Financial Condition Reporting for South African Short Term Insurers

1
Financial Condition Reporting for South African
Short Term Insurers
Emile Stipp Sam Isaacson 24 November 2005
2
Contents

• Background and scope
• Overall framework
• Insurance capital charge
• Investment capital charge
• Putting it all together
• Reserves
• Results
• Recommendations
• The way forward

3
Background and scope
4
Where it all started

• Started at the end of 2001
• Current one size fits all approach does not
  make sense move to a risk-based approach
• Paper presented at ASSA convention in 2002
• Working group discussed initial draft guidelines
  and made various changes
• Tender for recalibration sent out in November
  2004, awarded to Deloitte AIS Insight ABC in
  March 2005

5
Terms of reference, reliances and limitations

• Deloitte and Insight ABC appointed by FSB to
  calibrate formulae for Financial Condition
  Reporting for Short Term Insurance industry in SA
• Hence
• Construct formula
• In accordance with industry data in STAR returns
• And principles of Dynamic Financial Analysis
• And recommend new capital requirement for Short
  Term Industry the Industry Calibration
• The main limitations
• Data not contained in STAR returns, particularly
  in respect of non-proportional reinsurance
• Inaccurate data
• Held an industry workshop introducing concepts on
  11 August
• Sent two sets of individual results to each
  company in South Africa
• Had meetings with individual companies,
  reinsurers, cell captives and Actuarial Society
  STIC for feedback

6
Terms of reference, reliances and limitations

• This is an industry calibration there will
  always be overs and unders
• The question is whether it is more accurate than
  25 of net written premium currently required
• The model does not guarantee solvency under all
  circumstances
• And individual results for individual companies
  may be more or less appropriate depending on
  unique factors
• Some companies with specific needs
• Reinsurers
• Cell captives
• Those operating on behalf of Government, with
  effective Government guarantees
• Selected companies in niche markets
• Models could not be calibrated for each of these
  separately due to
• Lack of homogeneity
• Lack of data

7
Overall framework
8
The three models
9
The three models
10
Compared against current model

• The above framework is preferable, as the current
  model does not take into account
• Risks faced by company classes of business
  written, size of insurer, combination of classes,
  details of reinsurance programme, expenses etc
• And requires level of capital which is prudent
  for some and not prudent for others
• Advantage of current model it is simple
• We hope that industry calibration will be simple
  to apply in practice as it can be built into STAR
  returns
• And hence not require companies to apply detailed
  formulae
• And they can test different levels of capital
  required under different circumstances
• Industry calibration is balancing act between
• Greater complexity due to desire to allow for
  individual circumstances of companies VS
• Desire for simplicity

11
Schematic representation of framework
12
Comparisons

• The above approach is consistent with
  international trend towards risk-based capital.
  Adopted, among others, by
• Australia, USA, UK, Germany, Switzerland, Canada,
  Holland
• Also in line with principles established by
  International Actuarial Association
• And in line with investment capital charge
  adopted for long term insurers

13
Components of the framework

• Capital charge
• Insurance capital charge
• Investment capital charge
• Reserving
• Best estimates
• Prescribed margins

14
Data

• Intensive data cleansing process
• Used data from 1075 STAR returns

15
Insurance capital charge
16
Insurance Capital Charge

• Two primary investigations of Star return data
• Underwriting risk (ULR)
• Reserving risk
• Using net ULR and net reserving risk measures
• But calibrated to gross written premium (GWP) and
  gross unearned premium (GUPR)
• GWP is estimated for year following date of
  solvency calculation
• And GUPR can be regarded as looking back
• building blocks approach
• Determined initially for each of the 8 classes of
  business on a stand-alone basis

17
Insurance Capital Charge
18
Insurance Capital Charge
19
Insurance Capital Charge

• We ran model for notional companies and used
  interpolation to determine results for particular
  company

20
ULRs - means

• For all classes of business, relationship between
  mean ULR and GEP

21
ULRs standard deviation

• For all classes of business, relationship between
  std dev ULR and GEP

22
Underwriting cycle

• Allowed for implicitly. Evidence of cycle

23
Reserving risk

• Expresses development of claims relative to
  reserves

24
Reserving risk

• Expresses development of claims relative to
  reserves

25
Required number of simulations in DFA model

• Determined for each level of sufficiency in the
  end ran 79.2m simulations

26
Stand-alone capital

• Gross stand-alone capital surface at 99.5
  sufficiency

27
Diversification and correlation

• Allowed for explicitly
• And on the basis of data
• Writing more than one line of business reduces
  capital requirements (diversification effect)
• Correlations between lines of business
  generally increases capital requirements
  (correlation effect)
• This should discourage companies from dumping
  everything into the Miscellaneous class
• Finally, there is allowance for investment return
  on assets backing liabilities to reduce insurance
  capital charge

28
Investment capital charge
29
Overview

• To allow for fact that assets backing liabilities
  and capital may decrease in value to such an
  extent that solvency is threatened
• Hence, to the extent that market asset values are
  subject to variation, additional assets need to
  be held
• We performed this analysis using internationally
  the recognised Smith Model (TSM) stochastic
  asset model calibrated for South African market
• For different types of asset, determined
  additional assets to be held, at different levels
  of sufficiency
• First step is allocation exercise least risky
  assets allocated first (cash)
• If particular asset runs out, allocate from
  more risky assets
• Order of riskiness Cash, bonds, property,
  equity, other assets

30
Equity example

• Equity factors at different levels of
  sufficiency

31
Other risks to allow for?

• Operational risk
• Allowed for implicitly
• If operational risk can be measured, company
  should do something about it rather than setting
  aside additional capital
• Unfair to penalise whole industry for operational
  inefficiencies in some companies by requiring
  everyone to set aside additional capital for
  operational risk
• Also, did not have sufficient data to make
  explicit allowance for operational risk

32
Putting it all together
33
Framework

• Insurance and Investment Capital Charges are
  combined recognising that there is some
  correlation between insurance and investment
  risks, but not 100
• And capital should be grossed up to allow for
  type of investments
• Allowed for in formula
• But only 50 of g2 allowed for
• This is similar to the intended approach in
  Germany, as it deals with fact that insurance
  catastrophe may well affect investment market,
  but investment market crash may have no relation
  to insurance risks

34
Reserving
35
Liability estimation IBNR and OCR

• IBNR
• Use best estimate
• Table determined for companies who do not do own
  IBNR calcs
• Percentage of claims run off after each year
• OCR
• Companies case estimates

36
Liability estimation Prescribed Margin

• Added to IBNR and OCR best estimates
• To take up to 75th percentile
• Based on formula
• Where

37
Liability estimation UPR

• Aim is also to have best estimate
• But if UPR calculated using 365ths method,
  already includes margin of prudence given that
  premium includes profit margin
• We have assumed this margin takes us up to 75th
  percentile
• But still need to estimate the implicit margin
  for giving credit towards total capital
  requirements

38
Recap we have all elements of framework
39
Results of calibration
40
Overall results reserves
41
Overall results reserves

• Reserves may be understated for niche insurers
  and reinsurers due to longer run-off patterns
  inadequate data to calibrate separately for them

42
Overall results MCR
43
Overall results Total Capital Required (MCR PM)
44
Comparison different levels of sufficiency
45
Comparison with adjusted shareholders assets

• Adjustment takes into account prescribed
  reserving method and a release of the 10
  contingency reserve
• Only show companies with shortfall several have
  adequate capital

46
Applicability of industry calibration

• We believe it should generally be applicable to
  typical insurers
• But with cell captives, does not adequately
  reflect
• Non-proportional reinsurance (and expenses
  sometimes included in reinsurance)
• Lower risks due to structuring of business e.g.
  recapitalisation built into contracts
• Very limited data for cell captives
• Industry calibration more appropriate to third
  party rather than first party cells
• Recommendation that cell captives submit capital
  requirements on the basis of a certified model
• Extensive consultation with cell captive market
  still required

47
Applicability of industry calibration

• Industry calibration should be applied with care
  for reinsurers
• Reinsurance business may be more risky / volatile
  than traditional
• Longer tail business due to reporting delays
• Format of data in STAR returns not appropriate
• Recommendation for reinsurers
• At least reserving calculation on certified model
  basis, given longer run-off
• Modify STAR returns
• Impractical to calibrate specifically for them,
  given small number in SA
• Niche insurers may also not be appropriate
  depending on nature of business
• E.g. some have different run-off patterns (up to
  15 years)
• Some have specific arrangements e.g. Government
  guarantee
• For companies in run-off, model may give
  artificially low result for capital certified
  model should apply to capital calculation

48
What level of sufficiency to adopt?

• We believe 99.5 level too prudent at the outset
  rather build up to it over time
• Allows companies to build up capital, also using
  investment returns on increasing assets, and
  gives time to collect more data and test 99.5
  level before implementing it finally
• Following tables show years of net profit vs
  capital required at industry level
• Imperfect measure, but perhaps gives some
  indication of when industry should move from 98
  to 99 to 99.5
• Also, bear in mind with certified models that
  companies with most significant shortfall will
  probably apply to reduce capital requirement

49
What level of sufficiency to adopt?

• At 98 level, number of years net profit
  required to build up to capital (e.g. for 55 of
  market by number of companies and 66 of market
  by net premium 0 years)

50
What level of sufficiency to adopt?

• After 4 years, 87 of market (by net premium) is
  at 99 level (if using only net profit)

51
What level of sufficiency to adopt?

• After 6 years, 90 of market (by net premium) is
  at 99.5 level (if using only net profit)

52
Recommendations
53
Recommendations

• We recommend that
• Industry calibration should be done in accordance
  with the methodology and assumptions outlined
  above
• STAR returns to be expanded to include more data
  on non-proportional reinsurance, reinsurers, cell
  captives and co-insurance
• Reinsurance data may be used to refine
  calibration of model in future
• No explicit allowance should be made for the
  underwriting cycle
• No explicit allowance should be set aside for
  operational risk
• Specific attention to be given to developing
  framework for certified model for cell captives
  (esp first party cells), reinsurers, some niche
  insurers, companies in run-off
• Further consultation required with these sectors
• Despite this, industry calibration should be used
  as benchmark for all sectors
• FSB should also establish framework for internal
  models
• Grossing-up factor on insurance capital charge to
  be limited to 50, to avoid requirement being too
  onerous

54
Recommendations

• We recommend that
• Capital requirement implemented at 98 for 4
  years, then 99 for two years thereafter, then
  99.5
• Collected additional data to ensure that
  refinements can be made by the time 99.5 level
  reached
• Determine intervention points we recommend a
  fairly low intervention / action point, say 1.1.
  times MCR, with more extensive interventions when
  falling below MCR
• Every company falling short of MCR should agree
  with the FSB
• Whether they will use certified model or internal
  model
• Whether they will apply for special dispensation
  even if using certified model or internal model
• How the company will reach capital required as
  agreed with FSB
• To make application of model easier for
  companies, use spreadsheet contained in STAR
  returns
• Our recommendations are first step FSB now has
  to decide how to take it forward

55
The way forward
56
The way forward

• Recalibration report will be published
• FSB working group will discuss its findings and
  other outstanding issues (e.g. internal models,
  reporting etc)
• Working group will discuss draft wording for
  changes to legislation to implement FCR
• The usual process to change legislation will be
  followed
• Implementation - not before 2007