Insuring Long-term Care Chap 6: LTC Experience Monitoring Flashcards
Introduction
- Nature of LTC
- Aspects of LTC that make it distinct from Life/Medical
1. Nature of LTC product complicates performance analysis – specifically as it takes a long time for claims to emerge
- Assumptions must be projected many years into the future
▪ Persistency – portion of policyholders continuing coverage
▪ Probability of meeting benefit triggers for LTC claim
▪ Morbidity – length of claim and portion of benefits used
2. Aspects that make LTC experience analysis distinct from Life or short-term Medical
- Long duration
- Limited credibility
- Persistency and morbidity focus
- Claim severity not known at onset of claim
Data Needs and Considerations
1. Start with insurer’s own experience
2. Main components for LTC experience analysis: policy coverage history and claim history
3. Data Sets
a. Policyholder Data
▪ Characteristics by policyholder (age at issue, DOB, marital status, sex, policy number)
▪ Policy coverage features (benefit period, inflation protection, benefit trigger, etc)
▪ Daily benefit amount
▪ Coverage start and end date
▪ Policy termination reason (if applicable)
b. Claims Data
▪ Dates of service
▪ Claim status
▪ Payment amounts
▪ Care setting (e.g. skilled nursing, assisted living facility, home health)
▪ Assumed claim reserve for each open claim
* Use data to summarize policy persistency and details related to claims incurred
4. Characteristics of LTC block of business that affect suitability for assumption setting
- Size – larger number of policyholders and claims gives credible persistency and morbidity quicker
- Age – older avg age leads to more LTC claims and more credible mortality data
- Duration – longer duration leads to more underwriting “wear-off” and more LTC claims and insured mortality (most policies require insured to pass medical underwriting)
5. LTC relatively new compared to other medical products
- Partial or limited credibility – new product and long-tail nature of product
- Insurers often supplement own data with industry datasets from SOA or consulting firms
Use for Experience Studies
1. Use of studies depends on purpose – can be experience monitoring/reporting or assumption development for future projections
a. Usually rely on actual to expected comparisons
▪ Actual – reflects company data
▪ Expected – based on assumptions from industry data
b. Detail and granularity depend on intended use
2. Experience Monitoring and Reporting
a. E.g. Want to figure out why actual costs are higher than expected
▪ More policyholder claims, staying on claim longer, or utilizing more benefits?
b. Actuaries– want to understand current experience and help future experience
c. State Regulators – protect constituents from insurer insolvency
▪ Review rate increase requests and review Actuarial Guideline 51 submissions
▪ Help decide if rate increases are warranted and assess financial stability of insurer
d. Both parties seek understanding of where and why experience deviates from current
assumptions, credibility and future implications of deviations
▪ Time and budget constraints may necessitate higher level analysis
3. Projection Modeling
a. Require persistency and morbidity assumptions to estimate future claims and premiums
b. “Total Life” approach
▪ Status (healthy vs disabled) not explicitly tracked
▪ Assumptions apply to all lives
▪ Incidence, continuance and utilization not explicitly in model
▪ Details are aggregated and expressed as a “claim cost”
- Insurers apply specific adjustments (from actual to expected analysis)
o Apply to entire claim cost, instead of specific component
- Persistency assumptions also applied to total lives
c. “First Principles” approach
▪ More popular in recent years
▪ Track policyholder status and allow for more detailed modeling
▪ Separate persistency assumptions for active and disabled lives, separate morbidity and
interaction between persistency and morbidity can be modeled
▪ Allows for more detailed company-specific adjustments
Types of Experience Studies
- Claims Incidence
- Claim Termination Rate
- Utilization
- Policy Persistency
1. Helps to assess reasonability of developed assumptions and adjustments
a. Validate whether assumptions developed at more granular basis will produce reasonable answers when combined together
2. Claims Incidence
a. Estimated number of claims to be paid
b. Summarizes rate at which insureds commence claims
c. For consistency of actual and expected incidence – actuary must determine definition of a claim in the dataset
▪ Some may record a claim at incurral date, others may record at date of payment (after elimination period)
▪ Exposure basis also important
- Often based on active (healthy) exposure base (note – a policyholder currently on claim can’t start a second claim)
- Some models use total life incidence rates
d. Actuary must decide if historical incidence patterns are expected to continue/are appropriate
▪ “Shock incidence” – adverse selection that may occur following a rate increase
- Healthy insureds leave plan and less healthy population stays
e. Incidence Rate = Number of Claims / Active Exposure = Number of Claims / Years Inforce-Years on Claim
3. Claim Termination Rates
a. Rate at which policyholder terminates their claim
▪ Through death or recovery
▪ Benefit exhausts (end of benefit period) – often excluded, as these are calculated within projection models
b. Often studied on monthly basis
c. Compared against industry continuance tables
d. Continuance = 1 – Claim Termination Rate
▪ Continuance and claim termination often used interchangeably
e. Main variables – claim start and end dates, indicator variable of open or closed claim
▪ Denominator is claim exposure (time spent on claim)
▪ Numerator is claim terminations
f. Actuary must understand data – does data contain claims terminations in the elimination period?
4. Utilization
a. How much of the benefit is used
b. Often reimburse only on actual expenses incurred (up to a daily limit)
c. Utilization studies sometimes called “salvage”
▪ Salvage = 1 – Utilization Rate
d. Measure proportion of policy benefits paid out while insured is on claim compared to max possible payout
e. Utilization savings – either in dollars or days
▪ Dollars – refers to cost of LTC services, relative to daily benefit
▪ Days – refers to number of days of services received
- Home health services not always received every day
- Skilled nursing facility and Assisted living facilities usually assume 100% utilization as claimants reside in the facility (no day’s salvage expected)
f. Calculated at aggregate level (across claimants)
g. Expected claims = Max Possible Paid Amt x Expected Utilization
h. Similar analysis should be performed on other splits of the business where utilization tends to vary (e.g. inflation protection)
i. Total utilization – typically only pertains to indemnity or expense reimbursement benefits
▪ Cash policy benefits – pay max benefit for each day member is benefit eligible
▪ Indemnity – pay max benefit on days member receives service (only days utilization is
applicable)
▪ Expense reimbursement – subject to days and dollar utilization
5. Policy Persistency
a. LTC is lapse supported – i.e. lapses help, thus fewer policy terminations (lapses) than expected can make it less profitable
b. Persistency patterns help provide number of policyholders, age and risk composition of population
c. Policy coverage terminated in a few ways:
▪ Active – stop paying premiums necessary to keep policy in-force
- Voluntary lapse, active life death, non-forfeiture election
▪ Disabled – utilize entire benefit (benefit exhaustion) or dies (disabled life death)
d. Policy conversion to non-forfeiture status – not a termination, but still of interest to insurers
▪ Policyholder no longer pays premiums
▪ Accrues benefit equal to premiums paid to date (significant reduction in potential liability
for insurer)
▪ Conversions following a rate increase studied to determine future behavior in response to rate increases (called a contingent non-forfeiture option – CNFO)
e. Assumptions needed for mortality and lapse (for both Total Life and First Principles projections)
▪ Distinctions between lapse and mortality can impact projections
- Mortality increases with age and duration
- Lapse rates reach an ultimate level and remain relatively flat thereafter
f. Type of persistency used based on availability of historical data and credibility
g. Total Termination Studies
▪ Based on aggregate data, doesn’t distinguish between deaths, voluntary lapses or benefit exhaustion
▪ When historical data is unavailable or lacks credibility for reason of termination
▪ Developing assumptions from aggregate data – balancing act between quantifying impacts from mortality and lapse
- Lapses can vary more significantly by product and policyholder characteristics than mortality
h. Develop “implied lapses”
i. If deviation by age still persists, likely due to mortality and not lapse
j. Mortality and Lapse Studies
▪ Separate mortality and lapse studies to explicitly reflect the patterns
▪ Mortality underreporting
- When policy is unpaid and lapses, insurer doesn’t always know if it’s a voluntary lapse or a death
- For CNFO where no more premium is paid, can be difficult to track if insured is still alive (and eligible for benefits)
▪ Total life models – don’t distinguish between active and disabled status
- Lapses unlikely to occur when policyholder is receiving benefits
- Total life lapse rate should reflect underlying mix of active vs disabled
Conclusion
- Monitoring experience – key facet of LTC product management
o Insight into how morbidity or persistency contribute to patterns - Used to inform future results through pricing and projection assumption development, current rate increase requests, and demonstrations of current experience diverging from initial assumptions
- Internal management, regulators and consumers – all have vested interest in performance of LTC block of business