuals
▪ The methods to estimate credibility of experience for manual rate development range from subjective judgment to formal procedures using credibility formulas
▪ When the method is strictly subjective, actuaries make educated judgments to decide if the experience is credible enough for manual rate development
* If results are stable across multiple time periods, or if results are consistent with expectations based on other sources such as industry experience and/or professional judgment, then the experience may be deemed credible
* Thresholds could make the process more objective, for example by requiring a minimum number of policies or a minimum number of claims
* The advantages of using subjective methods include simplicity and flexibility
* One drawback is that it may be difficult to justify subjective decisions
▪ One example of a formal process to determine if LTD experience is credible for manual rate development is provided here. The example demonstrates limited fluctuation credibility concepts
Challenges in Applying Credibiliy in LTD
- Non-independence of claims
- Hetergeneous Claims
- Competitive Pricing Pressures
- Claim Duration
- Benefit from other Sources
- Outlier Claims
- Regulatory Requirements
- Estimating Parameters
1. Non-independence of claims
a. Most traditional credibility models, including limited fluctuation and least-squares models, assume that exposures in the experience are independent random variables
b. LTD exposures, often claims, are not believed to be completely independent
c. Economic factors, employment characteristics, industry-specific dynamics, and claim management practices all contribute to non-independence o Non-independence of claim occurrences (i.e. incidence)
▪ External factors such as work conditions or the economy affect many members of a group, and these factors are correlated with disability incidence
▪ Work conditions contribute to the well-being of employees. Physically demanding work conditions can trigger a disability for several EEs of the same group
* For example, suppose a group of firefighters suffer injuries while combatting a fire. The disabilities are not independent since they were caused by the same event
▪ Similar disabilities may be prevalent in some groups due to the nature of work. Group- specific dynamics contribute to the non-independence of disability claims
* Back injuries are common in industries where lifting heavy objects is part of the work routine
▪ Economic factors
* Unfavorable trends in disability incidence coincide with economic recessions
* Certain employees file claims during a recession (and would not file for claim
otherwise) because they fear being laid-off and having a total earnings loss
* Multiple claims can be motivated by a single event – in this case, a downturn in the economy
d. Non-independence of claim terminations
▪ Recovery trends are impacted by external factors like changes in claims management practices or changes in the economy
▪ Spikes in recoveries often occur when benefit admin resources are increased
e. How does non-independence of LTD claims impact credibility?
▪ One argument: a concentration of non-independent risks increases the credibility
* Disability trends among a group of construction workers are likely to persist due to the nature of their work. Then the risk class of construction workers may be considered “credible” due to the individual risks being correlated
▪ The other view: non-independence reduces credibility because it increases the volatility of the expected outcome
* Claim terminations are affected by changes in claims management practices, which makes them more volatile (and less credible). This was the approach taken in the development of the 2012 GLTD valuation standard
2. Heterogeneous Claims
a. Historical claims are used to estimate future claims by assuming the experience is homogeneous.
The underlying assumption is that claims will emerge similarly as in the past. Reasons why this assumption may not be true:
▪ Changes in the demographic mix of employees over time alters DI trends
▪ External factors like economic recessions can impact the experience
▪ Changes in underwriting or claim management practices can shift the experience
▪ Changes in plan design may result in different claims experience
b. Some ways to address these issues include
▪ Factoring in potential demographic changes
▪ Constant-rating historical premium in the experience calculation when plan designs change
c. Heterogeneous risk characteristics of the individuals in a group also complicates the application of credibility in LTD
d. Experience rating relies on using past experience to predict future experience.
▪ On the one hand, groups with more homogenous demographics could be viewed as being more credible due to their uniform risk characteristics
▪ On the other hand, groups with a greater demographic mix of employees may be considered more credible because there is greater spread of risk when compared to groups with more homogenous demographics
e. Credibility models based on expected claim volumes capture some of these group dynamics
f. Credibility models based solely on life years may not deal with these issues
3. Competitive Pricing Pressures
a. Competitive pressures make it challenging to apply credibility in LTD pricing
b. There is pressure to give the experience more credibility than what would be prescribed if a purely theoretical approach were taken, because it is difficult to increase rates on cases that have had good experience in a competitive market environment
4. Claim Duration
a. LTD claim durations range from one year to several decades long, depending on diagnosis, definition of disability, limitations, and other factors. This creates credibility challenges
b. LTD claim experience is more volatile in the early durations of claim for several reasons:
▪ LTD claim terminations in early durations are dominated by recoveries
* There is a strong correlation between recoveries and cause of disability in early durations, resulting in recovery patterns in early durations that vary significantly by cause of disability
* LTD recoveries are impacted more by claim management practices than deaths, resulting in terminations in early durations that differ from insurer to insurer
▪ Benefits from other sources are typically awarded within the first few years of claim, creating irregular payment streams in early durations
▪ The change in definition from an “own occupation” to an “any gainful occupation” definition usually occurs within the first few years of claim, resulting in a spike in recoveries
▪ The maximum benefit period for mental & nervous claims is usually 24 months
c. Claim terminations and benefit payments are more stable in the later durations, which results in reduced volatility. These dynamics make it difficult to apply credibility in LTD
d. Some credibility models address the issue by lowering the exposures required for full credibility in later durations, like the limited fluctuation model in the 2012 GLTD valuation standard
5. Benefits from Other Sources
a. LTD benefit payment streams typically change over time
▪ The approval of SSDI benefits or the loss of Workers’ Compensation benefits impacts net benefit amounts. This adds complexity to the credibility framework, because the expected severity of claims is volatile
b. The credibility of a block of LTD could depend on how many claims in the experience are SSDI recipients
▪ Claimants who receive SSDI awards are more morbid than claimants not awarded SSDI, because the SSDI definition of disability is more stringent than most contractual definitions in LTD
▪ Then the termination experience for claims with SSDI benefits could be different than the experience from claims that have not been awarded SSDI benefits
6. Outlier Claims
a. When the credibility formula is based on claim amounts, very large outlier claims artificially increase the credibility of experience
b. Even when the formula is not a function of claim amounts, outlier claims can complicate credibility
▪ In the pricing of LTD, very large claims shift the experience rate
▪ If the group is credible or partially credible, these outlier claims can create serious pricing challenges
▪ These claims are challenging because the probability of a claim of the same magnitude happening again could be very small, because maybe only a fraction of the group are
highly paid individuals, but nonetheless these claims do happen
c. Very large outlier claims can be dealt with in a number of ways in experience rating
▪ They can either be left in the experience or completely removed
▪ Must decide whether to adjust the credibility of experience that includes or excludes outlier claims. Outliers are unusual occurrences, so perhaps the experience that includes outliers should be considered less credible
▪ Another way of dealing with outlier claims is to implement risk pooling in the experience rating
* Risk pooling simultaneously caps claims in the experience that are above a threshold, and loads the rate in proportion to the exposure to risk
* This process could adapt the credibility formula to take pooling into account
7. Regulatory Requirements
a. Some states have adopted credibility requirements as part of the supervisory process
b. These requirements may apply to manual rate development, for example modifying existing pricing factors, or to experience rating
c. It can be a challenge for some carriers to meet the required credibility criteria
d. For example, in rate filings that involve changing a pricing factor for a provision, there may not be enough experience to support a rate change based solely on the company’s own data
8. Estimating Parameters
a. Estimating parameters like confidence intervals and full credibility thresholds is often based on a combination of subjective opinion and empirical testing
b. When credibility is based on a confidence interval, then the confidence factor (e.g. 85%) and allowable error (e.g. 5%) are usually determined subjectively. The choice may depend on:
▪ The credibility application (manual rate development versus experience rating versus valuation)
▪ The quality of the experience data
b. Empirical methods are typically used for estimating credibility thresholds and determining factors that influence the credibility of experience
c. Predictive modeling can identify the key drivers of experience, which can then be incorporated into the credibility framework