PwC *connectedthinking David Richardson Director Asia Actuarial Services PricewaterhouseCoopers 8 th September 2008 Enhancing Insurance Regulation and.

PwC*connectedthinking

David Richardson

Director

Asia Actuarial Services

PricewaterhouseCoopers

8th September 2008

Enhancing Insurance Regulation and Supervision



HISTORICAL BACKGROUND



• Insurance legislation designed to try to ensure solvency through implicit margins and conservative assumptions

• Assets at lesser of book or market value

• In Singapore/Malaysia, property taken into account at a value 30 years ago

• This was thought safe – the safer the better

• Taxation authorities didn’t help since a property revaluation would lead to a tax bill

• For life companies, this was coupled with a statutory net premium valuation of liabilities

• Net premium valuation involves calculating premiums based on prescribed mortality, morbidity and discount basis and then subtracting the present value of future net premiums from the present value of the sum assured and attaching bonus, if any, on the prescribed basis



• No explicit account is taken of management expenses, surrenders, lapses or distribution expenses. No explicit account is taken of expected future bonuses paid to policyholders or expected future dividends to be paid to shareholders

• The prescribed discount rates were what were thought to be low interest rates at the time so as to be safe and implicitly provide for future reversionary and terminal bonuses.

• The prescribed mortality rates were deliberately historical so as to implicitly provide margins

• The prescribed net premium valuation basis for Thailand requires the use of a discount rate equal to the interest rate used in the pricing of the life insurance products subject to a maximum of 6% p.a. The prescribed mortality table is TMO 86 for policies issued before 2002 and TMO 97 for policies issued from and after 2002.



• Current mortality experience among a number of life insurers is about 50% of TMO 97

• A Zillmer adjustment of 5.5% is permitted for ordinary life business and 6.5% for industrial life business

• The defects are huge lack of transparency and an inability of the insuring public to compare the financial health of one company with another

• Capital was locked away inefficiently and unrealised investment gains were passed to neither policyholder nor shareholder

• Without unrealised investment gains in the accounts, insurance was more expensive than it should have been and insurance company products were looking increasingly uncompetitive relative to banking and unit trust products



• The discount rate for the net premium valuation was fixed in times of high interest rates and could not cope with an economic cycle of low interest rates and what was thought to be conservative was nothing of the kind

• The same interest rate was prescribed for valuing participating policies as well as non-participating products – illogical

• “One size fits all” net premium valuation ignored any differences in rates of future bonuses or in the difference in management expenses, lapses, investment returns etc between different companies

• Even a mortality margin for life insurance valuations didn’t help the proper emergence of reversionary bonus because the margin which is based on the difference between sum assured and reserve decreased with policy duration whereas reversionary bonus values increase with duration

• Certain risks such as asset/liability mismatching risk and borrower default risk were not recognised at all and no reserves established to reflect these risks

hHISTORICAL BACKGROUND


• Governments were becoming increasingly concerned – the demise of Equitable Life in the UK and HIH in Australia

• Basel 2 provided an impetus for banks to decide on risk charges for borrower default based on individual risk standing rather than apply a flat 8% across the board

• Questions were posed:-

–Is there sufficient capital to support policy guarantees?

–Sufficient capital to support companies with high new business growth compared to less dynamic companies?

–Sufficient capital to cover mismatching of assets and liabilities?

–Sufficient capital to back the risk of corporate bond default or risk of market declines for equities and property?

• Change was inevitable. Already happened in Singapore, India, Australia and Malaysia.



MALAYSIA/SINGAPORE DIFFERENCES


*connectedthinking PwC


• In Singapore and Malaysia, RBC for general insurers is similar

• Both Singapore and Malaysia adopt projected cash flow on the life side to determine BE and additional PAD ( PRAD in Malaysia). The PAD is to cover liability fluctuations up to 75% confidence level

• With the Par fund in Singapore, there is introduced ‘Surplus Account”. The allocation to shareholders is by a 90/10 rule by which the shareholders are entitled to 10/90 x cost of policyholders’ bonuses

• Shareholders may withdraw the Surplus Account only if capital requirements are met. Remaining assets in the Par fund are available to meet policy liabilities.

• The policy liability of a Par fund is set to equal the policy assets i.e. assets less surplus account subject to (a) minimum condition liabilities i.e. the guaranteed liabilities discounted using a risk free discount rate (b) guaranteed liabilities + non-guaranteed liabilities discounted using the best estimate of the investment return of the fund ( Singapore) or the yield on A2 rated corporate bond ( Malaysia)



• In Malaysia, there is no “Surplus Account” concept for the par fund

• The liabilities of the par fund in Malaysia are the sum of the present value of guaranteed benefits (risk free discount rate) and the present value of non- guaranteed benefits i.e. bonuses. There is no adjustment for policy assets



• The discount rate for the GPV is different. Singapore uses the actuary’s best estimate of fund investment return. Malaysia uses AA rated bond yield

• Minimum CAR =120% ( Singapore) 130% ( Malaysia)

• Neither have liquidity tests e.g. assume projected future cash flow income sufficient to pay out benefits in all circumstances

• With regard to general insurance, Malaysia says that if discounting of liabilities is used, explicit claims escalation assumptions should be used. Singapore just leaves it to the Actuary to decide

• In the calculation of C1, the adjustment to mortality rates for insurance policies with guaranteed premiums and for annuities refers to a specified mortality table based on insurance company experience 1997-2002. All other mortality, expense, dread disease etc adjustments are based on the insurers best estimate ( usually PAD is 50% of the adjustment. On the other hand, in Malaysia, the adjustments refer to best estimate throughout but at a higher level e.g. non-guaranteed premium 120% of best estimate rates whereas Singapore uses 112.5%.



• For general insurance liabilities in Malaysia, the maximum diversification effect i.e. fund PRAD is not less than 50% of the sum of the individual PRAD by line of business. No such limitation is imposed in Singapore.

• Risk free discount rate in Malaysia is based completely on the yields for Government securities by duration up to 10 years. Thereafter based on the 10 year term.

• In Singapore, risk free discount rate is based completely on the yields for Government securities by duration up to 10 years. Thereafter, a stable long term risk free discount rate is determined based on the following: (i) compute the average closing yield of 10 year SGS since inception (ii) compute the average yield differential between 10 year and 15 year SGS (iii) derive an estimated long term yield by adding (i) and (ii) (iv) compute average closing yield of 15 year SGS over past 6 month period (v) allocate 90% weight to the yield in (iii) and 10% weight to the yield in (iv) and round up to the nearest 25 basis points to arrive at the LTRFDR. Use LTRFDR for durations of 15 years or more and interpolation between 10 and 15 years



• If the policy assets are short of either of the 2 floors, assets must come from surplus account to support the policy assets. This deduction may be recoverable in future when policy assets exceed the 2 floors

• On bonus distribution, Singapore RBC says that the cost of bonus will be calculated using the MCL basis.

• There are 2 capital requirements (a) fund solvency applicable to each insurance fund and (b) capital adequacy requirement applicable to the insurer overall

• There are 3 components to determine capital adequacy requirements.

• C1 is the liability risk charge obtained by applying specific risk charges to premium and claim liabilities ( general insurers) and by applying specific risk margins to the various assumptions affecting policy liabilities ( life insurers)



• C2 relates to asset risks based on exposure to bonds, equities etc but it also relates to asset/liability mismatching risk

• C3 refers to concentration risk in certain types of assets, counterparties or groups of counterparties

• Total risk requirement = C1+C2+C3 = TRR• Amount of capital to meet TRR is financial resources• For a fund, the financial resources > TRR• For all insurance funds ( except par funds), financial resources = assets-liabilities. For a

par fund, financial resources in Singapore = balance of surplus account + liability for non-guaranteed benefits

• For Malaysia, each insurer is required to determine CAR in its insurance and shareholders’ funds to support total capital required

• CAR is total capital available/ total capital required• For life insurer with par business in a separate fund, CAR = min (CAR all, CAR all except

par) reflects the ability of life insurers to adjust the level of future bonuses and also preserves the principal that surplus of par fund cannot support non-par business



• Total capital available – Tier 1 + Tier 2 capital in Malaysia Tier 1 is permanent, no maturity date, cannot be redeemed, non-cumulative, issued and fully paid up.

• In Singapore, total capital available = Tier 1 +Tier 2 + provision for non-guaranteed benefits( in par fund)

• In both Singapore and Malaysia, asset inadmissibility rules have been removed ( except for items such as goodwill, future tax credits etc)

• In Singapore, operational risk has no risk charges since it was felt that operational risks can be better dealt with by an insurer’s internal risk management systems and supervisory effort

• In Malaysia, operational risk has a specific risk charge of 1% of total assets

• In Singapore and Malaysia, there is a diversification effect by fund for general insurance liabilities but no diversification effect for life insurance liabilities or for assets. Solvency II specifies diversification effect for all.


Characteristics of a good RBC system and EWS



• Identifies risk faced by insurers and provides a clear framework on how they should be measured

• Asset risks should include credit risk, market risk and concentration risk

• Life liability risk should include mortality, management and distribution channels expenses, persistency, discount rates

• Non-life liability should include claims fluctuations through development years and the sufficiency of premium to cover unexpired risk

• Diversification risk is taken into account

• Initially, the framework should not be too complex. The system can be strengthened and enhanced over time.

Characteristics of RBC system


• Able to alert the regulators in good time of insurers who may be having financial problems

• Should be responsive without being an undue burden on insurers

• The key early warning indicator to the regulator is the Capital Adequacy Ratio

• However, because certain assumptions are based on industry data, additional indicators may be needed

Characteristics of EWS system


Project Plan



Overall timeline

Time in months*

N°

Activity (Details of each activity has been set

out in section 4) Sep 08

Sep- Oct 08

Oct – Nov 08

Nov – Dec 08

Dec 08 – Jan

09

Jan- Feb 09

Feb-Mar 09

Mar-Apr 09

Apr-May 09

May-Jun 09

Jun 09

1 Mobilisation and Stock-take 2 RBC formula 3 First Consultation with industry 4 RBC and EWS design 5 Second Consultation

Des

ign

6 RBC approval

Tes

t

7 RBC market testing

8 RBC roll out planning

9 EWS market testing

Ro

ll-o

ut

10 Post implementation review


Project phases

Phases Deliverables Main Activities

1. Mobilisation and Stock –take • New project structure• Agreed understanding of “where

we are”• Agreed Project Plan

• Formation of enlarged team• Approval of project plan• Workshops to identify outstanding technical issues• Preparation of “where we are now “report

2. RBC formula • RBC training number one, on international practice

• Agreed first draft RBC formula and asset and liability valuation rules

• Review Thai regulations and compare with other RBC regimes, including Malaysia and Singapore

• RBC training day on international practice• Draft asset liability proposals• Consider tax, accounting (including IFRS) implications• Approve asset liability proposals• RBC formula benchmark review• Draft RBC formula• Life/non-life RBC formula consistency review• Draft capital and fund solvency requirements • RBC Formula approval

3. First Consultation • Initial industry seminar • Stakeholders views on first draft

proposals

• Identify all relevant stakeholders for consultation• Agree consultation format and seminar agenda• Issue consultation invitations and launch seminar invitations• Prepare first consultation paper• Approve consultation paper• Consultation launch seminar• Review first consultation responses


Project phases


4. RBC and EWS design • EWS preferred approach• RBC proposed formula and

parameters• Transition proposals• Final models• RBC Training number two:

Regulator and industry training

• Collect data for parameterisation of RBC formula, including both assets and liabilities, taking into account possible future developments in asset classes

• Collate international RBC parameters• Life statistical analysis• Review existing non-life statistical analysis and develop as necessary• Consider international approaches to EWS • Define preferred EWS approach• First selection of RBC parameters• Compare international approaches to RBC transition and the issues

which arose• Define EWS information requirements and work flows• Collect data for market impact testing of RBC and EWS• Develop market experience and testing models• Conduct market impact testing of RBC and effectiveness of EWS• Complete EWS design• Complete RBC design

5. Second Consultation • Stakeholders views on final proposals

• Draft 2nd consultation paper to include RBC and EWS• Finalise draft of second consultation paper• Agree format of second consultation• Plan second consultation logistics• Initiate second consultation• Review second consultation responses


Project phases


6. RBC approval • Approved RBC framework and transition rules

• Draft regulations and actuarial guidelines

• Second draft RBC parameters• Revise and finalise RBC framework proposal• Revise and finalise EWS design• Prepare EWS manuals and templates• Ensure necessary approvals

7 RBC market testing • RBC training number 3• Identified issues for live RBC

environment

• Prepare final RBC regulations and supporting report• Prepare RBC roll out plan• Establish dry run help desk • Conduct training • Review dry runs results

8. RBC roll out • Training and help desk• Functioning RBC system

• Finalise RBC roll out plan• Amend transitional arrangements in light of all market testing• Establish help desk

9. EWS market testing • Training and help desk• Completed EWS framework• Industry report

• Test EWS workflow• Carry out “all market” dry runs• Analyse EWS dry runs • Develop internal report for discussion• Prepare amendments to EWS templates, workflows and manuals• Finalize industry report

10. Post implementation review To be performed by third party


Project Office



As-Is project office


To-be project office

* To be determined if necessary


Barriers to success/Key success factors



• Market education

• Market buy-in to the proposed framework

• Transition arrangements

• Integration of RBC and EWS

Success factors


ASSET RISK CHARGES



• The asset value in a RBC regime will be fair value or, for listed equities, market value

• We can split assets into 2 types: The fist type includes bonds, mortgages and preference shares for which there is a fixed interest return and a probability of default according to how risky is the investment.

• The second type includes property and shares which are subject to fluctuations in rental income, dividends and market values but not to default.

• The fair value of assets corresponds to the best estimate of liabilities and the probability of default ( for the first type) or of market value fluctuation (for the second type) enables the actuary to determine the appropriate asset risk charges to achieve a certain confidence level

• The general question to be answered for each investment type is this:

ASSET RISK CHARGES


• “ Given the current market value and a risk-free investment rate i.e. the investment rate generated by Government bonds of appropriate duration, what is the discount to current market value for which we have a certain level of confidence that the market value in 12 months’ time will not be below such discounted market value”

• The methodology to answer this question for listed ordinary shares is similar to the Black-Scholes methodology in the pricing of stock options and derivatives.

• The assumption is that proportional changes in the share price in a short period of time are normally distributed or that the actual share price at any future time has a log-normal distribution

ASSET RISK CHARGES


• The mean of the log-normal of the share price at any time t in future is given by the formula:

log(n) S(t) = log(n) S + ( µ - σ2/2)t

where S is the current share price, µ is the risk-free investment return rate and σ the volatility of the share price

• Also the standard deviation of log(n) S(t) = σ√ t

• One of the features of a normally distributed variable is that there is a 95% probability that it has a value within 2 standard deviations of its mean

• This provides a method of determining a suitable discount to market value at 31 December in any year such that we have 95% confidence that the actual market value as at 31 December in the following year is greater than such discounted market value

• It is normal practice with Black-Scholes to use a suitable stock market index as a proxy to the volatility of a share price

ASSET RISK CHARGES


• To illustrate the process, how did we generate asset risk charges for insurance companies in India?

• Tracked the closing prices in the Bombay SENSEX index for each trading day over a period of 10 years

• Taking the ratio of the index closing price in 1 day to the previous day’s closing price, we got the inter-day investment return

• We took various time periods such as 90 days, 180 days and 360 days to determine the average volatility of the investment return – 20%. The volatility itself was quite volatile – so we conducted a number of sensitivity checks with volatility ranges from 10% to 30%. The greater the volatility the greater the discount

• The risk-free investment return was the rate of investment return on Indian Government securities of 364 days duration

ASSET RISK CHARGES


• The risk free rate was 9.5%. We conducted sensitivity tests using rates 7% - 13%• At 95% confidence interval, the results were:

ASSET RISK CHARGES

Volatility 10.0% 15.0% 20.0% 25.0% 30.0%

Interest Rate Discount to market value

7.0% 9.5% 17.1% 24.3% 31.1% 37.4%

8.0% 8.5% 16.3% 23.6% 30.4% 36.8%

9.5% 7.2% 15.0% 22.4% 29.4% 35.8%

12.0% 4.8% 12.9% 20.5% 27.6% 34.2%

13.0% 3.9% 12.0% 19.7% 26.8% 33.5%


• The discount to market value varies considerably by interest rate changes at 10% volatility but much less so at higher volatility

• The lower the interest rate for a given volatility the higher the discount to market value

• Our best estimate discount against market value was 25%

• For corporate bonds, historical default data from LIC

• Once a bond defaulted on a dividend payment, it defaulted for all payments afterwards

• Simple probability of default

• Most corporate bonds not rated

• Assumed that coupon rate was correlated with rating but cannot group all bonds with a certain coupon rate over the 10 year period

• It’s the difference between the coupon rate and the risk free rate which determines if a bond is speculative or not

ASSET RISK CHARGES


ASSET RISK CHARGES

• Next we split the minimum and maximum coupon rates in a year into 7 equal intervals corresponding to ratings AAA to B-

• Assuming a uniform distribution of default over the bond duration, we obtained annualised default probabilities

• Similar exercises were conducted for residential and commercial mortgages, real estate and listed and unlisted preference shares

ASSET RISK CHARGES


GENERAL INSURANCE LIABILITIES



• Concept in risk-based capital is liabilities determined on a best estimate basis

• 50% chance of being too high and 50% chance of too low

• Margin in Singapore, Malaysia and Australia is to increase the 50% chance of being too high to 75% chance i.e. liabilities measured at a 75% confidence level.

• On top, we have liability risk charges

• To be consistent with asset risk charges, these should increase the confidence level from 75% to 95%

• The actuarial determination of claim reserves and unexpired risk reserves is statistical. It examines historical claim payments to determine trends so that future claim payments can be estimated

• One actuarial method selects development factors based on an analysis of historical claims development factors



• The selected development factors are applied to cumulative claims data for each line of business for each accident or underwriting year for which data not yet developed

• This produces an estimated ultimate claims cost

• The claim reserve is the difference between the ultimate claims cost and cumulative paid claims

• This is best estimate. 75% confidence level figure – referred to as provision for adverse deviations (PAD) in Singapore- normally determined by simulation

• Simulation approach is as follows:

• Taking cumulative claim payments throughout development years from an accident/underwriting year for each business line, we determine the ratio of cumulative claims paid up to a development year divided by the cumulative claims paid up to the prior development year

• From the various ratios in a development year, the best estimate is selected



• The selected ratios are applied to the cumulative claims data to determine the “fitted” claims development triangle

• The “fitted” claims data determines the best estimate

• The “fitted” claims data is considered the mean of a normal distribution and actual claims payments are samples of claim payments taken from the underlying normal distribution

• If F is the fitted claims data and A the actual claims data, then

(A-F) x 1/√F is a unit normal distribution with mean zero and standard deviation 1

• The simulation – “bootstrapping” – consists in applying random numbers from 0 to 1 to the variable to more closely define the normal distribution

• Usual to run 50 simulations 10 times and take the average of each mean as the best estimate and the average of the PADs as the PAD

• The average of the PADs is divided by the average of each mean and expressed in percentage terms



• The percentage is applied to the best estimate to determine the appropriate PAD

• The simulation does not depend on the method to get best estimate but depends on a random selection of the differences between fitted claims and actual claims

• Illustration: Cumulative paid claims from accident year through development years

Converted Triangle Reshaped to Left CumAccident Yr 12 24 36 48 60 72 84 96 108

1998 1,835,329 3,587,773 3,932,318 3,971,322 4,009,753 4,101,513 4,117,946 4,129,419 4,131,1971999 2,061,541 4,031,489 4,127,649 4,343,880 4,644,558 4,737,900 4,745,713 4,764,7922000 2,765,751 4,321,197 4,557,032 4,846,339 5,197,292 5,238,904 5,244,6382001 1,038,264 2,559,271 3,192,100 3,670,469 3,843,598 3,868,4122002 1,168,296 5,258,246 6,665,277 7,586,690 7,787,0412003 1,150,092 4,296,560 5,799,263 5,985,4142004 794,433 1,849,857 2,386,0662005 857,856 1,682,8612006 1,603,512



• Ratio of cumulative paid claims for one development year to the cumulative paid claims for previous development year and determine various averages of the factors

Age-to-Age FactorsAccident Yr 12 - 24 24 - 36 36 - 48 48 - 60 60 - 72 72 - 84 84 - 96 96 - 108 To Ult

1998 1.955 1.096 1.010 1.010 1.023 1.004 1.003 1.0001999 1.956 1.024 1.052 1.069 1.020 1.002 1.0042000 1.562 1.055 1.063 1.072 1.008 1.0012001 2.465 1.247 1.150 1.047 1.0062002 4.501 1.268 1.138 1.0262003 3.736 1.350 1.0322004 2.329 1.2902005 1.9622006

Averages12 - 24 24 - 36 36 - 48 48 - 60 60 - 72 72 - 84 84 - 96 96 - 108 To Ult

Simple Avg All Yrs 2.558 1.190 1.074 1.045 1.014 1.002 1.003 1.000

Latest 7 2.644 1.190 1.074 1.045 1.014 1.002 1.003 1.000 Latest 5 2.998 1.242 1.087 1.045 1.014 1.002 1.003 1.000 Latest 3 2.675 1.302 1.107 1.049 1.012 1.002 1.003 1.000

Medial Avg All Yrs x1 2.400 1.191 1.072 1.048 1.014 1.002 1.003 1.000 Latest 7x1 2.489 1.191 1.072 1.048 1.014 1.002 1.003 1.000 Latest 5x1 2.843 1.268 1.085 1.048 1.014 1.002 1.003 1.000 Latest 3x1 2.329 1.290 1.138 1.047 1.008 1.002 1.003 1.000Volume Wtd

All Yrs 2.364 1.184 1.075 1.044 1.014 1.002 1.003 1.000 Latest 7 2.440 1.184 1.075 1.044 1.014 1.002 1.003 1.000 Latest 5 3.124 1.236 1.086 1.044 1.014 1.002 1.003 1.000 Latest 3 2.794 1.302 1.101 1.045 1.012 1.002 1.003 1.000

Time Wtd Avg All Yrs 2.703 1.242 1.085 1.046 1.011 1.002 1.004 1.000

Latest 7 2.725 1.242 1.085 1.046 1.011 1.002 1.004 1.000 Latest 5 2.892 1.265 1.089 1.046 1.011 1.002 1.004 1.000 Latest 3 2.591 1.304 1.099 1.045 1.010 1.002 1.004 1.000

Linear Proj Est All Yrs 3.119 1.396 1.120 1.048 0.999 0.999 1.003 1.000



Development Factor Selection12 - 24 24 - 36 36 - 48 48 - 60 60 - 72 72 - 84 84 - 96 96 - 108 To Ult

Prev. Selection 3.650 1.225 1.100 1.055 1.020 1.003 1.003 1.020 1.000Defaults 2.794 1.302 1.101 1.045 1.012 1.002 1.003 1.000 1.000Default Weight 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000User Selected 3.000 1.250 1.100 1.045 1.015 1.003 1.003 1.001 1.007Selected Result 3.000 1.250 1.100 1.045 1.015 1.003 1.003 1.001 1.007FacToUlt 4.437 1.479 1.183 1.076 1.029 1.014 1.011 1.008 1.007Percent of Ult 0.225 0.676 0.845 0.930 0.972 0.986 0.989 0.992 0.993

Estimated Ultimate

Accident YrPaid

AmountsFactor to Ultimate

Estimate of Ultimate

Earned Premium

Estimated Ratio

1998 4,131,197 1.007 4,160,115 5,718,513 72.75%1999 4,764,792 1.008 4,802,943 5,929,209 81.00%2000 5,244,638 1.011 5,302,492 6,517,091 81.36%2001 3,868,412 1.014 3,922,818 5,433,706 72.19%2002 7,787,041 1.029 8,015,007 8,975,181 89.30%2003 5,985,414 1.076 6,437,866 9,505,261 67.73%2004 2,386,066 1.183 2,823,078 5,248,383 53.79%2005 1,682,861 1.479 2,488,850 3,652,307 68.14%2006 1,603,512 4.437 7,114,492 6,802,708 104.58%Total 37,453,933 45,067,662 57,782,359

Estimated Ultimate Reconciliation

Accident YrPaid

AmountsIncurred Amounts

Case Reserves IBNR

Estimate of Total

ReserveEstimate of

Ultimate1998 4,131,197 4,160,737 29,540 -622 28,918 4,160,1151999 4,764,792 4,798,762 33,970 4,182 38,152 4,802,9432000 5,244,638 5,278,031 33,393 24,461 57,854 5,302,4922001 3,868,412 3,987,383 118,971 -64,565 54,406 3,922,8182002 7,787,041 8,545,184 758,143 -530,177 227,966 8,015,0072003 5,985,414 7,206,044 1,220,630 -768,178 452,452 6,437,8662004 2,386,066 4,346,845 1,960,779 -1,523,767 437,012 2,823,0782005 1,682,861 2,738,748 1,055,887 -249,898 805,989 2,488,8502006 1,603,512 6,537,106 4,933,594 577,386 5,510,980 7,114,492Total 37,453,933 47,598,840 10,144,907 -2,531,178 7,613,729 45,067,662



• Diversification effect: The best estimate and PAD are obtained separately for each line of business but the overall PAD is not the sum of the PADs by line of business

• Combining short-tailed and long-tailed business has the effect of reducing volatility

• Determined by combining claims data for all lines and running the simulation again

• The other RBC reserve for general insurance is the unexpired risk reserve

• Generally obtained by applying best estimate ultimate loss ratio to unexpired premium reserve. In addition PAD required on the URR

• Normal distribution assumption applied to assets to determine asset risk charges at 95% confidence limit

• The best estimate plus PAD takes us to 75% confidence limit.



• Liability risk charges – based on the same normal distribution assumption – will take us from 75% confidence limit to 95% confidence limit

• In Singapore, we determined liability risk charges by collecting claims data from a number of insurers and applied bootstrapping to the best estimate plus PAD after allowing for the diversification effect for both claim reserves and unexpired risk reserves

• This provides the liability risk charges. This is added to the asset risk charges (both credit risk and market risk)

• The asset concentration risk charges are simply 100% of an excessive investment – according to a pre-determined table – in any one asset

• Adding them up gives the total risk charges for a general insurer



LIFE INSURANCE LIABILITIES



• Determination of life insurance liabilities follows the same principles as applied to assets and general insurance liabilities

• The artificial net premium valuation of liabilities is replaced by an explicit prospective analysis of future expected cash flows arising from each policy

• The actuary should project cash flows forward over the expected contract term taking into account any options which could extend the contract term, such as term policies with renewable options

• The approach requires a projection of expected future payments and receipts taking explicitly into account actual premiums, investment income and investment expenses, mortality and morbidity benefits, surrender benefits, lapses, distribution costs, management expenses, claim expenses ( if not in management expenses), reinsurance premiums and recoveries, tax, options and guarantee costs



• It should be noted that this cash flow projection takes explicitly into account expected future bonuses for participating policies and expected shareholders’ future dividends

• These are on the actuary’s best estimate assumptions. An additional margin is required to allow for adverse deviation and this would be the actuary’s estimate of what the assumptions would be at 75% confidence interval.

• The liability risk charges are such as to increase the confidence limit from 75% to 95%.

• The liability risk charges are prescribed percentages to apply to best estimate mortality, morbidity, renewal expense, persistency etc

• To illustrate, if the actuary’s best estimate mortality assumption was 50% TMO 97 and his 75% confidence interval assumption were 58% TMO 97, the mortality component of the liability risk charges would be set at, say, 135% of the best estimate mortality assumption i.e. 67.5% TMO 97



• A logical distinction is made between liabilities which are guaranteed (non-participating business, basic sum assured, attaching bonus of participating business) and liabilities which are not guaranteed (future bonuses)

• The discount rate for guaranteed liabilities would be a risk-free interest rate i.e. the yield on Government securities of appropriate duration. The question of appropriate duration provides some difficulties since there are no Government bonds with durations as long as many of the liabilities

• Malaysia specifies that the risk-free discount rate should be equal to Government bond yields for durations up to 10 years and for cash flow after 10 years, the discount rate is the 10 years Government bond yield



• The justification is that with a normal yield curve, yields for durations greater than 10 years would be higher than the yield for 10 years duration

• Therefore conservative.

• The discount rate for non-guaranteed liabilities i.e. future bonuses and future shareholders’ dividends should be greater than the risk-free rate to reflect the non-guaranteed nature of these liabilities

• Singapore directs that the best estimate of the investment return of the fund should be used

• Malaysia prescribes it a little more by directing the risk-free rate plus a margin should be used. The margin is the difference between the yield of A rated bond of same duration and the risk-free rate



• Solvency II in Europe does it differently. The discount rate is proposed to be the discount rate which equates the discounted value of future expected asset cash flows to the current asset market value

• The fact that liabilities will generally be longer than the asset duration for life insurers introduces an asset/liability mismatching risk

• This applies to the value of the guaranteed liabilities and the value of fixed interest investments.



• Suppose V is the liability including PAD based on risk-free discount rate and A is the fair value of the assets

• Calculate V1 and V2 based on increasing and decreasing interest rates

• Both Singapore and Malaysia specify the increases and decreases in absolute amount terms by duration – such as 1.5% for 1 year to 0.8% for 20 years + but it might be an improvement to specify the increases and decreases in percentage terms by duration

• Calculate A1 and A2 based on increasing and decreasing interest rates

• Suppose A-V = S, A1-V1=S1 and A2-V2=S2, the asset/liability mismatching risk charge = the higher reduction in surplus under the increasing and decreasing interest rate assumptions



• Total risk charges for life insurers are the asset risk charges (credit and market risk) plus the asset/liability mismatching risk plus asset

concentration risk plus the insurance liability risk charges

• Liability risk charges are determined by V*-V where V* is the recalculated value of the liabilities based on the 95% confidence limit assumptions and V is the value of the liabilities at the 75% confidence level



CAPITAL ADEQUACY RATIO



• Key measure of the financial health of an insurer in a risk-based capital regime is the Capital Adequacy Ratio

• This is the ratio of capital remaining after taking into account the best estimate plus PAD liabilities divided by the total of the various risk charges

• The Capital Adequacy Ratio and the progress of it over the years is an early warning to the regulator for progressive action



• Both Singapore and Malaysia have set minimum Capital Adequacy Ratios

• Capital can be split into permanently available capital such as shareholders’ paid-up capital, retained surpluses and any property revaluation reserve - termed Tier 1 capital – and other capital of a less permanent nature

• The key feature of Tier 1 capital is that it is available to policyholders in the event of liquidation

• One point is that the investment of such capital has consequential asset risk charges and these should be added to the total risk charges before determining CAR



• The method to determine fair value of assets including derivatives is contained in IFRS 39. Therefore IFRS 39 should be introduced at the same time as Risk-based Capital

• Also IFRS 4 focuses on the differences between insurance and investment products and the standard of disclosure to assist the layman in assessing the financial health of the insurer

• This should also be introduced at the same time.

• Sufficient time should be allowed for testing the financial impact of Risk-based Capital to the industry in general

• It’s also important that the introduction of Risk-based Capital is not regarded simply as a compliance matter but as a powerful means of improving insurance companies financially to the benefit of shareholders and policyholders alike.


© 2008 PricewaterhouseCoopers. All rights reserved. “PricewaterhouseCoopers” refers to the network of member firms of PricewaterhouseCoopers International Limited, each of which is a separate and independent legal entity. *connectedthinking is a trademark of PricewaterhouseCoopers LLP (US).

PwC

Thank you

PwC *connectedthinking David Richardson Director Asia Actuarial Services PricewaterhouseCoopers 8 th September 2008 Enhancing Insurance Regulation and.

Documents

connectedthinking slide

prescribed discount

prescribed mortality

rates of future bonuses

supervision pwc

prescribed mortality

insurance company products

discount basis