Actuaries in a RunOff Environment

1
Actuaries in a Run-Off Environment

• AIRROC Education Session
• February 12, 2009
• Presented by
• Jason L. Russ, FCAS, MAAA
• Thomas A. Ryan, FCAS, MAAA

2
Actuaries in a Run-Off Environment

• Actuarial Roles Include
• Estimation of projected liabilities (How much
  will I have to pay?)
• Estimation of payout patterns (When will I have
  to pay it?)
• Assistance in realization of reinsurance assets
  (Can I get some of it back?)
• Modeling of financial results (Do I have enough
  in the bank to pay it?)
• Pricing of commutations (What assets can I
  monetize at a fair price?)

3
Goals of Session

• De-mystify some common actuarial terminology
• Provide an understanding of the challenges and
  limits of actuarial work in a run-off
  environment
• Prepare you to ask better questions when faced
  with an actuarial work product.

4
Specific Agenda

• Whats reasonable about a range of liabilities?
  Whats the best estimate?
• Dueling actuaries how do you reconcile their
  differences?
• What unique characteristics of run-off cause
  distortions in actuarial projections?
• How do actuarial methodologies differ when
  evaluating a portfolio vs. a specific contract?

5
I. Estimates and Ranges
6
Introduction

• When evaluating liabilities, a point estimate
  is usually a necessary evil, a single value
  needed for accounting purposes or to complete a
  transaction.
• Many actuaries prefer to discuss ranges and
  distributions of estimates instead of a point
  estimate. These items add value in communicating
  the uncertainty surrounding an estimate.
• Critical to understand terminology, especially if
  audience is a buyer/seller of liabilities.

7
Common Types of Ranges

• Range of Reasonable Estimates
• Range of Possible Outcomes
• Both are referred to as reserve ranges but they
  are very different.

8
Common Types of RangesRange of Possible Outcomes

• Includes all the possible results of the claim
  process.
• Usually generated by statistics or simulations
  but can also be based on scenario testing or
  historical observation.
• A distribution generally describes all possible
  outcomes.

9
Common Types of RangesRange of Reasonable
Estimates

• Typically narrower than a range of possible
  outcomes.
• Produced by appropriate actuarial methods or
  alternative sets of assumptions that an actuary
  considers to be reasonable.

10
Other Range Issues

• No objective boundaries exist for a reasonable
  range. Subjective based on judgment of actuary.
• A range of reasonable estimates vs the range
  of reasonable estimates.
• Sometimes difficult to aggregate ranges for
  individual contracts or lines of business into an
  overall aggregate range - must account for
  correlation among segments.

11
Point Estimates What is the actuary intending
to measure?

• High/Low estimate?
• Mean, median, mode?
• 50th percentile, 60th percentile, etc.?
• New Actuarial Standard of Practice (ASOP43)
  helps guide actuaries in providing their
  measurement objectives
• Actuarial Central Estimate an estimate that
  represents an expected value over the range of
  reasonably possible outcomes.
• Best or actuarial estimates are not
  sufficient identification of the intended
  measure. They describe the source or the quality
  of the estimate but not the objective.

12
Key questions to ask an actuary about ranges

• What type of range are we discussing reasonable
  estimates or possible outcomes?
• How was the range calculated?
• What is the likelihood of outcomes outside the
  range?

13
II. Reconciling Actuarial Indications
14
Reconciling Estimates from Two Actuaries

• Two actuaries estimate asbestos loss for a
  portfolio
• Actuary 1 uses ground-up exposure models and
  estimates unpaid loss of 100M
• Actuary 2 uses top-down benchmarks and estimates
  unpaid loss of 300M
• What is causing the difference?

15
Need More Information!

• What details are each side willing to provide?
• May have to give information to get information.
• Cant ask for everything! Focus on where poor
  judgment most likely to occur.

16
Where to Focus

• First, attempt to use the other sides
  methodology
• Easiest way to uncover potential pitfalls
• Otherwise, turn to common areas of disagreement
• Both to attack and defend

17
Ground Up Issues

• Usually wont get full details
• How are underlying estimates made?
• How does model perform on closed claims?
• Investigate largest claims
• Non-products claims
• Pure IBNR claims
• How good is data?
• Handling of reinsurance

18
How Are Underlying Estimates Made?

• Reflect ultimate number of plaintiff filings, or
  limited number of filing years?
• How recent is underlying data?
• How complete is underlying data?
• Expenses vs. indemnity

19
How Does Model Perform On Closed Claims?

• Good to test as defensive measure
• Consider how closed claims may differ from open

20
Investigate Largest Claims

• Both those with largest case reserves and those
  with largest model indications
• Enlist aid of claims people / legal
• Go beyond the numbers what are possible
  outcomes?

21
Non-Products Claims

• How is it reflected?
• What is potential?
• Any known exposures with large potential?
• Claims/legal opinions

22
Pure IBNR Claims

• How is it estimated?
• Implied frequency severity vs. recent activity
• Separately by defendant tier

23
How Good Is Data?

• Attachment points complete from ground up?
• Duplicate claims
• Limits
• Reconcile to financials

24
Handling of Reinsurance

• Modeling by claim or in aggregate?
• Probabilistic nature
• Interaction of fac and treaty

25
Top-Down Issues

• What ratios?
• What amounts?
• Stability
• How good is data?

26
What Ratios?

• Unpaid / average paid (survival ratio)
• Ultimate / total paid
• Ultimate / total incurred
• IBNR / case reserve
• IBNR / average incurred
• Check them all

27
What Amounts?

• Based on industry, or peer companies, or ??
• Is it an appropriate match?
• Estimates current?

28
Stability

• How do estimates change as data changes?
• Any distortions
• Recent settlements
• Large case reserve

29
How Good Is Data?

• Complete history?
• Reconcile to financials

30
III. Distortions Due to Run-Off
31
Introduction

• Most actuarial methods assume a stable
  environment past results are a reasonable
  indicator of the future.
• According to ASOP 43, the actuary should
  consider whether there have been significant
  changes in conditions that are likely to be
  insufficiently reflected in the experience data
  or in the assumptions used to estimate unpaid
  claims.
• In a run-off situation, many things can change
  significantly

32
Potential Changes in a Run-Off Book

• Timing of claim count reporting and settlement
• Average case reserve level
• Average amount of claim payments
• Timing and amount of reinsurance recoveries
• Underwriting standards/loss ratios
• Unallocated loss adjustment expense (ULAE).

33
Timing of Claims Reporting and Settlement

• For liquidations, stay on claims may be put in
  place, reducing reporting and closing of claims.
  Bar dates may also limit future claim
  reportings.
• Run on the Bank Effect Claimants seek to
  avoid reduced recoveries (or no recovery) by
  achieving faster settlements.
• Actuarial methods that use claim frequency may
  need to rely on historic frequency levels rather
  than data distorted by early stages of run-off.

34
Average Case Reserve Level

• Overwhelming number of new claims or cuts in
  claim staff may result in inattention to claim
  files resulting in inadequate case reserves.
• If new entity takes over management of run-off
  book, often there is a change in the claims
  management team (move to internal handling or new
  third party administrator). These changes often
  result in changes to reserve levels either
  higher or lower.
• As a result, actuary may want to rely more on
  paid loss-based methods rather than incurred loss
  based methods.

35
Average Amount of Claim Payment

• Average amounts paid per claim can decrease
• If claims settled faster, a discount for the time
  value of money may be applied
• Claimants may accept lower settlements from a
  run-off entity
• In a liquidation, limits of guaranty funds are
  often lower than policy limits
• Some actuarial methods rely on projections of
  claim counts and claim severity amounts. Paid
  claim severities may need to be adjusted to
  reflect impact of lower settlements.

36
Timing and Amount of Reinsurance Recoveries

• In runoff, the nature of an insurers
  relationship with its reinsurers may change
  drastically. There is no future relationship to
  jeopardize by behaving badly.
• Claim audits and intentional delays in payments
  may be used to induce commutations.
• Lower reinsurance recoveries may also result as
  lower claim payments may result in reduced
  piercing of excess of loss reinsurance
  retentions.
• Must factor in potential lower future reinsurance
  recoveries than those achieved in the past.

37
Underwriting Standards/Ultimate Loss Ratios

• Loss ratios for more recent years likely to be
  significantly higher than prior years as
  companies in trouble sometimes place less
  profitable business on books in order to generate
  increased cash flow.
• Also potential for drastic changes in retentions
  and limits.

38
Unallocated Loss Adjustment Expense (ULAE)

• Be clear on what is included in ULAE some
  consider all expenses as ULAE in a run-off
• Correct definition of ULAE may help in developing
  industry benchmark statistics.

39
Impact of Settlement StrategyActuaries needs to
be aware of strategy pursued

• Accelerated Exit larger claims settled faster
  to reduce uncertainty potentially at a premium.
• Leveraged Settlements Claims settled in bulk
  and packaged with claims for same cedant from
  other portfolios to provide overall discount on
  reserves however, individual settlements may be
  higher or lower.
• Aggressive Commutation recovery of commutation
  payments can distort data and triangles. If data
  related to commuted business is removed from
  actuarial data, it may result in reduced
  credibility of data for projections.

40
Impact of Run-Off Environment on Actuarial Methods

• Critical to consider items listed above as
  otherwise traditional actuarial techniques can
  produce inaccurate results (either higher or
  lower).

41
IV. Evaluation of a Portfolio vs. Specific
Contracts
42
Actuarial Projections for Individual Contracts

• Actuaries rely on law of large numbers.
• Generally estimates reasonable in aggregate, but
  not reliable for single contract.

43
Possible Methods

• Allocate IBNR using case reserves
• Allocate IBNR using total incurred loss
• Allocate IBNR using recent incurred activity
• Ground up models

44
Allocate IBNR Using Case Reserves

• Pros
• Case reserve volume could have correlation to
  potential future activity
• Cons
• Punished by case strengthening
• Large open claims can distort
• Sometimes case reserves are old, should be
  removed
• Contracts with no open claims get no IBNR but
  they have potential!
• Leveraged

45
Allocate IBNR Using Total Incurred Loss

• Pros
• Less leveraged than using case reserves
• Incurred loss could have correlation to potential
  future activity
• Cons
• Large settlements can distort
• May allocate IBNR where no potential exists,
  unless care is taken

46
Allocate IBNR Using Recent Incurred Activity

• Similar pros and cons as using case reserves or
  incurred losses
• Benefit of reflecting recent trends
• Trade off is could have high leverage distorted
  by unusual activity (or inactivity)

47
Ground-Up Models

• Model assumptions may be too low or high for
  individual contract review assumptions
• Handling of pure IBNR and non-products

48
Best Method?

• Use multiple methods
• Understand greater level of uncertainty
• Review using alternative methods (claims people,
  legal)

49
V. Questions?