Group 39, Group 40 & SN103 Claim Reserves Flashcards
Claim Reserve Definition
Estimate remaining to be paid on claim incurred on or before valudation date
Unpaid Claims Liability
- IBNR
- Unpaid claims in excess of IBNR
- ICOS, Resisted Claims
- Due & Unpaid
- Outstanding Accounting Feeds - CAE / LAE
1. Incurred but not Reported (IBNR)
- Loss date before valuation date but not reported yet to insurer
2. Unpaid Claims in excess of IBNR
a. Reported by unpaid
ICOS, Pending Claims, Resisted Claims
- Reported but not yet adjudicated
- Estimated by: (1) Average Claim * Inventory Count, (2) Large claims on seriatim basis, (3) Calculate IBNR + ICOS then split
Due and Unpaid (D&U)
- Reported, adjudicated, but not yet paid
- Estimated by: (1) Average claim * claims in check cycle, (2) Individually if there are only a few
Outstanding accounting feeds
- Overlaps with due & unpaid
- Estimated by: (1) Amounts in accounts payable system, (2) Averages
b. Open Claims - Benefits currently being paid
c. Present value of amounts not yet due (PVANYD)
3. Claims Adjustment Expenses (CAE), Loss Adjustment Expense (LAE)
- Internal cost of claims adjudication
- Estimated as % of unpaid claim liability
Estimating Incurred Claims
Economic & External Influences
- Unemployment, economic conditions
- Epidemics, Catastrophes
- Price trends
- Changes in medical practices
- Changes in contracts with providers
- New laws
- Cutoff dates for risk adjustment reporting
Claims Administration, Internal Company Practices
- Staffing levels (layoffs, vacations, weather events)
- Time required for complicated claims
- Computer system changes or downtime
- Seasonal backlog of claims submitted
- Cash flow considerations
Insurance characteristics
- Marketing and Underwriting affects the type of risk accepted
- Pattern of growth & Maturity of blocks influence incurred claims - (1) New bocks have longer lags and lower claims, (2) Mature blocks have shorter lags and higher claims
- Type of Benefit - (1) DI long runout, (2) Major medical fast runout, (3) Life insurance paid within a few months of death
- Incurral dating methods - must be internally consistent (e.g. between actuarial and claims area)
Plan Provisions
- Elimination Periods
- High Deductibles
- Pre-existing condition limitations
- Benefit maximums
Reserves Cells
- Reserves calculated by each homogeneous category of business
- Too many cells - (1) increases estimation error for small cells, (2) Increases time and resources
Margin for Uncertainty
- Conservatism in claims liability estimates
a. Development Methods - implicitly in choice of factors, or explicitly add margin to best estimate reserve
b. Tabular Reserve - (1) Margins applied to assumptions of tabular factors, (2) STAT use explicit margin, (3) modify standard tables with own credible experience + PAD
c. Projection Methods - margins in trend assumptions
d. Loss ratio projections - margins in loss ratio assumed, or explicitly add margin to best estimate reserve
- Consider multiple estimates and run-out studies
Data Integrity
- Common errors in claim data include:
a. Missing data
b. Misstated age or gender
c. Inaccurate Elimination Period or Benefit Period
d. Inaccurate information on benefit integration
e. Inaccurate cause of disability
f. Incorrect coding of claim status
g. Inaccurate incurral date
Product Specific
- Managed Care and Value-based Payment Arrangements
a. MC may require pre-certification for large claims, which helps with IBNR calculation
b. Consider initiatives or discounts affecting:
i. Utilization Levels
ii. Lag factors
iii. Charge levels
iv. Claim reserves
v. Provider Risk Sharing - if provider meets criteria, insurer may owe provider a payment at end of each year
Important Aspects of LTD and LTC
a. Periodic Benefits
- Monthly or daily amount
- LTD (indemnity) and LTC (actual expenses up to specified benefit amount)
b. Long-term benefit periods may extend many valuation periods
- LTD often age 65, LTC often a maximum length of time or dollar amount
c. Elimination Periods - time after incurred date before benefits accrue
d. Inflation benefits COLA
e. Offsets (aka Integration of Benefits) - LTD typically social insurance integration. LTC typically Medicare LTC
f. Limitations and Exclusions
g. Partial disability benefits
h. Interest Rates
- STAT - use interest rate for whole life for policies issued in year of incurred date
- GAAP - based on companies own expected investment income rate on its assets.
i. Claim Expenses - Claim expense reserve equals % of tabular claim reserve.
j. Reserves for closed claims
- May hold a percentage of full claim reserve when a claimant has begun legal action against a company’s decision to close a claim
- Also some closed claims will reopen and receive benefits again for same disability. May hold a percentage of reserves on recently closed claims
k. Lump-sum, fixed, or variable payments
l. Criteria for benefit eligibility
m. Waiver of Premiums - LTD often no extra reserve. LTC usually an additional reserve.
Disability Income Benefit Claims
a. Continuance tables instead of lag tables
b. COLA inflation protection
c. Benefit Integration with other sources
d. Partial & Residual Benefits - if able to work part-time, pay some DI benefit
e. Survivor Benefits - on death, pay designated beneficiary
f. Benefit Limitation
LTC
a. Case Reserves
- For small blocks
- Evaluate each claim to set reserve instead of using morbidity assumptions
b. Institutional or home-based care
c. Non-level daily benefits - different levels of care on different days of the week
Claim Reserve Methods - Lag or Development Methods
Completion Factor, Claim Triangle
1. Description: Estimate progession of payments using an assumed runoff (probability of claim termination based on past experience) adjusted to reflect current environment and admin practices
2. Assumes future payment patterns will be similar to recent past
3. Benefits: Transparent logic, requires limited data & computational tools
4. Characteristics of coverages
- Short-term benefits having processed claims (not capitation)
- May also be used for long-term claims
- Sufficient volume to obtain stable results - blocks can be combined to achieve credibility if patterns similar; volume needed depends on benefits covered, duration since plan inception, growth rate of business
- Ability to record incurred date and paid date for all claims
- Consistent lag patterns from incurred date to paid date; Requires claims data by service date and payment date
5. Necessary data
- Short incurral periods relative to runout periods
- Earned Premium or number of contracts for adjustments
- Claim data grouped into time periods; (1) Periods claims were paid & Periods claims were incurred, (2) If want to exclude pending, substitute report period for paid period
6. Create a claims development pattern
- Claims by incurred period and paid period
- Cumulative claims by incurred period and paid period
- Age-to-ultimate development factors
- Age-to-age development factors (chain link, chain ladder); (1) Uses more recent service months because you don’t wait for ultimate development, (2) Less influenced by cost trend, (3) Smoothing - Simple average, removing bumps, Weighted Averaging, Harmonic & Geometric, Dollar Weighted vs. Ratio Weighted, Per Member Age-to-Ratios
- Developing fully incurred claims
7. Adjustments and Projection Enhancements
- Most recent completion factors can produce highly variable outcomes. Dividing relatively small paid dollar amounts by a factor close to zero can magnify minor differences into large swings in projected fully incurred claims
- Projection Method - Replace completion estimate by projection of exposure X (per member per month costs PMPM) for most recent periods
- Loss Ratio Method - Replace completion estimate by expected or historical loss ratio X Premium
- Credibility Weights - blends either projection method or loss ratio method with completion method
- Trends - will affect PMPM or loss ratios over time
- Impact of seasonality
(1) Understand patterns of cost by month and decide if they form a reliable projection adjustment,
(2) Benefits plans with large deductibles reduce carrier liability during the fitst part of the year,
(3) Reduced incidence of elective hospitalizations around holiday periods,
(4) Teachers show increased use of elective surgies during the summer,
(5) Increased use of services annually during flu season,
(6) Workday seasonality - more opportunities to obain serves in a month with 5 Mondays - Risk adjustment measures and pharmacy claims can be used as leading indicators
- Setting Initial Lag Factors -
(1) May set last lag factor to be less than 1.00 if we believe residual claims may still develop,
(2) May set completion factors to be greater than 1.00 if the expected runout will be “recoveries” - Adjustment for claims processing changes
- Reinsurance Coverage, (1) For stop loss, consider impact of large claims on reserve estimates, (2) For coinsurance, keep track of contract percentages
- Inflation
- Changes in medical practices
- Compare to earned rates and exposure units for same time period for reasonableness
Claim Reserve Methods - Tabular Method
Reserve Methodology for Open Claims
- Common for known long-term claims (LTC, LTD), not used for IBNR or ICOS
- May be required by regulatory standards
- Start with listing of all reported claims, including benefit plan and claimant information
- Valuation systems may group similar characteristics for calculations
- Applies factor to individual claims or waived premiums
a. To calculate PV of remaining payments, may use
i. Life annuity values
ii. Continuance probabilities - estimate duration of payments given possibility of termination, lapse, mortality
iii. Commutation function tables
b. Factors may be based on
i. Age and gender of insured
ii. Elimination Period
iii. Cause of claim
iv. Length of disablement on valuation date
v. Remaining benefit period - Formula
a. Monthly benefit includes
- Optional benefits
- Benefit integration, government offsets
- Cost of living adjustments
b. Continuance is probability will not recover or die
- Middle of month assumption
- Tables vary by sex, EP, age at disability, claim duration
- Standard LTD tables
c. Interest discount is time value of money - Pending Claims
a. Calculation requires pending factor
- Probability remain disabled and will be approved for payment
- Includes interest discounting
b. If still in elimination period, pending factor * tabular reserve at end of EP
c. After EP, pending factor * (tabular reserve at end of claim duration + accumulated past claims that have not been paid)
d. Sometimes pending claims combined with IBNR development
Claim Reserve Methods - Estimation Methods
1. Factor Method - Similar to percentage of Premium Method for LTD
a. Disadv - assumes claim liability stable over time
b. Adv - simple and requires little data
2. Case Reserves (direct enumeration), Examiners’ Method
a. Estimate ultimate value for each reported claim less amount already paid
b. Open claims only, not IBNR
c. For small volume of claims, large claims, claims in litigation
3. Average Size Claim Method
a. Open claims only, not IBNR
b. Multiply number of reported claims by average size of previously closed claims
c. Not good for new blocks without credible average claim estimate
d. Not good if high variance in amounts paid
4. Projection Methods, or PMPM Method
a. Multiply projected cost per unit by exposed units
b. Applications - claim incidence low, data volume low, immature data, reasonableness check for other methods
c. Relies on accurate cost per unit assumption
5. Multiply projected cost per service by service counts
6. Multiply earned premium by estimated loss ratio or Loss Ratio Method
a. Short Term
b. Long-Term
- IBNR factor = target loss ratio - actual loss ratio
- IBNR = IBNR factor * earned premium
c. Too low of loss ratio assumption results in circular low projection of loss ratio
d. Applications same as projection method PLUS exposure data unreliable, and historical claim costs not reliable
7. Multiple triangles
a. Can split IBNR and ICOS calculation
b. IBNR triangle uses incurred period & report period. ICOS uses report period & paid period
8. Stochastic Approaches
a. Benefits - guidance for PAD, quantifies variance of earnings and reserves, volatility by LOB
b. Drawbacks - not necessarily more accurate, instills false sense of security, complexity, not every model can fit each LOB
c. Techniques
i. Parametric distribution fitting
1. Fit distribution to data or specify distribution based on judgement
2. Positive: Good if process stationary over time
3. Negative: If mean or variance not constant, use another method
ii. Ordinary Least Squares Regressions
1. Multivariate regression to investigate effects of trends & seasonality
2. Negative: Limited by assumption of normality and constant variance
iii. Generalized Linear Models
1. Positive: Allows for non-normality by allowing dependent variable to be bounded
iv. Stochastic time series models
1. Can use for PMPM estimates, in use in place of lag factors by extrapolating future claim payments
2. Positive: when values are correlated over time, and if cyclical pattern can use ARIMA model
v. Monte Carlo Sampling
1. Good for combining techniques above using sampling techniques
d. Considerations
i. Convariance of estimates
ii. May want to model components of reserve first, then combine
iii. Data: Availability of data, extent of validation that can be performed, appropriateness of data
iv. Access to statistical software
v. Specifying an appropriate model
9. Claim Settlement Expenses
a. Claim reserves should provide for expense of settling claims
b. LAE ratio = historical claim adjudication expense relative to claim payments
c. LAE reserve = LAE ratio * claim reserve
10. Emerging Issues and Techniques
a. Electronic commerce reduces lags
b. More claims paid today in 1st & 2nd lags although still not credible
c. Historical lag studies still needed to form cost base and trends
d. Today large volatility due to operational changes, reimbursement changes, and environmental changes
Follow-up Studies: Evaluating Claim Reserve Adequacy
1. Earnings trends masked by “catch up” if prior estimates were low and “releases” if prior estiamtes were high
2. May recast income statements of prior periods with new reserve estimates and restate earnings
3. Run-out studies or runoff studies
a. Tests reasonableness of prior period estimates, provides insights into data problems, operations changes, and leel of conservatism
b. Required in some financial statements and actuarial reports
c. Test (reserve estimate original valuation date) against (claims. runour subsequent to valuation date) plus (residual reserve)
4. A/E claim termination rate studies
a. Look at actual claim terminations versus expected
b. A/E > 1 indicates reserves adequate
c. May be used in combination with run-out study
d. US stat reserving rules require A/E termination rate studies for LTD
e. Considerations
i. Perform study by duration
ii. Credibility
iii. Types of terminations included (include recovery & death, exclude end of benefit period terminations)
iv. Exposure characteristics (i.e. separate studies for maternity)
v. Voluntary claims settlements: do not count lump sum settlement as a claim termination
5. Experience Studies
a. Evaluate long term assumptions
b. Important for products with contract reserves
c. Ccommonly involves a gross premium valuation
d. OK if (PV of future claims) < (PV of future gross premiums) + current reserves
e. Can result in change in assumptions for newly issued policies, restatement of reserves, and impact recoverability asset on balance sheet
6. Reporting applications
a. Regulatory (similar to claim reserve components outline)
i. Solvency
ii. Conservative
iii. Required formats for reports & computation
iv. Be familiar with jurisdiction of filing
v. Document any adjustments related to regulations
b. GAAP (similar to claim reserve components outline)
i. Realistic earnings patterns
ii. Less conservative than regulatory
iii. Reserve entries spread acquisition costs
iv. LTC, LTD biggest impact. Short term health products less impacted by different of STAT and GAAP
c. Experience reporting for employers and at-risk providers
i. Less sophisticated incurred claims calculation
- Settlement allows 3-4 months of claim runout, so size of reserve estimate small
ii. Sensitive to large changes that impact anticipated costs or trends
d. Valuations for acquisitions
i. Reserves material to establish value of business acquired
ii. May wait several months for claims run-out before final settlement of purchase price
iii. Active live reserve details essential and less easily reviewed
