Subjectst32005 2009

Faculty of Actuaries Institute of Actuaries
EXAMINATION
8 April 2005 (pm)
Subject ST3 General Insurance

Specialist Technical
Time allowed: Three hours
INSTRUCTIONS TO THE CANDIDATE
1. Enter all the candidate and examination details as requested on the front of your answer
booklet.
2. You have 15 minutes at the start of the examination in which to read the questions.
You are strongly encouraged to use this time for reading only, but notes may be made.
You then have three hours to complete the paper.
3. You must not start writing your answers in the booklet until instructed to do so by the
supervisor.
4. Mark allocations are shown in brackets.
5. Attempt all 8 questions, beginning your answer to each question on a separate sheet.
6. Candidates should show calculations where this is appropriate.
AT THE END OF THE EXAMINATION
Hand in BOTH your answer booklet, with any additional sheets firmly attached, and this
question paper.
In addition to this paper you should have available the 2002 edition of the
Formulae and Tables and your own electronic calculator.
Faculty of Actuaries
ST3 A2005 Institute of Actuaries
1 You are the actuary to a general insurance company which writes only employers
liability insurance.
List the data you require to carry out an assessment of the appropriateness of the risk
premiums. [8]
2 You are the general insurance actuary for a small insurance company which writes
only coach fleet and car fleet commercial motor insurances.
Describe the differing claims characteristics of these two types of fleets. [7]
3 Explain the importance of relevant, accurate data for a general insurance company
underwriting private motor insurance. [8]
4 You are the general insurance actuary to a company which writes only commercial
property insurance.
Describe the following characteristics of this class of business:
(i) purpose and benefits [2]

(ii) perils covered [1]
(iii) exposure measure [1]
(iv) claims [4]
(v) rating factors [2]
[Total 10]
5 You are the Chief Actuary working for a general insurance company that writes only
commercial property business. The Board has indicated that it wishes to take over a
similar company with whom your company is in direct competition, but has not yet
made any formal approach.
Outline the major actuarial investigations that you would undertake in order to assess
the viability of this proposal.
[10]
ST3 A2005 2
6 You are an actuary working for a small, general insurance company that specialises
only in long-tailed insurance risks in its local insurance market. The company has a
moderate level of free assets. The Board, who require a risk averse investment
strategy, has asked for a review of the investment guidelines given to the company s
investment managers. It is assumed that there are no regulatory investment
restrictions.
(i) Outline the instructions that would be given to the investment managers in
respect of the assets that they may or may not hold, including any relevant
limits. [6]
At a recent Board meeting, one of the members suggested that, given the long-tailed
nature of the risks insured and the current low historic level of the stock market, the
equity proportion of the investment portfolio should be increased to benefit from
future stock market increases.
(ii) Explain the advantages and disadvantages of his suggestion for the company.
[6]
Another Board member has suggested increasing the proportion invested in index-
linked Government bonds.
(iii) Explain the advantages and disadvantages of this suggestion compared with
that given to the previous Board member. [4]
[Total 16]
7 A global relief agency provides humanitarian assistance to victims of natural disasters

in poor countries around the world. It is looking into its sources of funding for large
natural disasters and is thinking of buying a catastrophe insurance policy. The policy
pays out a fixed sum of $100m if a large natural catastrophe leads to more than 500
deaths during the term of the policy. The premium for the policy is likely to exceed
$10m per annum.
(i) Discuss the advantages and disadvantages to the relief agency of purchasing
such a policy. [5]
(ii) Discuss the advantages and disadvantages to the Insurer of writing such a
policy. [4]
(iii) Two different insurance providers are competing to write this policy.
Company A thinks that the number of such events has a Poisson distribution
with Poisson parameter of 10%, while Company B estimates the Poisson
parameter to be 8.33%. Company A requires a return on capital of 10% while
Company B requires a return on capital of 15%. Both companies allocate
capital to this contract using the same methodology. The amount of capital is
set to ensure that the probability of ruin for this contract on a standalone basis
is less than 10%. Using approximations where appropriate, calculate which of
the two companies will have the higher premium. [9]
[Total 18]
ST3 A2005 3 PLEASE TURN OVER

8 You have just been appointed as the actuary to a small general insurance company
which writes two classes of business, A and B, throughout the territory. All business
is written through insurance brokers.
The only financial information you have been presented with is shown below. All
figures are as at end 2003.
Class A 2000 2001 2002 2003

Gross written premium 1,000 750 500 250
Net written premium 750 550 350 150
Net earned Premium 750 550 350 150
Claims (all after reinsurance recoveries)
Paid total to date 100 50 25 10
Current case estimates 400 250 175 90
IBNR/IBNER 150 200 150 100
Total claims cost 650 500 350 200
Expenses 75 65 55 45
Commission 150 113 75 38
Investment return 200 195 190 185
Profit 75 67 60 52
Class B 2000 2001 2002 2003

Gross written premium 1,000 1,250 1,500 1,750
Net written premium 950 1,188 1,425 1,662
Net earned Premium 950 1,069 1,306 1,543
Claims (all after reinsurance recoveries)
Paid total to date 650 620 650 560
Current case estimates 30 600 150 250
IBNR/IBNER 10 55 140 310
Total claims cost 690 1,275 940 1,120
Expenses 100 130 165 210
Commission 150 188 225 263
Investment return 30 38 47 55
Profit 40 (486) 23 5
(i) Comment on the financial characteristics (including financial trends) of each

of the classes of business the company writes, using any appropriate financial
ratios. [14]
(ii) State, with reasons, which classes of business A and B are likely to be. [4]
(iii) Describe the issues the company will face in the short term and any actions
you would recommend. [5]
[Total 23]
END OF PAPER
ST3 A2005 4
EXAMINATION
April 2005

EXAMINERS REPORT
Introduction
The attached subject report has been written by the Principal Examiner with
the aim of helping candidates. The questions and comments are based around
Core Reading as the interpretation of the syllabus to which the examiners are
working. They have however given credit for any alternative approach or
interpretation which they consider to be reasonable.
M Flaherty
Chairman of the Board of Examiners
28 June 2005
Institute of Actuaries
Subject ST3 (General Insurance Specialist Technical April 2005 Examiners Report
1 This question asked candidates to consider the risk premium. Those who spent time
talking about expenses, commission etc did not gain marks for such comments. On the
whole though this question was answered very well.
Policy Data
For each policy
Dates on cover
Dates of endorsements and change in cover
All rating factor details:
Trade / Industry / Occupation

Turnover
Number of employees
Health and safety policies
Source of business
Materials handled
Processes involved
Location
Payroll
Exposure
Cover / exclusions
Limit of indemnity
Past claims experience
Size of deductible
Details of risk premiums charged, may need to be a proxy
Claims Data
Date of occurrence
Date claim reported
Date of re-opening
Dates and amounts of payments
Estimates, if they exist, of amounts outstanding
Rating factor details as they were at the time of the claim and/or
Link to policy information
Type of claim, i.e. injury or disease
Type of peril
Page 2
Dates of settlement
Other data
Industry statistics to back up own data where it is sparse
Changes in underwriting / claims handling / policy conditions
Inflation statistics for wage inflation, court award inflation and care inflation
Information about any legislative changes which will have impacted in past and any
that may impact future experience
2 This question asked for the differing claims characteristics. Many candidates listed
claims characteristics without identifying how they differed between the two types of
business. Some candidates also described claims characteristics and then said that
they would be similar. Neither of these approaches would have gained many marks if
any.
Coach fleet may take longer to settle small claims as the insured may be unwilling to
take a coach out of service to rectify minor cosmetic damage.
This will depend on how busy the company is.
Lower theft claim frequency likely for coach fleet
Overall claim frequency could be higher or lower depending upon circumstances, e.g.
mileage travelled, area driven, area kept.
If a coach fleet has a total loss this will be a large claim.
Also if all coaches are stored together overnight there is potential for an aggregation
of risk, e.g. fire. This is less likely to be a risk for a car fleet. Leads to a more skewed
claims cost distribution.
Average claims size of coach fleet third party property damage claims will be higher
than for a car fleet.
This will include some total losses in respect of third party vehicles.
Hence settlement delays will be greater for coach fleets.
For a coach fleet there is a potential accumulation of risk
There will generally be fewer vehicles per fleet for coaches and hence more random
claims experience.
Page 3
Likely to be more overseas exposure for coaches and hence possible notification
delays and currency issues.
Fewer small claims for coaches as generally higher excesses on the policies.
May be different legislation impacting upon coach passenger claimants compared

with car fleets.
If a coach was full of passengers and had an accident could be many individuals hurt
or killed which could lead to much larger claims than if a car had a similar accident.
Dealing with such a coach claim would take a long time and be very expensive.
Again if a coach hit another vehicle it could do more damage than a car and so bodily
injury claims are more likely and likely to have a higher average cost.
3 There were many points that could be made in the solution to this question. Several
candidates however concentrated on the issue of over / under charging and effect
upon expenses. Hence they missed out on several of the other points. Marks for this
question were generally very low.
Policy, claims and expense data are essential for reserving

to ensure sufficient funds are retained to pay future claims
and to highlight any concerning trends
This is also likely to be required for insurance Regulatory purposes, accounts or

otherwise both to satisfy reserving requirements and to demonstrate that the company
is run in a sound manner, e.g. with regards to investment strategy, capital use and
planning.
Similar data is also required for pricing to ensure that there is no adverse selection and
to aim to maximise profits. More accurate / complete data enables better pursuit of
these goals
Relevant, accurate data is required both for assessing appropriate levels of reinsurance
and to provide to reinsurers in order to maximise the chance of obtaining the desired
cover for an appropriate premium.
Such data is also required for persistency, portfolio movements and quote strike rate
management information to assess the business you have on your books at any
particular time.
In a competitive market such as motor, improved data can give the edge in terms of
more accurately pricing risks to win more profitable business and help avoid the less
profitable categories. Without relevant accurate data, poor underwriting decisions
become more likely
Page 4
Customers will not appreciate providing irrelevant information and will not want to
take the time. So the company must restrict info requested to just the minimum to
enable underwriting, in order to ensure volumes of business are optimised.
Exclusions / Excesses must be recorded correctly to ensure that the correct policy
cover information is applied in the event of a claim.
This information underpins management decisions. Poor data may result in poor
decisions being made
Relevant, accurate information may help to prevent fraudulent behaviour
4 This question was generally well answered with some candidates scoring nearly full
marks.
Purpose is to indemnify the policyholder against loss or damage to their commercial

property e.g. office, shop, factory. In addition there may be business interuption cover
Benefit is the amount indemnified for loss or damage subject to limits or excess.
Perils covered are fire principally but also including explosion, lightning, theft, storm
and flood. In addition there may be business interuption.
Exposure used is the sum insured year

however there are two complications:
Stock amounts may fluctuate considerably. Stock may be covered on a declaration

basis, determined retrospectively with an adjustment premium.
No standard way of allowing for inflation in the policy.
Claims characteristics
Claims arise from sudden and determinable events (except subsidence)
Notification delays are short apart from subsidence
Different inflation rates affect buildings, contents and business interuption sections
Possible accumulations owing to location
Generally low claim frequency compared with other property insurance
Many claims likely to be impacted by reinsurance arrangements
Good estimates of claims amounts can be made
Page 5
Settlement delays are generally short but can be increased if there is a need to verify
the value of the stock held in a commercial property or the validity of a claim.
Claims distribution is skewed as claims are fairly consistent in size with few total
losses but more varied than domestic properties as the properties insured are less
homogeneous.
Exposed to moral hazard especially in an economic downturn.

Claims may be reopened.
Large claims will be subject to negotiation / arbitration about the amount of the claim
Rating factors
The main rating factors are monetary value, location, trade / business. Other rating
factors which may be used are:
EML
Age of building
Fire protection equipment
Number of floors / floor area
Surveyors report / score
Construction type
Excesses
Indemnity period for business interuption
Security / links to police station
Claims history
5 Alternative approaches to this question gained marks regarding the detailed amount
of information given as the examiners felt that the question did not make it clear when
the actuarial investigations were due to take place. Marks were not given however for
investigations of a non-actuarial nature. Even allowing for the alternative approaches
most candidates mentioned only a few of the investigations that would be carried out.
Many candidates concentrated on aspects of getting the data rather than
investigations e.g. trends.
Claims related:
Frequency and Average Cost trends

by Accident year, Development year and Payment year
to highlight any potential new developments
cost per unit exposure analysis
large loss ratio
Probably only have statutory accounts and returns available
Page 6
Need to investigate reserving trends in conjunction with the claims related trends
e.g. UPR, (A)URR, OSCR, IBNR

compare with previous years. Are there any noticeable trends?
Also, investigate trends in
claims handling costs and associated reserve.

solvency levels and free reserves
statutory solvency requirements / levels of coverage
market share
portfolio movements
new business levels / premium volume
premium rates charged if obtainable / industry premium levels
the asset mix and any associated changes
Funding of the takeover

Alternative use of funds
Restrictions on purchase anti-competitive laws.
Other accounting ratio investigations important
Loss/Claims ratio, Expense/Combined ratio, Commission rates, Investment

returns, Profit margin, Return on capital employed, Share price / p.e. ratio
reinsurance purchased, recoveries made, reinsurers security

goodwill
taxation and regulation
policy conditions
Check for any trends in the wording of audit statements has there been any apparent
weakening in the sign-off
Investigate recent or possible future legislation changes that may impact the business.
Investigate similar trends in other potential companies
Investigate the credit standing of the company
Investigate the combined model office
Investigate the synergies that the combined operation will have.
Investigate the effect upon the benefit schemes of the two organisations.
Investigate the impact upon data owing to different ICT and product structures.
Investigate any savings in claims handling expenses
Page 7
Investigate the benefit of any diversification by location, different industries and

different materials handled etc.
6 Part (i) of the question asked for relevant limits. Most candidates failed to give
sufficient limits with many not giving any at all. Some candidates did give limits but
selected very high proportion of equities which even for a GI company writing long-
tail business is not appropriate, especially if it is small. A lot of the answers were
regarding general comments about investments for a GI company rather than the
instructions given to the investment mangers. Parts (ii) and (iii) were generally well
answered.
(i) Content will include:
guidelines for the split of assets, giving ranges for each major asset class
this will most likely give maximum and minimum ranges
indicate mean duration, mean maturity, maximum maturity permitted, etc.
Cash (20% 30%)

Fixed interest (50% 70%)
Index linked (20% 40%)
Equity (0% 20%)
Indirect property (up to 10% of the 20% max for equity)
No direct property investment
Domestic currency only
>80% of Fixed interest must be Govt guaranteed
The remainder of Fixed interest must be AA or better
Benchmark for investment return
Limit equities to bluechip or equivalent
Funds under management should be split broadly equal between at least two
different investment managers
For Cash no more than w% of funds under management with any one
deposit holder.
For other assets no more than x% of the insurers total assets should be
invested in any one company in aggregate, etc.
The insurer should not hold more than y% of any particular issue
The insurer should not hold more than z% of any one company s total debt in
aggregate (counter-party risk), etc.
Page 8
(ii) Advantages
Returns may be higher, in which case the company would improve its overall
investment performance
Equity may be better matched for real, long term liabilities
If currently low level of equities then to increase holding could be appropriate

for diversification
Disadvantages
Returns may be lower, in which case the company would see worse
investment performance
Equity has a greater volatility of potential returns, which means there is an

increase in risk
Equities too long to match most liabilities
Capital requirements are likely to be risk-based, which means equity treated as

higher risk
Equity is less liquid, which is disadvantageous in GI business where funds

may be needed at short notice to cover unexpectedly high levels of claims
Equity is less secure, so greater risk of default
(iii) Advantages
Inflation link offers some protection for real liabilities

Highest level of security
Long term should be a suitable match
Lower volatility, so generally lower risk than equity
Possible lower level of dealing expenses
Page 9
Disadvantages
Appropriate term may not be available

The inflation link may not be suitable (wrong inflation)
Overall expected returns are lower
Possibly less liquid / smaller volumes available
7 Most candidates managed to make a reasonable attempt at the first 2 parts of this
question which involved a non standard product. Part (iii) proved very difficult for
most candidates and hence overall the marks for this question were generally on the
low side. The solution below to part (iii) was not the only one that would have been
accepted by the examiners, in particular using different terms in the Taylor Series
expansion would have given a different answer.
(i) Advantages
Guaranteed funding in the event of a large catastrophe enabling the agency to

concentrate on relief efforts rather than fund raising.
Certainty about the amount available for relief efforts.
Enables better planning and budgeting as there is no need to build up

contingency funds etc.
Controls cash flow
Disadvantages
$10m dollars may be perceived to be a high annual cost for insurance and the
relief agency does not have the funds.
Contributions may not be as forthcoming if contributors realise that their

money is being paid over to an insurance company.
If event is near the end of the term of the policy may be difficulty in
determining if the 500 deaths occur during the term of the policy.
If there are no large catastrophes during the year, then the agency will lose the
premium.
The insurance provider is likely to add margins and profit loadings on the
premiums and thus the premium for the policy is likely to be higher than the
expected benefit to the agency.
There may very few providers of such policies in the market i.e. Demand may
outstrip supply and this may lead to the insurance provider overcharging for
the policy.
Page 10
There may be delays in receiving the funds from the insurance provider as it
may want evidence of the loss of life etc.
The agency may be able to raise the required funds anyway from public
sympathy for the victims of the disaster.
Man made disasters such as civil war would not qualify for benefit under the
policy.
The relief agency will be working to prevent loss of life and so may in some
situations not get the benefit under the policy due to its own success. In such
situations, the relief agency may request some financial compensation. This
may lead to disputes with the insurance provider.
Need for a clear definition of what constitutes a natural disaster.

Many of the events that the relief agency is concerned with may not result in
great loss of life but may still be large catastrophes in terms of the numbers of
people requiring assistance.
Effects of inflation may reduce real level of cover
Risk of insurer default.
Number of deaths is not indicative of size of loss
Payment is irrespective of number of deaths once it reaches 500.
(ii) Advantages
Fixed benefit payable on occurrence of an event leading to a valid claim.
Short tailed
Selling such a policy would bring in income from a new source and possibly
open up the possibility of selling such policies to other relief agencies.
There may be marketing advantages to being associated with the relief agency.
The policy would bring catastrophe exposure to parts of the world that the
provider is unlikely to be selling much insurance to. Therefore it would not
aggregate much with its other catastrophe exposures (i.e. diversification).
Pricing may be attractive as few companies in the market would be willing to

write such policies.
Page 11
Disadvantages
The definition of loss under the policy is somewhat subjective. There may be
difficulties in establishing the loss of life from an event. This could lead to
disputes with the relief agency.
It may be difficult to price such a policy as good data may not exist on the
frequency of such events (i.e. there is a greater risk that claims could exceed
premiums than in other situations).
The relief agency would have an incentive to exaggerate the loss of life from a
particular disaster. Independent verification may be difficult to obtain.
An event may occur in a part of the world where the relief agency does not
operate. In such a situation, the insurer would end up paying out even through
the relief agency does not spend that money providing any assistance to the
victims.
Difficult to reinsure at reasonable rate
Volatile claims experience
Liquidity issue in event of a claim
Political downside if do not pay out.
(iii) Loss cost for Company A = 100 * 0.1 = 10m

Loss cost for Company B = 100 * 0.0833 = 8.33m
Company A Premium = 10m + 10% * Capital

Company B Premium = 8.33m + 15% * Capital
Page 12
The Capital is set such that exp ( Capital * R) is approximately equal to 0.1.
where
R is the root of M(R) cR = 0

M(R) 100 (1 + )R = 0.
M(R) = exp(100R) = 1 + 100R + 5000R2 using the Taylor Series expansion
Therefore R is the root of 1 + 100R + 5000R2 1 100R 100R 0.

R = 0.02 where is the safety load
Using the approximation probability of Ruin = exp( R * Capital) = 0.1

exp( 0.02 Capital) = 0.1
Capital = 115.13m /
In the case of Company A, = 10% * Capital / 10m

2 = 115.13m / 100m =
sqrt(1.1513) = 1.073
So Premium = 10m * (1 + 10 * 2.073 = 20.73m
In the case of Company B, = 15% * Capital / 8.33m

2 = 115.13m / 55.53m =
sqrt(2.0731) = 1.4398
So Premium = 8.33m * (1 + = 8.33 * 2.4398 = 20.32m
Hence company A will be more expensive.
8 Most of the candidates made a good attempt at this question, particularly part (i). The
examiners were encouraged by this as this type of question has been poorly answered
in the past. It demonstrated an understanding and interpretation of
accounts. There was some confusion regarding at which point in time the figures
related to, even though the question stated that all figures are as at end 2003.
(i) Class A
GWP is decreasing sharply
NWP is decreasing even faster than GWP.
This could be because the reinsurance premium is rising as a % GWP or that

the company is buying more reinsurance as the size of the portfolio decreases.
NEP = NWP so all the policies are written on the first day of the accounting
period.
Loss ratio is increasing over time.
Page 13
Small proportion of claims are paid a long tail class
Lots of IBNR/IBNER which increases (as % of WP) as get closer to the future
as would expect.
Expenses are not falling as fast as premium.
Could be due to claims handling costs as claims are long tail.
Commission is stable at 15% of GWP, or increases from 20% to 25% as % of

NEP
Investment return is decreasing very slowly.
Investment return is high as % of GWP
Again consistent with a long tail class as substantial reserves will be held and
these will be fairly stable.
Profit has decreased in amount over time but as a percentage of GWP has
increased over time.
This is a result of the decrease of GWP with a stable reserve pot and higher
proportion of investment return.
Class B
GWP is increasing sharply, in fact at the rate that it is falling in class A
NWP is also increasing at the same rate as GWP.
Percentage of GWP paid in reinsurance is constant at 5%.

NEP for 2000 is the same as NWP.
Could mean that same volumes were written in 1999 as in 2000.
NEP for subsequent year suggests that the average policy is written mid way
through the year i.e. that policies are written evenly over the year.
Commission is again stable at 15% of GWP, or around 17% of NEP.
Expenses start at 10% of GWP and increase slightly over time to 12% by
2003.
Investment return is relatively stable at circa 3% of premium. Increasing

slightly as a % GWP.
Page 14
Proportion of claims paid is far greater than class A i.e. this class is shorter
tailed.
IBNR/IBNER moves over time as expected.
Looks like there is a large notified but not paid claim in 2001.
2001 loss ratio for this year deviates sharply from the average.
Profitability has declined over time as % GWP, consistent with the rapid
growth company has undertaken.
Growth at this speed is often not possible without a decline in profit.
(ii) Class A EL, PL anything else long tailed with reasons
Class B any short tailed class, household due to large claim / weather /
subsidence event in 2001
(iii) Issues
Expenses are increasing over time. Need to investigate why this is happening
and identify any opportunities to reduce expenses.
Reinsurance costs on class A seem to be increasing. Need to investigate why

and consider whether current reinsurance arrangements are the most
appropriate.
Profit for A as % of WP increasing in revenue accounts but underwriting

profitability of business written is decreasing, cannot rely on the investment
returns on long tailed class. Look at bringing class B back into profitability.
Changes in mix of business within each class owing to rapid expansion /

contraction.
Investigate whether the reduction in volume in class A and increase in class B

is intentional
Investigate investment strategy.
Investigate capital requirements
Staff and systems, class A will need experienced staff to handle a class that is
clearly decreasing over time. Staff may leave for another company once they
see they have no long term future. Need plan to retain and reassure staff
otherwise will lose ability to service business.
Page 15
EXAMINATION
9 September 2005 (pm)

booklet.
supervisor.
question paper.
ST3 S2005 Institute of Actuaries
1 (i) Explain the difference between accident year accounts and funded accounts.
[2]
(ii) Give four examples when funded accounts could be appropriate. [2]
[Total 4]
2 (i) Define the terms:
(a) suretyship
(b) moral hazard
[2]
(ii) Explain how a general insurance company writing suretyship business can
minimise the risk of moral hazard in respect of the cover it provides. [4]
[Total 6]
3 You are a consulting actuary who has been approached by a general insurance
company, which writes more than one line of insurance business.
(i) You have been asked to comment on the recent relative profitability of the
different lines of business for this company. State, with reasons, the types of
investigations that you will need to carry out in order to do so. [3]
(ii) On completion of the above task, you are asked to comment on the company s
reinsurance structure. List the types of investigations that you would carry
out. [4]
[Total 7]
4 A reinsurance company writes a book of catastrophe reinsurance contracts to an

expected combined ratio of 60%. It estimates that its aggregate claims distribution is
compound Poisson with and the claim size distribution is exponential with
mean of $1m.
(i) Calculate the minimum amount of capital it needs to ensure that its ultimate
probability of ruin stays below 0.5%. [5]
(ii) Ignoring investment income, calculate the return on capital that the reinsurer
would generate if it held the amount of capital that you calculated in (i) above.
[1]
(iii) Suggest potential practical limitations of the above solution. [1]

[Total 7]
ST3 S2005 2
5 You are an actuary working for a general insurance company that writes a wide
variety of classes of insurance. You have been asked to attend a cross-functional
group looking at how to improve the company s defence against fraudulent claims.
As part of your preparation you have been asked to suggest ways in which fraudulent
policyholder behaviour may be reduced.
Outline the suggestions you would make. [14]
6 The government of a small developing country wants to encourage the development

of the agricultural sector in its country. One of the measures it has taken is to
establish a specialist insurance company to provide crop insurance to farmers in its
country.
The insurance company would charge an annual premium and provide compensation
to farmers for crop failure resulting from drought, disease or pests during the policy
year. The compensation provided will equal the sum insured less the proceeds from
the sale of the crop.
(i) List the rating factors that the insurance company might use to set premiums.
[3]
(ii) Describe the characteristics of claims that the insurer can expect to receive. [5]
(iii) State the factors that will influence the level of capital that the insurer will
need. [5]
[Total 13]
7 You are an actuary working for a general insurance company that has been in business
for three years, writing only motor insurance third party liability, in a country that
does not allow claims equalisation or catastrophe reserves.
(i) State the required technical reserves likely to appear in the management
accounts and the matters you would consider in determining the reserving
methods to calculate these reserves. [18]
(ii) Discuss how you would overcome any particular issues or difficulties faced in
applying the methods discussed in (i). [6]
[Total 24]
ST3 S2005 3 PLEASE TURN OVER

8 You are a general insurance actuary working for an insurance company which writes
only household business.
(i) Describe, with examples, the adjustments that may be made to the base
experience to obtain the burning cost premium. [10]
(ii) Explain the adjustments that may be made to the burning cost premium to
calculate the premium the customer is charged when initially taking out a
policy. [10]
(iii) Describe any further adjustments that may be made to the premium in (ii) to
derive a renewal premium. [5]
[Total 25]
END OF PAPER
ST3 S2005 4
EXAMINATION
September 2005
Subject ST3 — General Insurance

EXAMINERS’ REPORT
Introduction
The attached subject report has been written by the Principal Examiner with the aim of
helping candidates. The questions and comments are based around Core Reading as the
interpretation of the syllabus to which the examiners are working. They have however
given credit for any alternative approach or interpretation which they consider to be
reasonable.
M Flaherty
29 November 2005
© Faculty of Actuaries
© Institute of Actuaries
Subject ST3 (General Insurance Specialist Technical) — September 2005 — Examiners’ Report
1 Most candidates had no problems with this question.
(i) Accident year accounts consider all income earned and outgo incurred in a
year and permit the release of profit at the end of the year. Funded accounts
consider the business written in each year and do not permit the release of
profits until the end of a subsequent year (usually the third year).
(ii) Where underwriting year is fundamentally important
Lloyd’s market
Reinsurance written on a policies incepting basis
Significant delays in premium payment and claim settlement and in making

recoveries
Marine and aviation business
Regulatory requirement
2 Credit for (i) part b was given to students who used the alternative definition of
‘Making the insured event more likely to happen as a result of being
insured, e.g. a household contents policyholder not taking as much care in checking
that doors and windows are locked.
Many candidates could not define suretyship and hence missed out on a few of the
points in part (ii) or added points that were not relevant to this type of insurance.
(i) (a) Insurance to provide a guarantee of performance or for the financial

commitments of the insured.
(b) The risk that an insured may attempt to take an unfair advantage of the
insurer, for example by suppressing information relevant to the
assessment of risk or by submitting a false claim.
(ii) Good underwriting
Suitable policy wording / exclusions
Provisions of suitable cover …

… including any limits and excesses
Data sharing with other companies
Analysis of claims information … may highlight patterns
Regular contact with insured to build confidence and understanding
Page 2
Expert claims handling.
Double trigger
Obtain relevant info regarding financial status and obligations of the insured
3 Most candidates managed to make a reasonable attempt at this question.
(i) Reserving analysis to check the expected ultimate liabilities for each line of
business
An analysis of the expenses and levies attributed to each line of business to

check their allocation
An analysis of cashflow for each line of business in order to work out the net
present value of underwriting profits from each line of business.
Determine the required level of capital to support the business
Allocation of the required capital amongst the various lines of business

Determine the return on capital for each line of business.
Investment risk based on asset risk to assess default risk
Investigation of bad debts on each class
Investigation of reinsurance to assess cost effectiveness
Investigate the effect of the insurance cycle for each class to see effect of
likely profitability
Investigate change in mix of business within each class to assess effect upon
changing profitability
Investigate the effect that the occurrence of large claims may have had on each
line of business
(ii) The amount of risk that can be retained safely having regard to the insurer’s
solvency position.
The extent of likely exposure to accumulations of risk
The need for catastrophe reinsurance and the appropriate upper and lower
limits for such cover
The extent for possible need for reinstatement covers
Page 3
Competitiveness of reinsurance prices and availability of cover and technical

assistance
Check on strength and solvency of reinsurer.
Competitors reinsurance structure
Regulatory requirements
Alternatives to reinsurance
Cost and effectiveness of commuting existing covers
Past reinsurance cost / benefit
4 This question generally resulted in either almost full marks or nil. Many candidates
gaining only 1 or 2 marks with the initial definitions but not being able to carry out
the calculation.
(i) Probability of Ruin is approximately equal to Exp (−RU) where
U = Capital required
R = adjustment coefficient = α − λ / c = α θ /(1 + θ)
μ = $1m.
α = 1 / μ = 1
θ = 1 / 0.6 − 1 = 0.6667
R = 1 * 0.6667/(1 + 0.6667) = 0.4
Probability of Ruin = Exp (−0.4 * U) < 0.5%
• 0.4 * U < ln (0.5%)
• 0.4 * U < −5.2983

U > 13.25.
I.e. the capital required is greater than $13.25m
(ii) Return on Capital = 0.2 * 0.6667 / 13.25 = 1.0%
(iii) Model not appropriate
Approximations used not correct or not appropriate
Regulatory requirements may be different
Ignores other business that may be written by the company
Page 4
Future experience different from past
5 This question was on the whole better answered than the examiners had expected. The
better candidates were able to show that they had considered many ways in which the
fraudulent behaviour could be reduced.
X-check policy cover dates against date of accident
Look for multiple claims from the same claimant, surname, address, postcode, etc.
Tighten underwriting criteria
Check previous claims history of the policyholder at inception
Introduce some form of experience rating
Increase / introduce excesses
Spot checks on claims of various sizes
Set up confidential fraud line with possible incentives
Write indemnity only policies
Set up fraud department
Attempt to identify any oddities about the claim, such as flood damage on a dry day
Corroborate info provided by claimant with independent sources. E.g. witnesses
Review previous claims history — any repeat claims in internal data
Consider data sharing with other companies — E.g. highlight any double coverage
Collaborate with police, media and other insurers to advertise penalties for fraud
using specific example cases previously discovered
Use of in-house / appointed repairers / loss adjusters
Use of own home service sales force
Settle claims by replacement items rather than cash settlement
Ensure policy wording is as tight as can be with appropriate declarations of facts

provided
Send out claims handler to view claim incident where economically viable
Page 5
Ensure claims handlers are well trained and receive regular refreshers
Maintain good links / relations with relevant police and related authorities and share
information
Require original receipts to demonstrate value

Require medical evidence from recognised professional
Highlight penalties for fraud on policy wording
Use of voice recognition techniques and recording of telephone calls for purposes of
lie detection / claim verification
6 Again the examiners were pleased with the general standard of answers for this
question with candidates demonstrating that they could apply their knowledge to a
non standard GI product.
(i)
• Type and mix of crops grown
• Geographic region in which the crops are grown
• Size of the farm
• Availability / method of irrigation
• Pest control techniques used
• Level of excess
• Claims history
• Sum insured.
(ii)
• Most claims will be reported at the end of the growing season although
some will be notified during the season.
• Most claims will be settled quickly, i.e. very soon after the growing season
• Claims size will be related to the size of farms in the country.
• Claim frequency likely to be low in most years
• There will be significant accumulations of claims from adverse weather

conditions / pest epidemics.
• Accumulations might occur from the same geographic region or type of

crop.
• If a period of drought / disease / pest extends over several years, then the
insurer could face several years of high losses.
Page 6
• Potential for fraudulent claims if a farmer does not tend his crops properly
and then claims under the insurance policy.
• Potential for moral hazard as farmer does not aim for highest sale price for
crop as insurance will pay anyway
• Short tail, therefore relatively little impact from claims inflation.
• Quick reinsurance recoveries
• Potential for claim disputes regarding definition of drought / cause of crop

failure.
(iii) Likelihood of accumulations and catastrophes.
Volatility of claims experience
Level of uncertainty from poor data quality for pricing
Liquidity risk
Credit risk from reinsurers / policyholders / third parties
The level of premiums charged in relation to expected claims (loading factor /

risk margin)
Desired level of ruin probability.

Expected volume of business
Level and variability of expenses
Level and variability of investment income that the insurer is expected to

generate
Reinsurance structure and price.

Regulatory requirements. There may be minimum capital requirements.
Rating agencies. The insurer may wish to achieve a particular rating.
Implicit or explicit guarantees from the government with regard to capital

support in future.
Required profitability
Dividend policy
7 Credit was given in either part of the candidates’ answer for valid comments listed
below in both parts (i) and (ii).
Page 7
Although most candidates were able to list the required technical reserves, there was
generally not enough discussion as to the matters to be considered in calculating such
reserves. Most candidates did comment upon the fact that the company had only been
in business for 3 years but did not go into enough detail as to how to deal with this.
A few candidates failed to realise that third party liability insurance also covered
third party liability to property damage and instead only considered third party
liability to bodily injury claims.
(i) All technical reserves to be stated on a gross and net basis
Consideration of any regulatory issues
Consideration of level of margin in reserves
Consistency with existing methods
Consider currency of reserves
Unearned Premium Reserve
365ths method most accurate …

… when risk is evenly spread
Calculates unearned premium by multiplying office premium …

… by ratio (365 — # days since inception) / 365
But requires computer records, which should be available
24th / other similar methods less accurate
Risk may not be evenly spread though …

… need to investigate claims experience to date …
… and probably industry data …
… to devise a suitable earnings formula
DAC
UPR may be directly reduced for acquisition costs …

… or a DAC can be created as an asset to offset the overstatement
Additional Unexpired Risk Reserve
Required where initial premium is considered insufficient to cover the

estimated ultimate claims outgo
Might assume not required, if confident in original pricing basis
However, could compare with industry claims experience …

… there may be recent trends apparent since business written
Page 8
O/S Reported Claims Reserve / IBNER
Case estimation likely to prove most fruitful …

… as low frequency, high severity expected for this class …
… and because the company is very young
Separate allowance is required for IBNR
Chain ladder techniques unlikely to be of any use for bodily injury…

… as too little data available …
… and not even close to being fully run-off
Chain ladder could be considered for property damage but need to consider
any changes in settlement procedures
Bornhuetter-Ferguson method may be helpful though…

… but must be confident in ultimate loss ratios
Allowance for discounting
If using an average cost per claim method then need to allow for changes in nil
claims
IBNR
Could take a simple proportion of premium / o/s claims reserve …

… however, not very robust as loss ratio may not be stable / affected by
adverse claims experience
Delay table method not viable as insufficient stats available
Projection method likely to be of most use, …

… but need an ultimate loss estimate …
… perhaps there are industry stats available to help.
Re-opened claims reserve if not included within O/S reported reserve
Claims Expense Reserve
With case estimation of o/s claims can include allowance for direct expenses
Or can look at industry estimates to assess a proportional addition
Indirect expenses will need to be based on prior estimates from business plan /
model projections …
… as the business is still very young
Page 9
(ii) Split the data between bodily injury and property damage
Try to use several different methods for estimating each reserve to ensure that
the resulting reserves are sensible
Approach reinsurers for technical assistance, as they are likely to have

significantly more experience on which to base estimates.
Actuarial consultants can provide similar assistance.
More frequent reviews of claims and expense stats will ensure that most up-to-
date info can be fed into the process
Future inflation may be different to past inflation in which case would need to
allow explicitly for inflation in projections
May have been trends in the past which need to be allowed for, e.g. in respect
of legislative changes. All claim costs would be required to be put onto the
same basis.
There may have been a large individual or catastrophe claim in the three years.
Would need to adjust the data for this by e.g. removing such an item from the
base data and making a separate allowance at the end.
Obtain any industry data
8 This was a bookwork question aimed at a particular class of business. Many

candidates managed to only mention a few of the adjustments required and
even fewer examples.
Part (iii) of the question called for a comparison of a renewal premium with a new
business premium at a point in time. Some candidates interpreted the question to
mean how the renewal premium for an individual policyholder would be derived from
the new business premium the policyholder was charged the previous year. The
examiners considered this interpretation but did not consider it valid.
(i)
• Unusually light / heavy experience. Claims experience fluctuates over
time. An example in household is that some years suffer from storms
more than others. If the experience of the base period does not appear
typical the insurer must choose another base period, aggregate more years
experience or apply an adjustment factor to the affected base year. Factor
will be subjective.
• Large / exceptional claims. Have to decide whether to leave claims in,

truncate them or remove them. This will depend on the extent to which
Page 10
claims of this type are expected to occur in future. Examples in household

would be a large public liability claim or a total loss of a large property.
• Trends in claims experience. Investigate any trends in base data to see if

they are likely to occur in future. If so need to project the trend and
include. If not likely to occur in future then exclude. Rate of increase of
cost of building work is likely to be a trend in household business.
• Changes in risk over time can be very difficult to deal with. Could be
dealt with as trends. Or may separate them out, project and combine with
explicit assumptions about the future mix of these risks. This will also
need to be done for different types of claims if the mix of claims is
changing significantly. Example is that due to global warming the risk
presented by weather is changing over time with wetter more flood prone
winters.
• Changes in cover can also be difficult to allow for. If perils are no longer
insured the insurer may be able to exclude from the base data all claims
that wouldn’t be covered in future. Example if subsidence stopped being
covered then all claims of this type could be excluded from the analysis.
However if a new peril is introduced then it is harder to make an
adjustment. Here external data will need to be used such as market stats,
government data or consumer stats.
If the change is to the level of limits or excesses then it is more

complicated. May be possible to allow for increases in these by looking at
individual claims e.g. if household excess increases from $50 to $100 then
add the old $50 excess onto amount insurer paid out and then adjust for the
effect of the new excess before analysing the claims.
If the detailed claims information is not available then an approximation

will have to be made using the claims cost distribution. Many insureds
will not inform the insurer of losses that occur below the excess point.
• Changes in the impact of reinsurance will need to be allowed for e.g. cat
XL retention changes assuming the calculations are performed net of
reinsurance.
(ii) Once a burning cost premium has been calculated it will need to be projected
forwards to give the future risk premium. This is a central risk estimate of the
cost of future claims.
Claims inflation needs to be allowed for from the mean payment date of
claims in the base period and the mean payment date of claims arising during
the exposure period of the new rating series.
Ideally the insurer will be able to adjust claims values by a specific index for
the loss type.
Page 11
In order to get to a risk premium rate the projected claim cost must be divided
by a corresponding projected value of the exposure. Again need to project at
an appropriate rate of inflation. This could be very different to what is used
for the claims.
Commission is usually a percentage of premium and would be loaded in this

way. If a different way of paying commission is used this should be reflected
in the premiums.
Expenses, divide into fixed and variable. Often load these separately. May
also here load only the expenses incurred in writing a piece of new business.
Also include tax, levies and any industry wide compensation schemes.
Allow for smoothing across rating cells.
Allow for new business and/or joint policy discounts.
Allow for any regulatory issues which may affect what rates may be charged.
Investment return should be allowed for by discounting expected claims

payments and expenses to the date on which the premium is received. As
household is not a long tail class this is less important. Allowance should be
made for whether premiums are payable monthly or annually.
Reinsurance costs should be allowed for. Usually loaded as a percentage. As

household insurance as a minimum there would be catastrophe XoL cover.
Contingencies should be allowed for. Often a percentage loading, this can be

considered with the profit loading.
Return on Capital / Profit Margin. Need to load a profit margin having

regards to what is required to give a reasonable return on the capital needed to
support the risks underwritten. Rate of return should correspond to the risk to
which the capital is exposed.
Competitive considerations may affect premiums. Insurer may decide to take

a smaller return in order to sell more policies and therefore make a lower $
profit than otherwise. Need to consider where we are in the insurance cycle.
Also look at competitors prices and assess likely volumes. Household
insurance is not as price sensitive as motor insurance. Compare to previous
rating series and assess effects of changing to new set.
(iii) Renewals different as:
• Expenses could be different, costs less to renew a policy than write a new
one. Could reflect in premium.
• Case by case underwriting could be applied at renewal if not possible at

new business e.g. business acquired through the internet.
Page 12
• Commission paid could be different
• Competitive considerations, if policyholder is likely to renew then may

increase premium to more than would charge a new customer. And vice
versa.
• Business plan, may prefer to retain business at a lower margin, especially

if seen as a good risk / will enhance risk profile.
• May be gradually moving to a new rating set so no direct correspondence

between new business and renewal rates.
• May have a team who have premium flexibility in order to maximise

retention of business, this will again make premiums different from new
business ones.
END OF EXAMINERS’ REPORT
Page 13
EXAMINATION
6 April 2006 (pm)

booklet.
supervisor.
question paper.
ST3 A2006 Institute of Actuaries
1 You are the actuary for a small general insurance company with a low solvency
margin. The insurer only writes commercial property and household risks which are
located in a small country.
(i) Explain the differing claims characteristics of these two classes of insurance.
[3]
(ii) Suggest, with reasons, the types of reinsurance that this company might
purchase. [7]
[Total 10]
2 You are the actuary for a general insurance company writing household buildings and
contents insurance.
(i) List the information you would seek to determine estimates of exposure for
future flood damage. [3]
(ii) Describe how the total exposure to a major flood loss could be assessed using
the information in (i). [7]
(iii) Discuss the degree of difficulty that you would expect to arise in obtaining the
information. [3]
[Total 13]
3 An insurance company specialises in writing commercial property insurance business.

As part of the company s business planning, it prepares financial projections using a
number of stress tests. One such stress test makes the following assumptions:
(a) A geographic region where the company writes 25% of its business is
subjected to a strong earthquake and
(b) A rise in interest rates causes a 20% fall in the bond market.
Describe the particular risks that the company would face under such a scenario.
[15]
ST3 A2006 2
4 You are an actuary working for a reinsurance company. You have been asked to help
price a new quota share reinsurance contract protecting a short tail class of business.
You have been given the following historical data from 2001 to 2005 underwriting
years on the business that is subject to the quota share.
Year Written Premium Incurred Losses Rate Change

£m £m
2001 120 144 N/A

2002 130 143 50%
2003 150 120 10%
2004 150 120 5%
2005 140 126 0%
Rate change figures indicate the percentage increase / decrease in premiums collected
from an identical risk from one year to another.
You also know that claims inflation has been running at 4% per annum over this
period for this class of business and that premium rates in 2006 are expected to fall by
5%.
(i) Calculate the loss ratio that can be expected for 2006 for this class based on all
of the historical experience and the rate change and inflation assumptions
given above, stating any assumptions you make. [9]
(ii) State other information that you would require in order to determine if this
quota share contract is expected to be profitable for 2006. [5]
(iii) The ceding company has told you that it expects its loss ratio in 2006 to be
85%. Suggest reasons why the ceding company s expected loss ratio for this
contract might be different from the value calculated in part (i). [5]
[Total 19]
ST3 A2006 3 PLEASE TURN OVER

5 (i) Explain how you would obtain an indication of the strength of a general
insurance company s claim reserves based on published information. [5]
Company ABC is a general insurance company. The following information is

available:
2001 2002 2003 2004 2005
Gross written premium (GWP) 2,000 2,500 3,000 3,250 2,000

Outstanding claims b/fwd 100 465 927 1447
Policy count at mid-year 30,000 32,000 35,000
All monetary amounts are $000s.
Unearned premium reserve brought forward into each year is 10% of the GWP in the
previous year. Claims paid in each year are 60% of the GWP in that year. Over the
period 2002 to 2005 commission has been paid to intermediaries at 8% of the gross
written premium. Expenses of writing and handling claims on the business are $5 per
policy and 2% of claims paid in the year respectively. Reinsurance costs are 5% of
gross written premium. Investment return is earned at a rate of 5%. Cost of capital and
free reserves are to be ignored in calculating investment return.
(ii) Using the above information, and stating any assumptions you make, calculate
the pre tax profit ABC made in 2002, 2003 and 2004. [9]
(iii) Comment on the results in (ii). [4]

[Total 18]
6 You are the pricing actuary for a general insurance company that underwrites event
insurance. You have been approached by a charity that is organising a fun run for up
to 2,000 people and wishes to arrange suitable insurance cover. To maximise possible
attendance the organisers will be accepting entries on the day. Your company is not
authorised to write motor insurance.
(i) Describe the types of insurance cover, excluding motor insurance, that the
event organising company may wish to purchase. [8]
(ii) List, with reasons, the information you might require from the event organiser
in order to establish the risk premium for this event. [10]
(iii) List, with reasons, the policy exclusions you would expect to see under each
of the products described in part (i). [7]
[Total 25]
END OF PAPER
ST3 A2006 4
EXAMINATION
April 2006

EXAMINERS REPORT
Introduction
interpretation of the syllabus to which the examiners are working. They have however given
credit for any alternative approach or interpretation which they consider to be reasonable.
M Flaherty
June 2006
Comments
Individual comments are shown after each question.
Institute of Actuaries
Subject ST3 (General Insurance Specialist Technical) April 2006 Examiners Report
1 (i) Commercial Property
More scattered cost distribution than household due to singular nature of

properties involved.
Settlement can be delayed by the need to verify the value of stock / need to
liaise with other insurers if only cover a part of the damaged property.
Also business interruption cover will be longer tailed as need to see how long
the business is out of action before the lost profits can be quantified.
Claim frequency tends to be lower for Commercial Property/higher for

Household
Household
Consistent in size and distribution with a small number of larger total losses.
Settlement usually quick.
A subsidence event may cause an uplift in the number of claims and average
cost would also increase (more dramatic effect than would be for commercial
property).
Possibility of geographic concentration.
(ii) Company is small with low solvency margin so protection of solvency

position is a primary concern.
All types of reinsurance will assist with protection of the solvency margin.
Stop loss reinsurance could be purchased to protect solvency margin.
Often it isn t available and certainly not at a reasonable price.
With all types of reinsurance the value for money should be considered, i.e.
the protection it brings, including security status of reinsurer compared to the
cost.
Commercial property is insured so some use of surplus reinsurance to allow

to write a variety of risks of a variety of sizes.
Also allows to fine tune exposure by ceding more of the risk in areas where
already have some properties on risk.
The small size of the insurer may indicate that quota share needs to be used to
allow more risks to be written, giving a more balanced risk profile.
Quota share reduces NWP compared to free reserves, helping the solvency
position.
Page 2
This may also encourage reciprocity, relevant as write Household and this may
be regionally based e.g. Building Society.
Only property insurance is written in a small country so cat XL will be

required to protect against insolvency in the event of a catastrophe occurring.
Assuming perils exist in this country (like flood and subsidence) that give rise
to large claims for individual properties then risk XL will be needed.
Also if there is potential for aggregations (from a defined peril or a geographic

concentration) then aggregate XL will be needed to cover the portfolio.
The need for both these types of reinsurance will be exacerbated as the
company is small.
Financial reinsurance to protect balance sheet/transfer risk to reinsurer
Any regulatory requirements must be complied with including those that affect
the choice of reinsurer / reinsurance programme.
Comments on question 1: A bookwork question generally reasonably well answered. Only a

few identified subsidence/BI settlement delay issues for Household/Commercial Property
respectively. The best candidates identified all the reinsurance types and how each type of
reinsurance helps protect the solvency position (which is a big issue for a small company).
2 (i) &(ii)
All relevant standard rating factors for Household Insurance (sum insured,
postcode, etc)
Any flood excesses
Height of property above mean sea level.
Number of floors of house.
For a flat which floor it is on.
Distance and height above closest major river/lake.
Height of sea defences/river defences.
Quality of defences (earth/concrete etc.).
Long term weather patterns/climate trends (e.g. predicted rainfall)
Previous experience of flood claims
Reinsurance coverage details
Trends in sea level as a result of global warming.
Total exposure to flood
Note difference between coastal flooding from the sea and river flooding.
Sea flooding due to increased volumes of water and salinity liable to be much
more destructive than fresh water.
Sea flooding will only affect properties fairly close to sea level
River flooding affects properties on the flood plain of rivers.
Page 3
Estimation for flooding
Obtain history of max water levels over say last 50 years

Project maximum likely water levels in future
Identify all properties which lie at or below this level
Obtain history of max sea levels over say last fifty years.
Project maximum likely sea level.
Identify all properties which lie at or below this level.
Use postcode/full address to determine location.
By use of ordnance survey maps, or other contour map identify which

postcodes are at risk.
Noting that not all of a postcode may be at risk make qualitative

judgement as to proportion of properties at risk within each postcode.
Noting that even a major flood will generally not result in each affected
property being a total loss use either historical data or judgement to
estimate average loss as proportions of sum insured.
Add together the sum insured for all exposed properties adjusted by the
two factors calculated above.
Any government directives
For alternative well reasoned approach to estimating flood exposure.
(iii) From Proposal Form
Easy to obtain standard rating factors from proposal form
Height and distance are not easily derived from postcode.
Height above sea level most homeowners probably do not know exactly so
difficult.
Number of floors should be simple.
Floor of flat no difficulty.
Distance and height above nearest river again most homeowners probably
do not know exactly so difficult.
With all questions on proposal form there may be a moral hazard if proposer
aware that flood exposed property may attract higher premium.
Page 4
Other sources
Height of sea defences can easily survey a particular part of the coastal
defences.
Quality of defences the coastline may be long so would be costly.
Global warming a lot has been written on this.
Global warming/weather patterns, lots of info already available
Previous claims experience, possible industry and own experience
Future sea levels timing unclear.
Broker or policy holder resistance to changing information required.
Comments on question 2:
For this question in particular points in answer to (i), (ii) and possibly (iii) were given if
answered in any section.
Answers to this question were generally poor. Only a few candidates suggested practical and
realistic methods of determining total exposure to a major flood loss or considered the
different types of major flood risk (sea, river, lake) and the different risk consequences of
each. Some candidates discussed burst pipes which isn't a major flood loss. In part (i) better
candidates thought through a variety of sources of information, but most just listed the
detailed policy and claim information that would be available on a home policy. For part (ii)
often candidates just explained how to do a rating exercise with no reference to flood. Some
candidates seemed to think one company would have enough historic claims data to rate
solely on their past experience hence demonstrating that they didn't understand the peril and
frequency with which it occurs. Part (iii) was poorly answered with many candidates not
going into sufficient detail about the degree of difficulty in getting each piece of information.
3 Risks relating to level of claims reserves
The company will have a lot of uncertainty about the level of reserves to set aside to
pay the claims from the earthquake.
The gross losses that it will have will depend upon:
The amount of damage caused by the earthquake.
The amount of exposure (in terms of insured properties and their sums insured)
that the company has in the affected region.
Coverage provided by the insurance policies that the company writes (property
damage, business interruption etc.). The company may find itself in dispute with
some of its policy holders if the policy wording is ambiguous.
Claims inflation following the earthquake may be higher than the company
anticipated when pricing the policies. (due to shortage of building materials and
labour).
Page 5
The company will have to incur increased claims handling costs e.g. loss adjusters
fees, etc.
Moral hazard from policyholders inflating claims and unscrupulous builders

encouraging this.
Lack of control by authorities: looting/arson; increasing losses.
Timing uncertainty: delays due to scale of event.
Political/altruistic pressure to pay claims up front.
Risks relating to reinsurance
Appropriateness of reinsurance protection e.g. Whether the company has purchased

catastrophe reinsurance.
Adequacy of cover. Is there any danger that the reinsurance cover is exhausted?
Possible increased cost of reinsurance in subsequent years.
Need to buy additional reinsurance for the current year (if any is indeed exhausted).
Availability of reinsurance following the earthquake.
Solvency of reinsurers.
Multiple exposures for reinsurer from other insurers.
Need for bad debt provisions on potential reinsurance recoveries.
Possible disputes with reinsurers.
Risks relating to investments
If the company holds bonds as part of its assets, then it will suffer a fall in the market
value of its investments.
The company may have to sell those assets at a depressed price to pay its claims from
the earthquake.
The company may have to change its investment strategy in line with its lower
solvency level.
The company may have to increase contributions into its staff pension scheme if the
scheme has suffered losses due to the bond market fall and this fall is not offset by an
increase in expected future investment income.
Page 6
Risks relating to new business
There will be uncertainty about the premium levels that the company will be able to
charge after the earthquake.
The company would be expecting premiums to rise but they may not rise as much as
expected.
Also the company may not have enough capital to write the level of premium that it
wants for future periods.
Business risks
Contributions to pension fund increase/salary rises impacting contributions.
The company may be downgraded by the rating agencies if it suffers a significant

loss. This would affect its ability to write business in the future.
Company insolvency
Regulatory intervention
The company may not be able to meet its regulatory solvency requirements and be
subjected to regulatory intervention.
The company may have to raise capital at a depressed share price (share issue) or
higher interest rates (debt issue).
The company may have to pay into an industry compensation fund if other insurers
become insolvent.
The company s own operations may be affected by the earthquake e.g. damage to
offices, power outages, disruption to communications etc.
The increased level of claims after the earthquake may lead to increased frequency of
operational losses e.g. mistakes in paying claims, fraud, etc.
The company s key business partners (e.g. brokers) may be affected by the earthquake
and could lead to loss of business or defaults amongst the company s debtors.
Brokers and policyholders may choose not to renew cover with this insurer due to its
weaker financial strength.
The company may have to post collateral to back its obligations to policyholders (e.g.
Letters of credit or other security arrangements)
Comments on question 3: Candidates who scored well tended to focus on different types of
risk, not just claims, reserving, reinsurance and investment issues, but also considering
operational and other third party risks. Many candidates tended to focus on one aspect such
as claims or solvency and overdevelop this at the expense of other aspects.
Page 7
4 (i)
Underwriting premium
Year Inflation index index Inflated losses on level premium loss ratio
2005 1.04 =1.04 0.95 =(1-0.05) 131.0 133.0 98.5%
2004 1.08 =1.04^2 0.95 =0.95*(1+0) 129.8 142.5 91.1%
2003 1.12 =1.04^3 1.00 =0.95*(1+0.05) 135.0 149.6 90.2%
2002 1.17 =1.04^4 1.10 =1.00*(1+0.10) 167.3 142.6 117.3%
2001 1.22 =1.04^5 1.65 =1.10*1.5 175.2 197.5 88.7%
738.3 765.3 96.5%
Weighted average loss ratio: 97.2%

Simple average loss ratio: 96.5%
Assuming 4% future inflation
Assuming similar policy conditions/coverages
Assuming incurred losses include IBNR/IBNER
Assuming no change in business mix
(ii) More information about rate changes for 2006

Projected volumes of business in 2006 (Written Premium)
Information on large or unusual exposures
Information of large loss experience.
Any IBNR, particularly for 2005.
Commissions including profit commissions, brokerage, overrider etc.
Loads for internal expenses both fixed and variable.
Taxes and any other levies.
Any costs of retrocession.
Investment income.
For this we need to know payout pattern and premium receipt pattern as well
as investment yields on suitable assets.
Relevant comments on changes in any of above.
(iii) The past 5 years may be too short a time frame to capture the extremes of the
loss experience.
For example, if this portfolio of business has had unusually heavy claims
experience in the last 5 years, then the average loss ratio over a longer time
period may be lower.
There will be differences of opinion from the buyer s and seller s perspectives.
Page 8
Errors in data supplied or errors in calculation
Oversimplification or differences in calculation/method used
Claims expenses/reinsurance coverage changes of direct insurer.
Difference in definition of premiums and claims (i.e. gross/net of commission)
There may be difference in opinion of:
Coverage provided in 2006 vs historic years.

Type of claims incurred.
Claim inflation.
Legislation or regulation that affects claims.
Propensity to claim.
This means that the future experience may be very different from the past.
The company may be planning to target a particular segment of the portfolio

and hence increase its exposure to that part.
There may be changes in underlying economic conditions / crime rates that

affect the claims experience.
Comments on question 4: Candidates responses were mixed. Some candidates did not
produce a loss ratio as asked in the question. A number of candidates did not read the
question that the data was by underwriting year and decided to convert premiums to accident
year using certain assumptions. Some candidates did not use all the information given (e.g.
only using 2005 data to estimate 2006). Several candidates inflated premiums in the same
way as claims rather than the correct method of adjusting for rate changes. Very few
candidates actually provided any appropriate assumptions thereby losing out on marks. Parts
(ii) and (iii) were reasonably well answered, with the best candidates giving a range of
possible different reasons why the figures differed in part (iii).
5 (i) Compare results against competitors writing similar business
Examine individual accounting items gross and net of reinsurance.
Examine ratios, within and across years.
Less information will be available as only looking at published information.
Credit rating.
Prior year reserve adjustments
Level of discounting if disclosed or used
Evidence of additional contingency reserves
Page 9
Ratios to be considered are:
Notified claims reserves to premium income

IBNR claims reserves to premium income
Claims expenses reserves to outstanding claims reserves
outstanding claims reserve to claims paid
(ii) Accounts
2002 2003 2004
GWP 2500 3000 3250

UPR b/f 200 250 300
UPR c/f 250 300 325
Earned Premium 2450 2950 3225
Claims paid 1500 1800 1950

Claims reserve c/f 465 927 1447
Claims reserve b/f 100 465 927
Claims incurred 1865 2262 2470
Commission 200 240 260
Expenses $5 per policy 150 160 175

Expenses 2% claims pd 30 36 39
Expenses total 180 196 214
Reinsurance 125 150 162.5
Investment return 25.4 48.6 75
Insurance return 105.4 150.6 193.5
Assumptions
No AURR is required.
Assume no DAC allowed for or alternatively used correctly in calculation
No reinsurance recoveries are made or they are included in the figures given.
Using the number of policies at 30/6 is a fair approximation to number on risk

over the year.
Investment return 5% of average of total reserves for each year.
(iii) Claims ratio is stable at 76/77%.
Average written premium per policy increased from 2002 to 2003 and then
stayed the same 2003 to 2004.
Page 10
Average claims per policy are increasing at similar rate as loss ratio is stable
Expenses as % of premium decreasing slightly over time.
Will always be fairly constant per policy as most of expense is per policy
Insurance return (i.e. profit) is increasing in absolute terms also as % GEP and
GWP.
Average premium has increased but investment return increased more than
proportionately due to increasing reserves.
Relevant comment about strengthening of reserves
Relevant calculations to support strengthing of reserves comment.
Comments on question 5: Part (i) was generally not well answered. A number of candidates
thought that it was important to have strong free reserves or free assets when the question
clearly stated "claims reserves". Only a few candidates suggested comparing different ratios
between companies. Part (ii) was generally reasonably well answered except for investment
income where, in a large number of cases investment return was calculated on outstanding
claims only (technical reserves do include UPR, many candidates did not seem to know this!)
and for a good number of candidates on Profit which is just wrong! Again, very few
candidates stated any assumptions they were making. Part (iii) was poorly answered with
very few candidates scoring well: many candidates did not comment on the level of reserve
strengthening taking place nor any trends in the data year on year.
6 (i) Public liability

for claims made by members of the public for
damage, claimants costs and expenses
in respect of accidental bodily injury
or property damage.
Employers liability
for claims made by temporary staff, helpers and volunteers.
Negligence on the part of company (comment made under either EL or PL

cover)
Property damage all risks, covering

loss or damage to insured property at the event, including transit to and from
the event.
Cost of hiring replacement items for the event.
Accidental loss or destruction to money whilst in transit or in storage at any

authorised residence.
Page 11
Cancellation insurance
Covering cancellation, abandonment or postponement of the event due to

reasons beyond the control of the organiser.
Non-appearance cover
Covering the cancellation, abandonment or postponement of the event due to

the non-appearance of specified persons due to reasons beyond their control or
the organisers.
Theft by employees/Fidelity guarantee
Terrorism cover
Covering the cancellation, abandonment or postponement of the event due to,

or damage / injury caused by, terrorist activities
(ii) Sum Insured/ limits of indemnity for each aspect of cover.
Date of proposed event.
Duration of proposed event.
Length/Terrain of course
Location of proposed event.
Estimated number of participants.
Estimated number of onlookers.
Has the event been held before.
If so, what differences are there this year.
Past claims (or incidence) experience

as this may affect the likelihood of accidents.
Will there be trained marshals / police providing road supervision

as this will affect the exposure to risk of accidents for participants.
Numbers of helpers and volunteers

as this will affect the exposure to potential injury of employee
and also affects the exposure to risk of accidents for participants.
Details of any equipment that will be required

in order to determine the potential costs involved in the event of damage
and additional hire costs.
Page 12
Expected costs, budget, expenses and net profit from the event
in order to estimate the potential impact of cancellation
and the potential impact in respect of loss / destruction of cash.
Details of celebrities/officials due to attend the event (for non-appearance

cover).
Levels of excess per cover

Dangerous peripheral activities on day of event (e.g. firework display/fun
fair).
(iii) Any excess applicable under each element of cover.
Deliberate/criminal acts of negligence.
Public and Employers liability:
Participant liability as participants will be responsible for their own conduct

(and related insurance).
Road Traffic Act motor liability
Damage to property in the custody or control of the insured.
Property damage all risks, covering:
Theft / losses from unattended vehicles

Exclusion of losses not reported to police authorities
Loss / damage due to Terrorism
Theft where there is no sign of forcible or violent entry.
Damage due to nuclear, biological or chemical risks.
Cancellation insurance
Anything within the control of the organiser.
Anything arising from breach of contract.
Due to lack of or inadequate attendance.
Non-appearance cover
Page 13
Any loss within the control of the organiser or participant.
Any pre-existing medical conditions.
Terrorism cover may be excluded.
Comments on question 6: Many candidates adopted a scattergun approach listing all the
types of insurance they could think of. Better answers showed more discrimination and
showed they had thought about the question. Also poorer answers seemed to think that
insurers would collect individual information on all participants, which may be nice but not
practical nor economic. The best candidates were those that identified most of the relevant
covers that needed to be considered. A number of candidates mistakenly decided that PL
would cover fun run participants as well when, in reality, this would be an exclusion. A few
candidates also commented on insurance coverage of the charity's own premises when the
question was about event insurance. Part (ii) was answered reasonably well by most
candidates. Part (iii) was not answered well with only a few candidates having a grasp of
what exclusions would be made on any of the covers taken out.
END OF EXAMINERS REPORT
Page 14
EXAMINATION

booklet.
supervisor.
question paper.
ST3 S2006 Institute of Actuaries
1 Describe ways in which an insurance regulator can influence the investment policy of
a general insurance company. [3]
2 Define the following terms and explain why they are used in an insurance contract
(a) Risk attaching basis

(b) First loss
(c) Discovery period
[6]
3 (i) Define the term Average (as used specifically in non-marine general
insurance). [1]
(ii) Explain, with an example, why it may be used in non-marine general

insurance. [2]
(iii) Suggest alternative approaches available to a general insurance company in

situations where the term Average might apply. [3]
[Total 6]
4 State the criteria that are desirable for a risk to be insurable, explaining why these
criteria are desirable. [8]
5 (i) Explain the general principle of experience rating systems and comment on
the benefits and drawbacks of such systems to both the insurer and the insured.
[6]
A bus company operates nationwide, with both local bus services and inter-city
coaches. It has for some time self-insured its vehicles. The company has an
established department which deals with all claims made on the bus company. The
company is now seeking a quotation from an insurance company to cover the liability
arising from all its fleet risks, both from its passengers and from third parties.
(ii) Describe the statistics and other information that the insurance company
would ideally like to have from the bus company in order to assess the likely
future cost of insuring the fleet. [10]
[Total 16]
ST3 S2006 2
6 You are an actuarial entrepreneur in the process of setting up a new general insurance
company to underwrite motor insurance only direct through the internet. You are
midway through developing your IT requirements. The aspect you are currently
considering relates to the data you propose to collect and hold.
(i) Outline the data-related issues that you will need to consider in establishing
your IT requirements. [10]
(ii) Explain the problems that may occur as a result of using inaccurate data for
determining the initial premium rates that you intend to charge. [6]
[Total 16]
7 You are the reserving actuary for a small general insurance company. You are
considering the claims reserve estimate produced by your team for household
buildings and contents insurance, which has been calculated using purely statistical
methods and paid data only.
(i) Describe the main issues that may result in adjustments to the data being
required in order to ensure maximum reliability of your final estimate. [10]
When you compare your final adjusted statistical estimate for the claims reserve with
the sum of the case estimates from your loss adjusters you find that there is a
significant difference between the two totals.
(ii) Explain why this may have occurred and how you might investigate the
reasons for the difference. [10]
[Total 20]
ST3 S2006 3 PLEASE TURN OVER

8 A general insurance company underwrites two classes of business, personal motor and
commercial property. In personal motor business it expects to write £50m of
premium with expected claims of £30m and expenses of £15m. In commercial
property, it also expects to write £50m of premium with expected claims of £20m and
expenses of £10m.
The company has carried out an investigation to assess the aggregate claims
distribution from the two classes.
(i) Describe the steps that the company may have taken in this investigation. [5]
The company concludes that claims in personal motor have a standard deviation of
£10m and claims in commercial property have a standard deviation of £30m. The
company also concludes that both the aggregate claim distributions can be modelled
using lognormal distributions. The company also makes the assumption that claims
from the two classes are statistically independent and that the total aggregate claim
distribution (i.e. the sum of claims from both classes) can be approximated by another
lognormal distribution.
(ii) Calculate the mean and standard deviation of the total or aggregate claim
distribution for the company. [2]
(iii) Calculate the mean and standard deviations of the underlying normal
distributions for each of the classes as well as the total claim distribution. [5]
The company wants to ensure that it has sufficient capital to pay its claims even if
they reached the 99.5th percentile of the claim distribution.
(iv) Calculate the 99.5th percentile for claims in each class as well as the total. [5]
(v) Discuss the reasons why the assumption of independence between the two
classes of business may not be appropriate. [5]
(vi) Calculate a reasonable range of minimum amounts of capital, based on

different correlation assumptions, that the company needs in order to satisfy
this 99.5% solvency level. Ignore taxes and investment income. [3]
[Total 25]
END OF PAPER
ST3 S2006 4
EXAMINATION
September 2006

EXAMINERS’ REPORT
Introduction
M Stocker
November 2006
Comments
1 Restriction on the types or amounts of assets that can be taken into account when
assessing solvency.
Prevent the company from holding certain assets.

Prescription to hold certain assets.
Custodianship of assets
Requirement to hold mis-matching reserves.
Prescribing asset valuation basis for assessing solvency.
Risk-based capital requirements varying by asset class
Tax incentives/disincentives
Requirement to hold assets in domestic currency.
Comments on question 1: A bookwork question generally reasonably well answered.
2 (a) Risk attaching basis: A basis under which reinsurance is provided for claims
arising from policies commencing during the period to which the reinsurance
relates.
The insurer knows there is coverage for the whole policy period when written.
Alternatives are “losses occurring” or “claims made”.
(b) First loss: A form of insurance cover in which the sum insured is less than the
full value of the insured property, so that the policyholder has to bear any loss
in excess of the sum insured. It is appropriate in circumstances where the
policyholder considers that a loss in excess of the sum insured is extremely
unlikely or the item is effectively priceless. Commonly used in fire
business/commercial property.
This approach is used to establish a more relevant figure for the sum insured
on which to base the policy coverage.
(c) Discovery period: A time limit, usually defined in the policy wording (e.g. in
the sunset clause) or through legislative precedent, placed on the period within
which claims must be reported.
This term is used in policy wording to provide certainty regarding the length
of time the (re)insurer is exposed to the risk of receiving claims — i.e. it limits
the (re)insurers exposure. It generally applies to classes of business where
several years may elapse between the occurrence of the event or the awareness
of the condition that may give rise to a claim and the reporting of the claim to
the insurer. e.g. employer’s liability or professional indemnity.
Comments on question 2: Many candidates failed to define some or all of these terms. This
was somewhat surprising as the answers for this question come directly from the glossary of
terms in the core reading. Candidates should learn the meaning of the key terms in general
insurance before attempting to sit the exam.
Page 2
3 (i) Average: This relates to the practice of scaling down the amount of a claim by
applying the ratio of the actual sum insured to the amount deemed to have
been the appropriate sum insured.
(ii) Usually used in household buildings/contents insurance.
For household buildings/contents insurance, the sum insured can only be

obtained accurately by expert valuation which is expensive. By allowing the
insured to select his own sum insured means that the cost of property valuation
is not required which saves the insured money: to the insured’s benefit.
To the insurer’s benefit, in the event of a claim, if the house is over-insured

the insurer will not pay more than the actual claim, if under-insured the insurer
pays the appropriate percentage.
Ensures that the policyholder only gets the cover they have paid for.
Discourages under-insurance.
(iii) The insurer may:
• pay out in full
• pay out in full, but deduct an amount for the additional premium that
would have been due if correctly insured
• pay out in full, but deduct an amount for the additional premiums that
would have been due over several past years if correctly insured
• negotiate an alternative settlement
• decline to pay out at all/void policy
• improve underwriting involvement to ensure correct SI/introduce

escalation clause
• use different exposure/rating factors to eliminate the potential for this to

occur
• provide better upfront education of policyholders with regard to

(under)insurance issues
Comments on question 3: Most candidates understood the meaning of the term average, and
could think of one or two alternatives approaches available to a general insurance company.
The better candidates were able to generate a range of alternative approaches. In part (ii)
few candidates attempted to answer the specific question on why average might be used.
Page 3
4 For a risk to be insurable:
The policyholder must have an interest in the risk being insured.

This distinguishes between a wager and insurance.
The risk must be of a financial and reasonably quantifiable nature.

Financial as claims settlement will be financial.
This means the insurer can determine how much to pay out in the event of a claim and
how much risk they are accepting.
Individual risk events should be independent of each other.

This is so that there is a spread of risk.
The incidence and/or magnitude of the insured event must be uncertain.
There should be a relatively low probability of the event occurring. Events that are
certain to happen cannot be insured as payout is certain, need to be able to spread risk.
There should be a large number of potentially similar risks which can be pooled in
order to reduce the variance and hence achieve more certainty.
There should be an ultimate limit on the liability undertaken by the insurer.

This is so the insurer knows what they are liable to pay out in claims.
Moral hazards should be eliminated as far as possible.

These are difficult to quantify, result in selection against the insurer and lead to
unfairness in treatment between one policyholder and another.
There should be sufficient existing data/information to enable insurer to estimate

extent of risk and likelihood of occurrence.
This is so that the insurer can come up with a price.
Comments on question 4: Bookwork: well answered by most candidates.
5 (i) Experience rating systems are systems by which the premium for an individual
risk takes into account the claims experience of that individual risk.
Some systems relate to numbers of claims, e.g. NCD schemes,

Some systems relate to amounts as usually used for fleet rating purposes
Some systems are retrospective.

Some systems are prospective.
Benefits:
Small claims may be discouraged (benefits Insurer).

Gives the policyholder an incentive to be cautious and take precautions to
avoid claims (benefits Insurer and Insured).
Page 4
Rewards good claims experience and penalises bad claims experience

(benefits Insurer and Insured).
Insurer must experience rate if this is standard market practice to avoid being
selected against.
Premium charged should be more representative of the risk of the policyholder

having a claim (more so by amount rather than by number) (benefits Insurer
and Insured).
May encourage customer loyalty.
Insured could be bad risk and be lucky (benefits Insured).
Reduces administration for experience rating by amount (benefits Insurer and

Insured).
Can increase insurance companies’ control of underwriting for large risks

(benefits Insurer).
Drawbacks:
Operates against the insurance principle of spreading costs (drawback

Insured).
Can make policyholders feel aggrieved if after many years claim free driving
they have a claim and are penalised for it, e.g. in NCD systems (drawback
Insurer and Insured).
Can make policyholders feel aggrieved if they lose discount through a claim
that was not their fault.
Penalty for having a claim may be a large increase in premium, e.g. loss of
NCD (drawback Insured).
Increased administrative costs for rating by individual claims (drawback

Insurer).
Limited ability to distinguish between high and low risks if on average one
claim is made by a policyholder once every five years (drawback Insurer).
Does not achieve objective of rewarding better risks as the reward is given for
not making a claim and not for being accident free (drawback Insured for
NCD systems).
Page 5
(ii) The statistics the insurance company would ideally require to analyse initially should
be sub-divided by:
regional area and by intercity services separately

data over the past ten years would be ideal
with data further sub divided by type of bus
Particular consideration for the following is required in respect of each

exposure period:
The number of miles travelled.
Number of passengers per year.
Number of vehicle years.
Similar information (expected mileage, vehicles, number of passengers etc) is

required for the future policy period.
Claims Data
Historical claims information is required.
A one to one link is required between statistics on exposure and statistics on
numbers of claims.
Strict definition of year of claim required, by date of accident or notification.
The following statistics are required:

Data are required at individual claim level
Type
Paid
Outstanding
Historical claim development data required:

Dates of reporting / settlement
Changes in case estimates over time
Additional information required is:

The method of claim recording.
The method of claim handling
Age of bus fleet.
Speed of renewal of bus fleet.
Experience of drivers increasing or decreasing.
Information on drivers (previous driving experience, endorsements,
convictions).
Turnover of staff.
One person operated buses or driver plus conductor.

Changes in bus routes over the past year/planned changes for the future
Additional safety features in buses, if any, incorporated over the past years.
Terms of insurance required, limits, excesses, etc
Additional information used to assess:

Any adverse/beneficial trends that are likely to develop in the future that will
tend to increase/decrease the number of claims and/or the cost of claims
Page 6
Reasons for any change in the experience over the past few years that can be
detected in the statistics.
Comments on question 5: Weaker candidates produced generic lists of statistics that could
be used in pricing a motor risk. These lists generally failed to consider the particular features
of the risk in question, for example, that the bus company has been self insuring for some
time. As a result, some candidates listed statistics that were irrelevant to the particular risk,
or missed out useful statistics. In addition to tailoring their answers to the risk described in
the question, the best candidates clearly structured their answers. For example, strong
answers separately described claims and exposure statistics, and the need for
correspondence between the two.
6 (i) How much data

… needs to balance what’s essential to avoid discouraging potential
policyholders and cover likely future data requirements.
IT issues – how much storage capacity will be required

System structure should be flexible enough to permit future changes
Need to design secure, logical and efficient IT systems
… to provide reassurance to potential policyholders and encouraging new
business.
Aim to have advanced internal cross-checks

… to ensure majority of potential inaccurate info is corrected at source.
Aim to ensure info is entered in a consistent manner

… to ensure the info captured can be analysed meaningfully.
Need to design exception reports to investigate potential oddities.
How will the data be captured and/or verified.

e.g. Proof of possession, Claims History, access to car registration database,
proof of entitlement to NCD.
What information will other users require

e.g. reinsurers, regulators, auditors.
and the uses to which the data will be put
e.g. reserving, rating, marketing, etc.
Need to ensure backup arrangements exist …

… and disaster plans.
Aim to ensure designed for ease of data retrieval after input

… to ensure ease of reporting and analysis.
Data security issues.

Data Protection Act issues.
Payment security issues.
Page 7
(ii) May charge incorrect premium rates

… which may result in larger than planned volumes
… at loss making rates / worse than expected
… or smaller than planned volumes
… leading to insufficient coverage of fixed / start-up costs
Or have incorrect rating structure

… resulting in adverse selection.
May result in early solvency pressure

… at a time when capital is most scarce.
May result in loss of regulator confidence

… leading to subsequent business restrictions / closure.
May result in loss of market confidence

… with longer-term business retention implications.
Inappropriate reinsurance purchased.

May result in loss of reinsurer confidence.
Comments on question 6: Candidates that scored poorly on part (i) often failed to read the
question carefully enough, and as a result produced lists of rating factors for motor
insurance. Also candidates often did not generate the breadth of points, getting bogged down
in the detail of just one area. In part (ii), most candidates correctly identified that use of
inaccurate data could lead to incorrect premium rates. Better candidates went on to identify
the consequences of mis-pricing with regard to capital, regulation, reinsurance and market
confidence.
7 (i) To avoid distortions in any statistical analysis it is important to ensure that all
information is converted to a consistent basis for consideration.
Changes in cover
… these will affect the relative exposure levels between different cohorts and
hence the likely claims trends.
Inflation … differences in both past and future levels

… may affect the analyses if not stable through time.
External changes affecting exposure.

e.g. legislative or building regulation changes
… may affect both the likelihood and severity of claims.
Change in geographic spread or premium rates

… may affect the mix of business.
Large one-off claims

… where these may distort the analysis.
Page 8
Significant / Catastrophe event.

e.g. Weather related / Subsidence
… where these may distort the analysis.
There are also many issues that may affect the claims runoff and distort any
trend.
e.g. reporting delays due to internal or external issues
or mass sickness or postal strikes.
Handling / settlement delays.

Internal system changes.
Seasonality / Regional bias.
Changes in claim settlement procedures.
Split data into homogenous groups, by class or claim type.
Data cleaning/removal of errors.
Reinsurance if net paid data is being projected.
Consistent treatment of claims handling expenses.
Tail factor.
(ii) Different basis of reserving, e.g. loss adjusters may have under/over reserved,
actuary may have included margins.
Loss adjusters may each have different views on costing and inflation issues
… both of these can be investigated by means of an audit / review process
… with particular focus on the largest case estimates.
Loss adjusters have more information upon which to base their estimates.
… Comparison of the largest differences can be investigated at cell level to

see if there are any particular isolated large discrepancies.
Statistical estimates may include loss adjustment expenses

…example of investigation for loss adjustment expenses
Statistical estimates may include IBNR/unexpired risk provision

… example of investigation for IBNR/unexpired risk provisions
Statistical estimates may include reinsurance recoveries

Statistical estimates may have been discounted
There may be errors in either or both statistical and case estimates

… which can be investigated by rigorous review of both sets of estimates.
IBNR may be negative due to computer generation of too many case estimates
for small claims, which later become redundant on the system.
Case estimates may be out-of-date

… which may be highlighted through investigating the dates on which the case
estimates were established.
Page 9
There may be currency errors

… which may be highlighted through exception reporting.
Statistical adjustments are approximated across groups of claims rather than

assessed at individual claim level.
To investigate any inaccuracies arising from this, statistical estimates can be

calculated on several different bases using several different methods
… to ensure internal consistency / identify any flaws
… ensure the estimates are robust and not affect significantly by “rogue” cells
… to help understand any oddities in the results.
Statistical estimates may be more accurate if, for example, they were based on
incurred claims rather than paid claims, or premiums were used.
Investigate the accuracy of data used in the statistical analyses.
There may have been changes in loss adjustment procedures.
Investigate whether loss adjusters look at all claims or whether case

outstandings on small claims are set using a formulaic approach. Check
reliability of formula based approach.
Comments on question 7: Part (i) was reasonably well answered with many candidates
identifying a range of issues that could require adjustment. Many candidates found part (ii)
more difficult than part (i) even though this is a question that has been asked a number of
times in the past. Stronger candidates appreciated that both statistical methods and case
estimates are based on judgemental assumptions, and considered the possibility of different
views being taken on these assumptions. Some candidates did not describe the investigations
at all and therefore scored poorly. Generating a breadth of reasons for the difference was
again a problem for many candidates.
8 (i) Collect claims data over an appropriate historical time period.
The time period chosen should be long enough to have credible data and it
should be recent.
Subdivide the data into homogenous groupings (e.g. by peril or exposure

period) taking care to ensure credibility within the cells.
Adjust data to allow for:
Claims inflation over the historical period to now.
Changes in exposure over the period.
Changes in policy conditions, limits and deductibles.
Changes in legal / regulatory conditions.
Page 10
Projection to ultimate.
It may be advisable to allow for large/catastrophe losses separately
Fit a range of probability distributions to the data within each grouping.
Find separate loss distributions for frequency and severity of loss within the
homogenous groupings.
Use of binomial, Poisson, negative binomial for frequency, whole numbers or

normal or translated gamma for severity.
Select appropriate parameters for the fitted distributions.

Combine the loss distributions for separate perils by closed form or simulation
methods to obtain an overall aggregate claim distribution.
The company might also have fitted a distribution (e.g. Lognormal, Pareto,
Weibull) to the aggregate claims directly.
Test goodness of fit.
(ii) μ A+ B = μ A + μ B = 30 + 20 = £50m
σ2A+ B = σ2A + σ 2B = 102 + 302 = 100 + 900 = 1000
Standard Deviation of A + B = 1000 = £31.62m
Mean of a lognormal distribution = eμ+0.5σ

2
(iii)
And variance = e2μ+σ eσ − 1

2
( 2
)
Std Deviation = eμ+1 2σ2
eσ − 1
2
eσ − 1
2
Std Deviation = Mean .
eσ − 1 = Coefficient of variation = CV
2
Std Deviation / Mean =
(
=> σ = log e CV 2 + 1 )
and μ = loge(Mean) – 0.5σ2
where μ and σ are the mean and standard deviation of the underlying normal
distribution.
Page 11
Class A B Total
Mean = m 30.00 20.00 50.00
Stdev = s 10.00 30.00 31.62
CV = s/m 0.33 1.50 0.63
(
σ = log e CV 2 + 1 ) 0.32 1.09 0.58
μ = loge (Mean) - 0.5σ 2
3.35 2.41 3.74
If working in £, mean of A is 17.16, mean of B is 16.22 and mean of total is 17.55.
(iv) 99.5th percentile for a standard normal distribution is at 2.58.
So the 99.5th percentile is given by μ + 2.58σ
Class A B Total
(
σ = log e CV 2 + 1 ) 0.32 1.09 0.58
μ = LN(Mean) – 0.5σ2 3.35 2.41 3.74
99.5th percentile 4.19 5.21 5.24
99.5th Percentile claim 65.76 182.62 188.74
Alternative answers were allowed based on alternative number of decimal

places used for the 99.5th percentile of a standard normal distribution
(v) Claims experience between the two classes is likely to be correlated due to:
Both classes are likely to be affected by underlying economic conditions. For

example, at time of low economic growth there tend to be more fraudulent
claims, and theft claims.
Events such as windstorms, earthquakes, flooding tend to cause losses in both

classes of business.
Both classes would be subject to bigger losses in inflationary environments.
Legal changes may affect claims severity and frequency in both classes
e.g. changes in court procedures, statute of limitations, and other legislation
could affect both classes.
Similar claim handling procedures.
Higher crime rates would lead to more theft claims from both classes.
Page 12
(vi) If we assume independence between the two classes of business, then loss
given that losses are at 99.5th percentile = 100 – 25 – 188.74 = -113.74.
If we assume that the two classes are 100% correlated, then the loss given
losses at 99.5th percentile = 100 – 25 – 65.76 – 182.62 = -173.38.
Negative correlation is not likely.
Adjust for premium received (100)

Adjust for expenses (25)
Therefore capital required is between £113.74m and £173.38m.
Comments on question 8: Part (i) was generally well answered with better answers
discussing the aggregate distribution, poorer answers trotting out the standard bookwork.
Part (ii) was simple but not universally well answered. Parts (iii) and (iv) were not difficult
but many struggled or did not attempt. Part (v) was often poorly answered even though
completely independent of the calculations. A few candidates actually suggested that the
results of the calculations in previous parts of the question, which were based on assuming
independence, indicated that the classes were not independent without spotting the flaw in
this logic. Part (vi) was only answered reasonably by a small minority. Many candidates who
correctly identified the method required for the question went on to lose marks due to
calculation errors suggesting that they had not left sufficient time to check their answers.
Page 13
EXAMINATION
16 April 2007 (pm)

booklet.
supervisor.
question paper.
ST3 A2007 © Institute of Actuaries
1 You are an actuary working for a general insurance company with a small book
offering professional indemnity cover to solicitors through brokers. Your CEO has
noted that a competitor has recently been successful in growing its book by selling
professional indemnity policies direct to small firms of solicitors. He suggests that
you could increase your book with a major campaign marketing professional
indemnity cover direct to other professions.
(i) Outline the risks to the company of pursuing such a marketing campaign. [7]
(ii) Suggest how the risks identified in (i) could be mitigated. [5]
[Total 12]
2 A small general insurance company specialises in motor insurance.
(i) Outline the principles that you would consider in deciding in which assets the
company should invest. [9]
(ii) Explain how these principles differ from those of a large general insurance
company specialising in employers’ liability. [3]
[Total 12]
3 You are the actuary of a large general insurance company. You have been asked to
price a “cross-class” deal for a customer. The policy will cover the customer’s motor
fleet and public liability requirements. Another general insurance company has
written the public liability cover in the past.
The proposed structure for the policy is as follows:
Motor: the general insurance company will provide unlimited cover for any
individual loss.
Public liability: the limit of indemnity on any one individual loss is £250m.
The customer retains a deductible of £0.5m on each and every loss for the
complete programme subject to an annual aggregate deductible of £15m.
(i) Outline the concerns you would have with this proposed structure. [3]
The customer has provided you with a large database of their individual claims data,
as well as relevant exposure measures, for the past 10 years.
(ii) Explain how you would calculate a risk premium for this product using the
information on this database. [10]
[Total 13]
ST3 A2007—2
4 A motor underwriter has approached you for assistance with a new business premium
quote on a fleet of 100 heavy goods vehicles commencing 1 January 2008. She has
supplied you with the following unprojected historical claims data from the existing
insurer as at 31 October 2007.
Incurred Claim amounts in £000’s
Accident Own Third Party Third Party Earned

Year Damage Damage Personal Vehicle
Incurred Incurred Injury Years
Costs Costs Incurred Costs
2003 44 30 55 80
2004 56 32 61 88
2005 42 35 51 90
2006 70 50 35 92
2007 40 30 20 98
The following additional information is available:
The prospective insured has always renewed the policy on 1 January each year.
Damage inflation has been 4% p.a. for many years.
Personal Injury inflation has been 7% p.a. in each of the calendar years 2003 to 2005,
then 9% p.a. from calendar year 2006.
Incurred Claims as a percentage of Annual Ultimate Projected Claims are estimated

from internal data to be:
As at development month
10 22 34 46 58
Own Damage 70% 95% 105% 102% 100%

Third Party 45% 80% 95% 100% 100%
Damage
Third Party 30% 55% 75% 85% 95%
Injury
Commission is 15%. Expenses are £100 per policy, £10 per vehicle and 7% of claims
costs. Insurance Premium Tax can be ignored.
Profit and contingency loading is 5% of the overall gross written premium.
(i) Estimate the annual premium to charge the prospective client, using the data
provided, stating any assumptions you make. [11]
You have predicted that the average annual premium charged per vehicle during
January 2008 on your company’s existing account of 25,000 heavy goods vehicles
will be £3,750. You decide to recalculate the premium using a credibility approach.
ST3 A2007—3 PLEASE TURN OVER

(ii) Recalculate the annual premium assuming you use a credibility factor for the
fleet’s own experience as:
Z = minimum (1, 1 - σ/μ)
where
σ = the standard deviation of the yearly projected burning cost per vehicle
observed from the five year data
μ = the average of the yearly projected burning cost per vehicle observed from
the five year data [4]
(iii) Explain why the premium charged in practice may not equal the premiums
calculated in parts (i) or (ii). [5]
[Total 20]
5 You are the actuary of a large general insurance company that only sells insurance to
large international companies. The underwriters are considering entering the smaller
end of the commercial market through the creation of a new product that covers the
insurance needs of construction and engineering tradesmen who are either sole
traders, partnerships or limited companies with up to five employees.
(i) Describe the distribution channels through which this new product could be
sold. [8]
(ii) Compare the marketing methods in part (i) to those which would be used for
the insurer’s existing business. [2]
(iii) Describe the types of commercial insurance that these tradesmen may wish to
purchase. [10]
[Total 20]
ST3 A2007—4
6 You have been provided with the following financial information for general
insurance companies X, Y and Z for the accounting year 2006. All amounts shown
are in $millions.
Company
X Y Z
Gross Written Premium 50 2,000 250

Additional Unexpired Risk Reserve c/f 15 100 0
Gross Outstanding Claims Reserve b/f 20 800 750
Gross Claims Paid 35 700 150
Gross Outstanding Claims Reserve c/f 30 850 700
Non Acquisition Expenses 5 250 30
Investment Income 3 100 16
Current Assets at 31/12/2006 5 80 30
Current Liabilities at 31/12/2006 11 100 40
Investments at year end 125 3,500 1,000
Share capital at 31/12/2006 15 500 125
Acquisition Costs as a % of Gross 30% 15% 20%
Written Premium
(i) Construct the balance sheet for each of the three companies as at
31 December 2006, stating any assumptions made. [5]
(ii) Derive underwriting, solvency and return on capital employed ratios for all
three companies, stating any assumptions made. Taxation should be ignored.
[9]
(iii) Comment on the results in part (ii). [4]
(iv) Define five other insurance related ratios that could be derived from the data to
compare the performance of the three companies and state each of their
objectives. [5]
[Total 23]
END OF PAPER
ST3 A2007—5
EXAMINATION
April 2007

EXAMINERS’ REPORT
Introduction
M A Stocker
June 2007
Comments
Subject ST3 (General Insurance Specialist Technical) — April 2007 — Examiners’ Report
1 (i)
• Biggest danger is charging insufficient premium — this is a new
distribution channel
− exacerbated by lack of data since only have a small existing book
− which might not cover the relevant professions
• Lack of expertise in the professions CEO may be interested in
• May be difficult to attract staff (e.g. underwriters) with enough experience
of PI
• Inappropriate rating structure that could lead to anti-selection
• Could annoy brokers of existing PI book — could lose this business
• Risk of loss of broker value (e.g. collating detailed underwriting and risk
information from clients, administering policies and claims)
• Significant set up costs for direct operation — might not be recouped if
insufficient volume of business
− This could be more likely as this is a “new” channel, and other
professions may prefer to place the cover through a broker
− The set up costs/ongoing expenses may be much higher than
anticipated especially if we do not have experience in selling direct
• There is a danger that a competitor reacts e.g. by a similar campaign
leading to lower volumes than anticipated
• Exposure to large losses
• Greater exposure to accumulation event as write more business and more
professions, but may be more than offset by diversification benefit
• Reinsurance risk — insufficient or unavailable coverage
• Regulation risk — e.g. channel authorisation
• May sell bigger volume than expected leading to too much risk/increased
capital requirement
• Reduced persistency (e.g. lower retention)
• Operational risk of setting the whole arrangement up
(ii)
• Problem of lack of data might be mitigated by using reinsurer data if using
reinsurance
• Industry data might also be useful if available
• Reinsurers could also provide expert advice in this field on rating and
underwriting
• Alternatively recruit more experienced staff
• Reinsurers provide assistance in mitigating the risk of large single losses
and accumulations (e.g. through per risk xl cover and cat cover)
• Use financial reinsurance or raise more capital to support balance sheet
• Difficult to mitigate the risk of annoying brokers. It may help to “sound
them out” beforehand or allow them to sell the new product as well
• Difficult to predict competitor action but company should monitor
competitors pricing levels
• Danger of low volume mitigated by market survey
• Full costing with experts on similar projects would help to assist in
correctly estimating expenses (or similar sensible comments on controlling
expenses)
Page 2
• Regular monitoring should ensure that volume controlled and addressed by

regular rating action
• Check with regulators that this class and marketing method are allowable
• Research reinsurance market to ensure sufficient coverage available

• Check that relevant IT systems, human resource and facilities are capable
of coping with new operation
• Carrying out pilots with “friendly” brokers will help identify issues early
Comments on question 1: Part (i) was generally well answered by most candidates. For
part (ii) not all candidates ensured that all of the points raised in part (i) were subsequently
addressed. Candidates who identified the financial, operational and reputation risks scored
particularly well in this question.
2 (i)
• Try to maximise investment return subject to meeting liabilities with
chosen level of certainty
• Match assets and liabilities by
− term
− amount
− nature
− currency
• Motor property damage claims are mainly short tailed, so need liquid
assets
− need to hold cash on deposit, very short dated assets such as short
dated government securities to match liability outgo
• Motor third party claims are longer tailed and costs are influenced by
inflation
− need to hold some longer dated real assets (index linked securities if
available or low risk equities)
• Consider regulatory requirements :
− restrictions on assets that can be held
− prescription to hold assets
− custodianship of assets
− mismatching allowed
• Since company is small, need to have extra consideration of the level of
uncertainty in reserves, so more secure, liquid assets required
• A small company might consider collective investment vehicles (e.g. unit
trusts, investment company shares)
• Investment likely to be in assets of small unit size (e.g. no direct property
investment
• Level of investment expenses of each asset type
• Tax efficiency of each asset type
• Availability of certain asset types
• Benchmarking against competition
• Availability of additional capital (e.g. parent company, shareholders)
• Diversification of assets held (within and between asset types)
Page 3
• Size of the free reserves (in excess of solvency requirements)

− As the company is small, the company is less likely to be able to
accept the risk of investing in higher risk/reward investments (e.g.
property)
• Expected growth plans and resultant needs to invest in the business
• Shareholders and management’s attitude to risk
(ii)
• Since company is large, assuming larger free assets, potential scope for
more aggressive investment strategy.
• Employers’ liability claims are generally longer tailed and costs are
influenced by inflation
− Greater need to hold longer dated assets providing real returns
− since better match by term for liabilities
• Equities and properties are an appropriate match
• Index-linked bonds (if available) for security and inflation hedge
• Potential investment in specialist areas such as large unit size, ventures,
brokers, derivatives
• Likely to handle investments in-house through specialist team of managers
giving greater control over investment choice
Comments on question 2: Most candidates scored well on this bookwork question. A few
candidates suggested investments in part (i) that were unlikely for a small company (e.g.
direct property).
3 (i)
• Once the £15m aggregate is exhausted and/or as there is no aggregate limit
on public liability, there is the potential for a single bad year
• Unlimited coverage for motor — potential for large single loss
• Large limit for public liability — potential for large single loss
• Do we have the required reinsurance coverage to protect the insurer
against these ?
• Need to clarify if the excesses/limits cover legal and other expenses
(ii)
• Model the motor and public liability accounts separately
− Group claims by property damage and bodily injury per cover
− Need to model the frequency and severity separately in order to apply
deductible
− Use client’s data as start point (since large dataset)
− Pick a base period
− Use from the ground up data
− Adjust for IBNR and IBNER
− Adjust the claims for inflation
− Adjust for change in exposure
− Adjust for trends in data
− Adjust for changes in underwriting and claims handling procedures
Page 4
− Adjust for any changes in terms and conditions over period considered
− Compare outcome with any internal portfolio/external benchmark data

- especially for large loss assumptions
- consider credibility weighting to portfolio/benchmark
• Consider any relationship between claims received under motor and public
liability covers to determine any correlation in experience
− Unlikely to be strong so probably model as independent.
• Could use deterministic modelling approach to determine parameter
estimates for frequency and severity for each cover
• Determine the mean values for both parameters
• Alternatively could model the outcome of the individual accounts using
stochastic modelling approach
• Carry out several thousand simulations and apply the product “rules” to
the outcome
• The average outcome to the insurer in the simulations will give the
expected loss cost to the insurer
• This would also provide the range of possible claims experience scenarios
which could assist in determining suitable reinsurance arrangements
Comments on question 3: This question was poorly answered by most candidates. Some
candidates were of the mistaken opinion that unlimited liability for motor is abnormal.
Many candidates suggested use of Generalised Linear Modelling techniques involving the
use of rating factors when this was clearly inappropriate for a policy for which all losses are
large losses. Furthermore, many candidates decided to cut and spread large losses when the
purpose of the cover was to protect the insured against large losses. This requires an
analysis of the claims size distribution to determine the impact of the deductible and
aggregate deductible on the expected claim frequency and severity. Very few candidates
explored benchmarking claims experience against internal/external claims data nor the use
of deterministic and stochastic methods in pricing this policy, thus missing out on a
significant number of the available marks.
Page 5
4 (i)
Assumptions
Assume claims inflation in 2008 = claims inflation in 2007

Development factors are on the same basis as claims stats
Projected Claims Costs (before claim inflation allowance)
Year Damage Costs Third Party

Personal
Injury Costs
2003 44 + 30 = 74 58
2004 55 + 32 = 87 72
2005 40 + 37 = 77 68
2006 74 + 62 = 136 64
2007 57 + 67 = 124 67
Amounts in £000
Claim inflation adjustments to 2008

Personal
Injury Costs
2003 1.045 = 1.217 1.072 ×1.093 = 1.483
2004 1.044 = 1.170 1.07 × 1.093 = 1.386
2005 1.043 = 1.125 1.093 = 1.295
2006 1.042 = 1.082 1.092 = 1.188
2007 1.04 1.09
Projected Claims Costs (after claim inflation allowance)

Personal
Injury Costs
2003 90 86
2004 102 99
2005 86 88
2006 147 76
2007 129 73
Total 554 422
Amounts in £000
Burning Cost per vehicle = (554 + 422) × 1000 / (80 + 88 + 90 + 92 + 98)

= £2,178
Page 6
Gross Premium to charge client = (number of vehicles in 2007 × burning cost

per vehicle × claims costs + policy expenses + vehicle expenses) /
(1 – commission rate – profit and contingency loadings)
= (100 × 2178 × 1.07 + 100 + 10 × 100) / 1 - 0.15 – 0.05)
= £292,683
Also identifying and allowing for any legitimate trends in the data, e.g.
improvement in PI peril
Alternative acceptable approaches scored equivalent marks
(ii) Yearly burning cost observations
2003: 175,872/80 = 2,198

2004: 201,106/88 = 2,285
2005: 174,499/90 = 1,939
2006: 222,903/92 = 2,423
2007: 201,429/98 = 2,055
μ = (2198+2285 + 1939 + 2423 + 2055)/5 = 2,180

σ = Square root of {(5 × (21982 + 22852 + …) – (2198+2285 + …)2)/(5 × 4)}
= 190
Therefore Z = min (1, 1 – 190/2180)

= min (1, 1 - 0.087)
= 0.913
Therefore revised gross premium = Z × 292,683 + (1 - Z) × (100 × 3750)

= £298,932
Note alternative approach: one could strip out the claims cost from the average
premium and then blend claims costs and reconstruct gross premium from that
(iii)
• The 5 year historical claims experience may be heavier or lighter than is
expected in 2008
• Potential large losses in historical data distorting the calculations
• Competitors pricing levels
• Insurer may be willing to take a reduced profit or slight loss on this
business as the policyholder has other insurance contracts with the
company that are highly profitable.
• Using the insurer’s own heavy goods vehicles experience may be
inappropriate, for example the account may have a different business mix
to that of the client (e.g. age of drivers, location of vehicles).
• Cover provided in 2008 differs from that in previous years (e.g. increased
own damage excess)
• Different policy wordings/restrictions expected to reduce claims
costs/numbers
Page 7
• Expected future external events (e.g. changes in legislation) that may

impact claims costs, expenses, commission or profit allowances
• Per policy expense allowance in main account may be disproportionately
higher than that required under this fleet contract
• Influence of broker/customer (e.g. volume of other business offered by
broker/customer)
• Regulators may restrict the price that could be charged
• Other assumptions not used in the calculation (e.g. investment return,
reinsurance)
• Management decision/growth plans/attitude to risk
• Position in the market cycle
Comments on question 4: In part (i), the examiners’ solutions provide one answer although
alternative solutions were given equivalent marks (e.g. taking an average of each year’s
burning cost per vehicle). Some candidates applied the profit and contingency loading to the
risk premium and not the office premium as stated in the question. Parts (i) and (ii) were
fairly numerical questions and these parts were generally well answered by most candidates.
However, some confusion arose over the time period to use for the inflation figures. Some
candidates also failed to realise that, as the incurred claims development data were as at 10,
22, … months there was no need for further adjustments to allow for the data provided
being as at 31 October 2007. For parts (ii) and (iii) the better candidates identified the
external factors such as market, distribution and regulatory influences on the price. For part
(iii) most candidates failed to generate enough ideas.
5 (i) Brokers
• A company which acts as an intermediary between the seller and the buyer
of the insurance product without being tied to either party
Banks, Building Societies and other financial institutions
• A company whose main activities include providing financing to small

businesses and can therefore cross-sell insurance on the back of loan
arrangements
Trade Associations
• A union whose main activity is to provide support and advice to

companies of a similar trade who can provide insurance products tailor-
made to their requirements
Internet
• The insurance company can develop a web-based sales point with the
customer entering all the relevant rating information through the internet to
obtain a quote for insurance
Page 8
Telesales
• A call centre arrangement managed by the insurance company to provide

in-calls and out-calls to potential clients
• In-calls can be through advertising in press or telephone directories
• Out-calls can be through leads generated from commercial tradesmen
databases
Direct mailshot
• The insurance company can directly target potential clients through the
posting of literature to small business tradesmen
Employed staff paid by salary or commission
• Staff of the insurance company visit the potential clients face to face to
discuss their insurance requirements based on their circumstances.
Affinity Groups (e.g. Trade Retailers, Training Groups)
• A company whose main activities are non-insurance related (e.g. a

building supplies wholesaler) but whose organisation has a significant
Commercial customer database to target sales.
(ii)
• Companies of all sizes (small and large) may use Commercial brokers as
they can offer advice on their specific insurance needs.
• Larger international companies with credible data attract individual

underwriting and brokers facilitate this, whereas a standard rating structure
approach is used for smaller risks
• Companies of all sizes could be a part of a trade association
• The remaining distribution methods are more likely to be used mainly by

small businesses due to:
− the relative speed and ease of obtaining low cost insurance

− the far greater propensity for clients to use the distributor for other
non-insurance activities
− the commodity nature of small business insurance products makes
them more appropriate for direct route
− cost considerations : lower unit delivery cost of internet when
compared to brokers is appropriate for low average premiums
(iii) Public Liability
• The insured is indemnified against legal liability for the death or bodily
injury to a third party
Page 9
• Or for property damage belonging to a third party

• Other than those liabilities covered by other liability insurance
Employers Liability
• The insured is indemnified against legal liability to compensate an

employee or temporary employee for the death, disease or bodily injury
suffered owing to the negligence of the employer during the course of
employment.
Contract Works
• Indemnifies insured against loss of or damage to contract works property

being worked on and materials
Plant insurance (Hired or Own Plant)
• Indemnifies insured against loss or damage to plant whether it is hired or

owned by the insured
Employees Tools All Risks
• Indemnifies insured against loss or damage to tools used in the course of

trade
Personal Accident/Sickness
• Indemnifies all people specified under the cover for loss of earnings in an
event of an injury or accident, whether temporarily or permanently out of
work.
Professional Indemnity
• Indemnifies insured against legal liability resulting from negligence in the

provision of a service (e.g. inaccuracies in architectural building design)
Vehicle insurance (vans, pickups, goods vehicles, trucks, lorries)
• Property Damage — indemnifies insured against loss or damage to their

own vehicles
• Third Party Liability — indemnifies insured against compensation payable
to third parties for damage to their vehicle or property or for personal
injury
Acceptable alternative valid covers :
Commercial Fire/Business Interruption/Offices

Fidelity Guarantee/Theft by Employees
Pecuniary Loss/Credit Guarantee/Third Party Failure
Page 10
Comments on question 5: Part (i) was well answered by the majority of candidates. A few
candidates listed distribution channels but failed to describe them as the question asked.
Candidates failed to generate many points for part (ii) and thus missed many points on the
marking schedule. The better candidates identified the differences between the commodity
nature of selling insurance to small businesses and the more bespoke underwriting
requirements of selling insurance to large international businesses. In part (iii), candidates
who scored well identified the specific property and casualty insurance risks of small
construction and engineering businesses.
6 (i) Assumptions
All yearly business
No reinsurance
Risks written uniformly across year
Risk is uniform across policy year
Company
X Y Z
Assets
Total investments 125 3500 1000

Current Assets 5 80 30
Deferred 8 150 25
Acquisition Costs
Total Assets 138 3730 1055
Liabilities
O/S claims 30 850 700

reserves
Additional URR 15 100 0
UPR 25 1000 125
Current Liabilities 11 100 40
Free Reserves 57 1680 190
Total Liabilities 138 3730 1055
(ii)
Assumptions
• assume GWP = GEP (i.e. business written in 2005 = business written in

2006)
• assume AURR as at 31/12/2006 = AURR as at 31/12/2005
• assume outstanding claims reserves include IBNR
Loss Ratio = claims incurred/GEP
Company X = (35 + 30 - 20)/50 = 90%
Page 11
Company Y = (700 + 850 - 800)/2000 = 37.5%

Company Z = (150 + 700 - 750)/250 = 40%
Expense Ratio = Acquisition Expense Ratio + Non Acquisition Expenses/GWP
Company X = 30% + 5/50 = 40%

Company Y = 15% + 250/2000 = 27.5%
Company Z = 20% + 30/250 = 32%
Underwriting Ratio = Loss Ratio + Expense Ratio
Company X = 90% + 40% = 130%

Company Y = 37.5% + 27.5% = 65%
Company Z = 40% + 32% = 72%
For solvency ratio:
Solvency Ratio = Free Reserves/GWP
Company X = 57/50 = 114%

Company Y = 1680/2000 = 84%
Company Z = 190/250 = 76%
For return on capital employed:
Return on Capital employed = (Earned Premium – Claims Incurred –

Expenses + Investment Income) / Free Reserves
Company X = (50 - (35 + 30 - 20) - 5 – 15 + 3))/57 = -21%

Company Y = (2000 – (700 + 850 - 800) - 250 – 300 + 100))/1680 = 48%
Company Z = (250 – (150 + 700 - 750) – 30 – 50 + 16)/190 = 45%
(iii) Comments
• Company X may have suffered from adverse claims experience due to its
higher loss ratio compared to the other companies.
• Each company may be writing different classes or mix of business, each at
a different point in their respective market cycle
• Company X expense ratio is higher due to higher acquisition expense ratio.
• The company is smaller than Y and Z and it may be spending money to

expand rapidly.
• Company X solvency ratio is higher than the other companies.
• This may be the result of a recent capital injection to expand the business.
• Company Z has the lowest solvency ratio, suggesting that the company is
less financed than the other companies.
• Or it may have more stronger valuation basis for its assets and liabilities
Page 12
• Company Z return on capital employed is the highest, supported by a

larger relative investment return compared to the other companies
• Company Y and Z both have high returns on capital employed, supported
by a good underwriting results
• Relevant comment comparing profitability and solvency
(iv)
• Investment Return = Investment Income/(Current Assets + Investments)
• This provides a comparison of the investment performance of the
companies.
• Gross Claims Paid/Gross Outstanding Reserves

• This provides a comparison of the relative speed at which reserves are
reduced by claims payments
• Gross Outstanding Reserves/Gross Written Premium

• This provides a comparison of the relative strength of the outstanding
reserves
• Additional Unexpired Risk Reserve cfwd / UPR cfwd

• This provides a comparison of the relative profitability of the unexpired
risk
• Current Assets / Current Liabilities

• This provides a comparison of the ability of each company to meet short
term liabilities without the need to realise investments
Acceptable alternative valid ratios:
• Loss Ratio
• Expense Ratio
• Profit Margin
• Total Assets/Total Liabilities
Comments on question 6: The examiners were disappointed by the standard of answers to

this question. In part (i) a number of candidates made assumptions that were inconsistent
with the data provided in the question (e.g. assuming “all policies were written on 1 January
each year with no UPR at year end” even though an AURR was being carried forward at
year end). Some candidates thought that share capital was an asset and failed to understand
that this formed part of the free reserves of each company. In part (ii), a number of
candidates did not know the definition of underwriting ratio with some incorrectly assuming
it was derived from the underwriting profit. In part (iii), the better candidates commented on
possible reasons for the level of the ratios in part (ii) and compared profitability and
solvency levels for each company separately. Most candidates merely commented on whether
the ratios were high/low for each company without providing any explanation as to what may
be driving the results. In part (iv), a number of candidates defined ratios that could not have
been derived from the data as specified in the question (e.g. % reinsured).
Page 13
EXAMINATION

booklet.
supervisor.
question paper.
ST3 S2007 © Institute of Actuaries
1 A small general insurance company writes only a specialised book of business. The
aggregate claims distribution is estimated to follow a compound Poisson distribution
where λ = 10% and the claim size distribution is exponential with mean of £0.5m.
The company starts with capital of £10m.
(i) Show that the loss ratio that the general insurance company must achieve in
order to ensure that its probability of ruin is less than 0.5% is 73.5%. Ignore
expenses for the purposes of this calculation. [4]
The predicted loss ratio for the business next year is 90%.
(ii) Comment on whether this is a problem for the general insurance company and
list the actions the company might take. [6]
[Total 10]
2 You are an actuary working for a newly established general insurance company. It
commences writing household contents insurance on 1 September 2006 writing
annual policies only. The company sells the following number of policies per month
in 2006:
Month Policies sold

September 1,000
October 1,500
November 2,000
December 2,500
(i) Describe the claims characteristics of household contents insurance. [3]
(ii) Calculate the average accident date for accidents occurring during 2006 by
considering the company’s exposure profile. Assume that policies incept on
the first day of the month in which they are sold. State any other assumptions
that you use. [4]
You have obtained benchmark accident year development factors by analysing a

sample of long-established general insurance companies offering similar policies and
with similar business mixes as your company.
(iii) Discuss the appropriateness of using this benchmark pattern to project the
2006 accident year to ultimate.
[4]
[Total 11]
ST3 S2007—2
3 You are the actuary of a large general insurance company that writes all types of
property insurance. The management is reviewing the reinsurance programme, which
has historically comprised facultative and treaty arrangements and has asked for your
comments.
(i) Define the terms facultative reinsurance and treaty reinsurance. [2]
(ii) Compare the advantages and disadvantages to the insurance company of using
facultative arrangements rather than treaty arrangements. [5]
(iii) Discuss the factors that you would consider when assessing the effectiveness
of the existing reinsurance programme. [6]
[Total 13]
4 You are a consulting actuary engaged by a small general insurance company which
writes only motor business. The company sells its policies through a broker network
and direct to customers. It has been attempting to grow its business and written
premium and policy volumes have both increased in the last two years. However, its
profit margin has fallen over the same period. The company has asked you to analyse
its business and identify reasons for its falling profitability.
Describe the actuarial analyses that you would carry out in each of the following areas
and discuss the features for which you would be looking in your analyses.
(i) Claims [5]

(ii) Premiums [4]
(iii) Mix of business [5]
(iv) Reinsurance [4]
[Total 18]
ST3 S2007—3 PLEASE TURN OVER

5 You are a pricing actuary of a general insurance company writing employers’ liability
for large companies. An underwriter has given you the following information on a
risk that he wishes to renew in 2007. Under the proposed terms of the policy there is
a limit for each and every loss of £5m and no deductible.
Underwriting Number of Notified claims

Year Employees (£000’s)
2003 2,100 1,200

2004 2,600 1,400
2005 2,400 1,200
2006 2,500 550
2007 2,800
The underwriter suggests that the underwriting year incurred development pattern for
similar risks is:
Underwriting % Developed
Year
2003 90%
2004 80%
2005 60%
2006 25%
(i) Calculate the risk premium for 2007 using a burning cost methodology
allowing for claims inflation of 10% per annum, indicating any assumptions
that you make. [8]
(ii) Explain why the answer that you calculated in (i) may not be an accurate
indication of the “true” risk premium. [5]
(iii) Further to the points raised in (ii), state what further information you would
require to calculate your theoretical office premium for the risk. [3]
The actual office premium quoted for the risk is different to the theoretical office
premium that you recommend.
(iv) Discuss the situations that could give rise to this, other than those already
raised in (ii) above, and outline the concerns you would have in each case. [5]
[Total 21]
ST3 S2007—4
6 You are an actuary of a large general insurance company writing private motor
business. You are estimating the technical reserves for this class of business as at
31 December 2006. You have been given the net paid accident year triangle for the
class as a whole, from which you have calculated development factors using the basic
chain ladder method, as shown below.
Net paid development factors on cumulative amounts
Development year
Accident year 0–1 1–2 2–3 3–4 4–5 5–6 6–7
1999 2.259 1.318 1.255 1.110 1.054 1.040 1.030
2000 2.265 1.262 1.189 1.134 1.049 1.046
2001 2.464 1.280 1.228 1.130 1.052
2002 2.341 1.337 1.209 1.065
2003 2.747 1.342 1.179
2004 2.692 1.239
2005 2.474
2006
Selected idf’s 2.500 1.300 1.200 1.100 1.050 1.040 1.030

Selected cdf’s 4.825 1.930 1.485 1.237 1.125 1.071 1.030
idf = incremental development factor

cdf = cumulative development factor
(i) Suggest reasons why actual ultimate claims paid may differ from those
estimated using the paid chain ladder method. [5]
(ii) Discuss other analyses and projections that you would wish to perform before
selecting your undiscounted outstanding claims reserve, including IBNR. [11]
(iii) (a) Outline the arguments for and against discounting of reserves.
(b) State the factors that you would consider in arriving at an appropriate
discount rate. [5]
The company has decided to discount its reserves for this class using an interest rate
of 9% per annum.
(iv) Calculate an appropriate discount factor to apply to the undiscounted reserves

for accident year 2004 as at 31 December 2006, stating any assumptions you
make.
[6]
[Total 27]
END OF PAPER
ST3 S2007—5
EXAMINATION
September 2007

EXAMINERS’ REPORT
Introduction
M Stocker
December 2007
Comments
Individual comments are shown after each part-question or question.
1 (i)
• Probability of ruin Ψ(U) ≤ e-RU (Lundberg’s inequality)
• where U = initial capital = 10
• R = adjustment coefficient = aθ/(1 + θ)
• where a is the exponential parameter i.e. = 1/0.5 = 2
• and θ = is the premium loading factor
• so 1/(1 + θ) = x, where x is the loss ratio
• Set 0.005 = e-RU and solve for x to find the required loss ratio
• R = 2(1 - x)
• 0.005 = e-20(1-x)
• x = 1 + ln(0.005)/20
• x = 73.5%
Comments on Q1(i). Some students showed that 73.5% loss ratio implies a
ruin probability of less than 0.5%. This approach gained equal credit. Many
candidates got full marks on this part.
(ii)
• 90% LR implies a probability of ruin much higher than 0.5%, so this may
be a problem if the insurance company requires a 0.5% or lower
probability of ruin…
• … for regulatory solvency/capital requirements, for example
• … or if this would adversely affect insurer's credit rating
• Lundberg’s inequality gives probability of ruin < 13.5%
• Being a small company, parameter uncertainty surrounding the predicted
90% LR is likely to be greater as it is being estimated on less data
• Lundberg's inequality is probability of ultimate ruin so concern will
depend to an extent on whether the 90% LR is expected to persist; the
current position within the insurance cycle may mean that market is
currently soft but expected to harden in the next few years
• The extent to which this is a problem also depends on the level of expenses
and commissions
• as well as the investment return expected on the assets held to back the
reserves
Possible actions:
• Investigate why the predicted loss ratio is 90%
• Consider why insurer requires a probability of ruin of 0.5%. If competitors
live with a higher probability of ruin then at a competitive disadvantage
• Consider reinsurance: might be able to reduce probability of ruin by using
reinsurance to reduce variability but will be passing on profit to the
reinsurer
• Reduce expense/commission ratio to compensate for higher LR
• Tighten claims handling/settlement procedures to reduce payouts
• Consider diversification into other lines (currently writing specialised
book)
• Enforce tighter underwriting criteria to exclude bad risks
Page 2
• Reduce coverage using lower limits and/tighter policy wording/higher

excess or deductible
• Increase premium rates
• Do nothing; if 90% LR is thought to be the soft part of insurance cycle,
management may make a strategic decision to write across the cycle
Comments on Q1(ii). Most students were able to list the various actions to
bring the loss ratio down. Not so many mentioned the insurance cycle and
virtually none pointed out that probability of ruin is to ultimate, not just over
the next year.
2 (i) Household contents insurance:

Claim event is usually sudden and easily determinable (e.g. burglary, fire)
Notification is normally prompt
Settlement is usually quick
Often just consists of a single payment although total losses (e.g. from fire)
may take longer to settle and be settled in parts
Claim amount can normally be estimated accurately
Claims tend to be fairly consistent in size and distribution
Claim amount tends to be low relative to buildings cover
Frequency tends to be high relative to buildings cover
As a class, can be exposed to accumulation/cat risk
As a class, very exposed to the risk of moral hazard
Frequencies closely linked to the economic cycle,
e.g. theft claims frequencies rise when unemployment rises
Claims costs tend to rise in line with price inflation
May have nil claims as an excess often applies
Comments on Q2(i). Bookwork on which most candidates scored reasonably well.
(ii) Assumptions:
Even risk profile over the year
Identical policies
No cancellations
Claims occur on average mid-month
(In each case, alternative assumptions are valid if correctly applied.)
Calculation:
(1) (2) (3) (4)
Policies Earned exposure = (Month – 0.5) ×

Month written policies on risk earned exposure
9 1,000 1,000 8,500
10 1,500 2,500 23,750
11 2,000 4,500 47,250
12 2,500 7,000 80,500
7,000 15,000 160,000
If a non-constant earnings assumption is used then column (3) needs to be

adjusted appropriately.
Page 3
Average accident date = sum(4)/sum(3) = 10.667, i.e. two-thirds of the way

through November 2006
Comments on Q2(ii). Some silly mistakes such as giving average accident

date in 2007 by ignoring the statement “occurring during 2006”. Various
approaches are possible, all given equal credit if correctly reasoned and
applied.
(iii) Long-established => writing business for many years. The benchmark
companies’ portfolios are likely to be quite stable without the rapid growth in
volumes that we have experienced. In particular, their pre-September exposure
is unlikely to be zero like ours.
Their aggregate earnings patterns are therefore likely to be much more even
across a year, so their average accident date is likely to be around mid-year
and our 2006 accident year will develop later than the benchmark accident
year
Therefore applying the benchmark development factors is likely to
underestimate the true 2006 accident year ultimate
So the benchmark is not appropriate for use unless first adjusted for this lag
effect.
General remarks on why benchmarks development factors may not be
appropriate:
benchmarks may have faster/slower claims handling procedures;
may have different reinsurance arrangements (if considering net);
policies though similar won't be identical;
business mixes though similar won't be identical;
benchmarks could contain large claims/cats which could distort
development pattern;
benchmarks may use different inflation assumptions;
benchmarks might/might not include ALAE
may not expect future claims to develop in same way as past.
Comments on Q2(iii). Virtually all students were able to recall the standard
points for/against benchmarks. Only a few linked the answer back to (ii) and
discussed why the unadjusted benchmark would probably under-project in this
case.
3 (i) Facultative reinsurance is a reinsurance arrangement covering a single risk,

commonly used for very large risks or portions of risk written by a single
insurer
Treaty reinsurance is a reinsurance arrangement that a reinsurer is obliged to

accept, subject to conditions set out in a treaty
(ii) Advantages of using facultative reinsurance

• The insurer can choose the risks that it believes are most advantageous to
cede or retain from the point of view of maximising profit and subject to
acceptable retention of risk
Page 4
− under an obligatory/obligatory treaty, the insurer is required to cede all

risks that meet the criteria of the contract
− under a facultative/obligatory treaty, the insurer can choose which risks
it wishes to cede
• Facultative reinsurance can cover risks that
− fall outside treaties
− fall inside treaties but outside the contract limits
• Easier for insurers to shop around for best rate/terms in market.
Disadvantages of using facultative reinsurance

• It is a time-consuming and costly exercise to place each and every risk
separately
• The insurer may not be able to accept and provide cover immediately until
it has had the opportunity to find appropriate facultative reinsurance cover;
under the treaty all risks covered by the contract are ceded automatically
• This may impact the company’s stance in the market with distributors and
customers
• There is no certainty that the cover required will be available when
needed; under the treaty the cover is predetermined in the contract
• Even if cover is available, the price and terms may not be acceptable;
under the treaty, the price and terms are all predetermined at the start of
the contract
(iii) Discussion of Factors:

• Look at the reinsurance profit and loss accounts (or suitable ratios, e.g.
recovery ratios) historically by year to understand the level of profit or
losses ceded
• Analyse separately for each treaty to understand the impact of each
arrangement on the overall result
• Analyse different levels of facultative arrangement (e.g. by size of risk) to
determine profitability by amounts ceded
• Check to see if the period has been atypical in terms of claims experience
and adjust accordingly
• Compare cost of other forms or levels of reinsurance that could have been
used.
• For example, would a Property per risk treaty with a higher or lower
retention have proved more profitable for the insurer (or other appropriate
example)
• What are the current reinsurers’ credit ratings? Is there a need to move
cover to more secure companies?
• What level of financial assistance are the current reinsurers providing? Do
these adequately cover acquisition and administration costs?
• What level of technical assistance is provided by the reinsurers?
• Can the insurer find more tailor-made solutions elsewhere?
• Any reciprocal arrangements in operation need to be assessed to determine
the profitability of these arrangements and their effectiveness in reducing
risk concentration
• Look at how the reinsurance programme reduces capital requirement
Page 5
• Check extent to which reinsurance programme covers accumulations of

risk in book
• Analyse the profit smoothing achieved by the reinsurance programme
• Consider any regulatory constraints or relaxations that the reinsurance
programme has caused
• Has the reinsurer imposed any conditions, e.g. minimum retention, certain
policy conditions?
Comments on Q3. Bookwork which was generally well answered. A few

students were confused between facultative reinsurance and fac/oblig treaties
and some mistakenly answered part (iii) as reasons for purchasing
reinsurance.
4 (i) Claims
Claims data needs to be collated into homogeneous groups and triangulated
Analyse claim numbers and average claim cost separately to identify source of
poor experience. This may be a result of increased frequency, increased
severity or both.
Analyse claims paid, outstanding reported claims reserves and IBNR
separately in case one has developed unusually
Analyse historic claims inflation to see if claims costs increasing faster than
expected
Investigate large claims experience. Poor profitability may be explained by a
few large claims in the last couple of years.
Where there is sufficient data, claims may be analysed by claim type to isolate
the claim type(s) developing poorly
Where possible, also analyse claims on quarterly/monthly basis to see if any
clearer trends emerge
Review the company's historic reserving (are their methods appropriate?)
There may be errors in their calculations leading to overstatement of reserves
Review strength of reserving basis. Are assumptions too conservative?

Alternatively, an historically weak basis may have resulted in the need to
increase prior year reserves
Investigate changes in claims handling procedures. Claims adjusters setting
case estimates may have become more cautious over time.
Claim settlement controls may have slipped resulting in higher payouts (e.g.
due to new claims handling staff being inadequately trained or increased
claims volumes and not enough staff to handle workload efficiently or other
valid examples)
Compare company claims development with external/industry benchmarks
Were there any accumulations of risk in the book which have now caused
losses?
Investigate claims for signs of fraud
Compare rate- and inflation-adjusted historic loss ratios to identify trends in
underlying risk
Could analyse loss ratios for new business and renewals separately to see if
experience of one group is worse than the other
Page 6
(ii) Premiums
Analyse claims together with exposure to determine adequacy of risk premium
rates
Split by rating factor (age/sex/vehicle type, etc.) into rating cells
For each rating cell, compare total cost of claims against theoretical risk
premium predicted by company's rating structure
Identify particular factor levels where inadequate premiums are being charged
which may indicate anti-selection.
Where sufficient exposure to perform credible analysis, split by claim type
(AD, TPPD, TPBI, etc.) to identify more accurately the inadequacy in the
rating structure
But company is small so credibility of data may be an issue
Consider changes in cover that may have led to increased claims
Analyse adequacy of each loading:
Expense loading may no longer cover costs (office space, staff costs) arising
from increased written volumes
Commission loading may be too low, e.g. because commission rates boosted
in order to attract more broker business
Profit and/or contingency loadings may have been cut to make premiums more
competitive
Use historic company and market data to investigate position in insurance
cycle. Falling profitability may be a feature of the market as a whole
Look at historic changes in market share. Changes in share can indicate
premiums out of line with market.
(iii) Mix of business

Investigate change over time of:
Mix of policyholders: proportion of higher risk policies may be increasing,
e.g. young male drivers, high performance cars, more policies in high theft
areas?
Mix of policy type, i.e. comp/non-comp, as these typically experience
different loss ratios.
New business/renewal ratio as acquisition costs are higher than renewal costs
but usually spread across all policies. Therefore higher than expected volume
of new business will mean lower contribution to expenses than expected
Source of business: broker/direct as broker business more costly to acquire
due to commissions. May have seen broker business increase and direct sales
fall.
Analyse profitability and persistency by source. Identify sources
(phone/internet) or particular brokers where the experience is poor.
Analysing lapse rates may indicate rating cells where we have lost profitable
business to competitors
(iv) Reinsurance
Analyse historic recovery ratio (RI recoveries/RI premium earned). Has it
fallen in the last two years?
The company may be failing to structure its RI programme appropriately for
the increased business, e.g. aggregate XL limit too low so vertical exhaustion
more likely
Insufficient reinstatements on treaties causing horizontal exhaustion
Page 7
The price of reinsurance may have risen disproportionately

RI recoveries may have fallen due to:
new exclusions, tighter policy wording imposed by reinsurers
increased deductibles
higher attachment points
Check for reinsurer failures and disputes. Bad debt provisions may have
impacted results.
Comments on Q4. Most candidates did not get many of the points available
for this question. Better answers covered a broad range of points and also
clearly explained why and how the proposed analyses would be useful in
investigating declining profitability. Poorer answers tended just to list the
various analyses possible without linking these to the specific details of the
question. This was especially the case in (iv).
5 (i) Assume that :

• Employee count definition remains constant
• No change in structure of company e.g. disposals/acquisitions
• No change in safety procedures/substances handled
• No unusual losses in the data
• No change in the terms and conditions of the policy e.g. limits, deductible
• Underwriting pattern provided by underwriter is reasonable
• Future risk profile is same as historic risk profile
Calculation:
Underwriting CL Ultimate Inflation Inflated Inflated ultimate

year index Ultimate per employee
2003 1,333,333 146.41 1,952,133 929.6
2004 1,750,000 133.10 2,329,250 895.9
2005 2,000,000 121.00 2,420,000 1,008.3
2006 2,200,000 110.00 2,420,000 968.0
average 950.5
weighted average 950.1
average excl. 2006 944.6
weighted average excl. 2006 943.9
• Make a reasonable selection: average or weighted average

• Taking e.g. weighted average, risk premium for 2007 is: 950.1 × 2800 =
£2.66m
(ii) Explanation:
• Frequency-severity approach would provide more information on the risk
and might improve accuracy of answer
• Number of employees may not be the ideal exposure measure. Wage-roll
might be better (number of employees okay for frequency)
• Insufficient data to give full credibility to answer
• Unlikely to be sufficient large losses in the data to give reasonable
allowance for future large losses
Page 8
• Or may be heavy experience in the data e.g. unusual large losses

• Need to investigate trends in the data
• Any of the assumptions identified in (i) may be incorrect
• Exposed to latent claims so may need to include a loading
Comments on Q5(i) & (ii). There were some strange calculations where
students tried to triangulate claims using the development pattern in order to
apply the inflation adjustment. Very few discussed alternative methods or
exposure measures in (ii). Otherwise the question was generally well
answered.
(iii) Further information:

• commission rate
• expense loading
• contingency loading
• return on capital required/profit requirement/ target loss ratio
• tax rate
• cost/structure of reinsurance
• investment return
Comments on Q5(iii). Bookwork. Almost all candidates scored full or near

full marks.
(iv) Situations:
• Could be a hard/soft market at present so we may be able to charge
more/less than the theoretical office premium under present conditions
(adjustment for insurance cycle)
• May make a decision to price over/under the competition (regardless of
insurance cycle)
• Regulations may restrict the premium we can charge
• We may wish to sell the cover as a loss leader or as part of a combined
package
• Underwriter may have “soft” information giving him additional
information about the risk that enables him to adjust the premium
• The broker may apply pressure to accept a lower premium
Concerns:
• Need to consider the premium charged for the risk in previous year as
customer would compare to this
• There will be less concern if a higher than theoretical premium is charged
in a hard market
• Although if actual rates are too high this might be an indication that we are
losing out to competitors on other profitable business and this may result
in volumes insufficient to cover fixed expenses
• The concern in a soft market will be how low rates will have to go and
how long the soft market will persist
• Charging less than the theoretical premiums may eventually lead to losses
and to capital/solvency issues
Page 9
Comments on Q5(iv). Generally well answered.
6 (i) Reasons:
Future claims inflation is not a weighted average of past inflation
Payment pattern may extend beyond development year 7 (i.e. a tail factor is
needed in the projection)
Payment pattern may have changed across accident years due to:
Change in incidence pattern of claims during each year, e.g. harsher than
normal winter causes more claims
Change in reporting delay, e.g. postal strike, bad weather
New claims handling procedures/systems
New legislation, e.g. TPBI settlements change from lump sum to
periodical payments
Large claims (or the absence of such) may invalidate payment pattern as they
usually have a different development pattern to non-large claims
The reinsurance program may have varied between accident years
Reinsurance recoveries may have been disrupted by defaults/disputes
The mix of business may have changed during the period due to:
change in balance of comp and non-comp business
rapid growth or contraction of business
change in mix of policyholders (e.g. age distribution, male/female split)
A change in any of these could influence reporting and settlement patterns
Random variation
The selected development factors may not be appropriate. There is
uncertainty surrounding the selected development factors:
At later developments they are calculated on relatively little data (one or
two years)
At early developments they are generally large so a small percentage error
in the factor can lead to a large numerical difference in the estimated
ultimate
Any sensible comment on the development factors given
The data may contain errors
Comments on Q6(i). Answers were reasonable but students could have

gained more marks by systematically going through the possible factors which
might invalidate the constant development pattern assumption.
(ii) If possible, a better starting point is to project gross claims and then net down
as a separate step. This avoids distortions to net patterns caused by changes in
RI programme over time.
Fit a tail to the net paid projection.
Use more historical data if available
If not, could try fitting a curve to the development pattern or consult
industry/external benchmarks
Also project on a monthly or quarterly basis to compare
Project gross/net notified claims as a check on the gross/net paid BCL
Notified claims includes case estimates so factor to ultimate is smaller. May
consider notified claims projection to be more reliable than paid projection
Split claims into homogeneous groups and project separately, e.g. by claim
type (AD, TPPD, TPBI, etc.), by comp/non-comp, by geographical location
Page 10
Advantages: this gives more homogeneous groups of data which are more
likely to develop in the same way.
Avoids distortions caused if mix of claim types changes in future
Disadvantages: each projection is on a smaller volume of data so less
credible, more volatile
Large insurer: may write business across several countries, currency may be
an issue:
conversion of payments at historic exchange rates will distort the
development pattern
better to compile triangle using constant exchange rate across whole
triangle
or project currencies separately if enough data for credible projection
Large claims (say > £250k) in private motor are usually TPBI.
These are likely to be longer-tailed than other claims and have a different
development pattern
Therefore sensible to remove large claims from triangles and project large and
non-large separately.
Definition of large should vary (by large claim inflation) between accident
years, otherwise you would effectively be removing more claims in earlier
accident years
Due to low frequency/high severity nature, claims development will be lumpy
and irregular so any statistical projection methods for large claims will be
problematic.
Individual case estimates probably best for outstanding reported large claims
reserve.
Large claims IBNR: could apply a simple loading to large outstanding if
historical data supports this or frequency-severity approach: project numbers
of notified large claims to get an IBNR number. Estimate an average cost of
large IBNR claims by comparing historic IBNR with reported sizes and
allowing for any trends, known changes
You need to be consistent in how you treat claims which are “large” at some
point but eventually settle for an amount less than “large”
For 2005 (and 2004) accident years:

The chain ladder method is very unstable (factor to ultimate is high for the
immature years)
The Bornhuetter-Ferguson provides a more stable estimate whilst taking some
account of actual development
Alternatively could use the Expected Loss Ratio method for these years
Other projection methods:

Inflation adjusted CL if historic claims inflation has varied considerably over
time and/or future inflation is likely to be different to the past
ACPC (and/or inflation adjusted version) to study development of frequency
and severity separately
Need to allow correctly for nil claims here
Stochastic methods (e.g. bootstrapping), to get an idea of the reserve
uncertainty or if a specified percentile is required
Page 11
The methods and assumptions used will be influenced by the purpose of the
reserving exercise
Comments on Q6(ii). Generally reasonably good answers.. Most students were

aware of the different claim types in private motor and why they need to be studied
separately. Some answers simply listed different projection methods without any
discussion of why they might be used. Unfortunately, quite a few students seem to
think that projected accident year paid amounts do not include IBNR.
(iii) (a) Arguments for discounting:

In the case of long-tail business investment income can be significant
It gives a more realistic position of the financial condition of the
insurance company
This is particularly relevant for management accounts
and for estimating claim costs for rating purposes
and allows easier comparison between different classes
Arguments against discounting:

For short-tail classes discounting makes negligible difference
Not discounting provides an automatic margin for uncertainty
For long-tail classes, the greater uncertainty justifies the margin
provided by not discounting
May be seen as a sign of weakness if market practice is not to discount
(where regulations allow) not discounting increases reserves and defers
the emergence of profit and therefore, tax
Where regulations don't specify it, discount rate is a subjective choice
requiring extra calculations/time
Regulations may prohibit discounting
(b) Factors to consider in choosing discount rate:

The purpose of the accounts. Management accounts would probably
use a more realistic rate than solvency/published company accounts
Regulations might dictate or limit the discount rate to use
The expected returns from the assets held to back the technical
reserves for this class
Proportion of non-investible assets held
Delay before receiving premiums
Investment expenses/tax
Consistency with inflation assumptions used for projections
Discount rate used last year
Any margin to reflect riskiness of class
Comments on Q6(iii). An easy question with most students scoring highly.
(iv) Use selected idf’s and assume no tail (or else select a tail and apply correctly
in calculations)
Assume payments occur on average mid-way through the year
Page 12
Development yr 2 3 4 5 6 7
Selected idf 1.200 1.100 1.050 1.040 1.030 1.000
1. Selected cdf 1.485 1.237 1.125 1.071 1.030 1.000

2. Cum % dev 67.4% 80.8% 88.9% 93.4% 97.1% 100.0
%
3. Increm % dev 13.5% 8.1% 4.4% 3.7% 2.9%
4. Increm % paid for 2004 acc 41.3% 24.8% 13.6% 11.4% 8.9%
yr reserve
Time t in years 1.00 2.00 3.00 4.00 5.00

5. Discount factors (1+i)-(t-1/2) 95.8% 87.9% 80.6% 74.0% 67.9%
6. Payment * discount factor 39.5% 21.8% 11.0% 8.5% 6.1%
7. Discount factor = sum of 86.8%

row 6
Comments on Q6(iv). Surprisingly few candidates were able to calculate a discount

factor correctly. This may be partly explained by a lack of time. Some scripts did show
evidence of time pressure towards the end of the exam.
Page 13
EXAMINATION
17 April 2008 (pm)

booklet.
supervisor.
question paper.
In addition to this paper you should have available the 2002 edition of the Formulae
and Tables and your own electronic calculator from the approved list.
1 Describe the role of the broker in transacting insurance in the London Market. [5]
2 You are an actuary who works for a large general insurance company that has been
writing personal lines business for many years and that has kept good marketing
records. The marketing director has asked you to model the effect on policy sales of
changing each of:
• the per policy profit loading and

• the amount spent on advertising.
(a) Describe the impact that these two items will have on the company’s total
profit and
(b) Explain ways in which a model may be set up to estimate the future profit
arising from changes in these two items.
[6]
3 You are an actuary working on secondment overseas in a country that has no guidance
in place for the actuarial profession. A member of your team has asked you about the
importance of guidance in your home country.
(i) State the advantages and disadvantages of having guidance in place. [5]
You have been asked to draft a formal report on the year-end reserving exercise
conducted by your team. The company’s board will also review the report. The
board consists of non-actuarial members only. You have been asked specifically to
include a detailed section on “methodology and assumptions” used.
(ii) Describe the sub-sections that you would include in the “methodology and
assumptions” section. [5]
(iii) List the other key sections that you would include in a formal reserving report.
[2]
[Total 12]
4 Describe how you would project the various components of the existing overall
liability outgo for a motor portfolio in each future time period. [13]
ST3 A2008—2
5 (i) State the two definitions of burning cost premium and show how they are
linked. [2]
(ii) You have ten years of premium and individual paid claims development data,
gross of reinsurance, for a particular product. Explain the steps you would
take to analyse and adjust these data for the purpose of reviewing the risk
premium for that product. [13]
(iii) For a separate project you have claim data for claims reported between 30
September 2006 and 30 September 2007. The average reporting delay is nine
months. You are setting the rates for annual policies sold between 1 January
2008 and 30 June 2008. Claims inflation was 3% in 2006, 4% in 2007 and is
expected to be 5% in 2008 and 6% in 2009. Using a claim at the mid-point of
the periods given, calculate the inflation adjustment needed for the 2008 rating
series stating any assumptions made. [4]
[Total 19]
6 (i) State the assumptions underlying the Empirical Bayes Credibility Theory
(EBCT) Model 1 and Model 2. [4]
X1, X2, ...Xn are the aggregate claims, or the number of claims, in successive periods
for a risk. Let ao, a1 , , ..., an be the values which minimise the expression:
E[(m(θ) − a0 − a1 X1 − a2 X2 − ... − an Xn)2]

where m(θ) = E[Xj│θ]
You are given
• s2(θ) = V[Xj│θ]
• E[Xj m(θ)] = V[m(θ)] + (E[m(θ)])2
• E[Xj Xk] = V[m(θ)] + (E[m(θ)])2
• E[ X 2j ] = E[s2(θ)] + V[m(θ)] + (E[m(θ)])2
(ii) For Model 1 show that the estimate of m(θ) is given by:
(1 − Z) E[m(θ)] + ZX
where:
n
X = Σ Xj / n
j=1
and:
n
Z= [5]
n + E[ s (θ)]/ V [m(θ)]
2

Company A’s records show the claim amounts and number of policies from its
household book of business over the last 5 years. You are given the corresponding
claim amounts and number of policies from company B, as shown below:
Company A Company B
Year £000’s Policies £000’s Policies
2003 300 100 300 110

2004 280 90 330 110
2005 310 70 260 90
2006 270 95 310 105
2007 240 78 300 105
(iii) Using EBCT Model 2 calculate E[m(θ)], V[m(θ)] and E[s2(θ)] and hence the
expected claim amount for company A, where m(θ) = E[Xj ⎢θ] and
s2(θ) = PjV[Xj ⎢θ]. [9]
(iv) Comment on your results given that using Model 1, E[m(θ)] = 290, V[m(θ)] =
60, E[s2(θ)] = 700 and the expected claim amount for company A is £287,000.
[1]
[Total 19]
7 You are a consulting actuary. You have been asked to review and comment on a
client’s reserving methodology, and to re-project some of the key classes of business,
in order to form a view on the client’s year-end reserve figures. The client is a large
international company with a variety of classes of business being sold in the US and
some countries in Latin America.
After reviewing their methodology you note the following:
• Incurred claims data are used for projections.
• Projections are performed at a net of reinsurance level.
• For the Bornhuetter-Ferguson (B-F) method, the expected loss ratio (ELR) is
calculated as a rolling average of the ultimate loss ratios, using the chain ladder
method, for the previous three accident years.
• The incurred claims data from the Latin American countries is converted into $US
prior to projecting. Due to concerns of possible hyper-inflation in these countries,
only the latest diagonal is adjusted. All previous diagonals are equal to the data in
$US from the previous year reserve calculations. This is done using the latest
exchange rates and by converting the movements in paid and outstanding claims
only.
(i) Comment on the advantages and disadvantages of this approach and highlight
any recommendations that you would make to the client. [12]
The company is writing some new books of business. For these classes the company
has based the latest ELR on the underwriter’s view.
ST3 A2008—4
(ii) State the advantages and disadvantages of this approach, again providing any
recommendations that you may have. [3]
You have decided to re-project the following data for a motor book of business. You
have been told that the claims inflation has been steady at about 5% for each year.
You have no information about premium rate increases, but you can approximate this
using the increase each year in the average premium per policy.
Year Earned Earned Incurred Incurred Selected Selected

Premium Policy Claims Cumulative Ultimate Ultimate
$000’s Years $000’s Development Loss Loss
Factor Ratio
2002 11,750 1,150 8,765 1.000 75% 8,765

2003 13,000 1,275 10,350 0.960 76% 9,936
2004 12,500 1,125 9,235 0.940 69% 8,681
2005 13,250 1,050 9,500 0.920 66% 8,740
2006 15,250 1,125 11,250 0.975 72% 10,969
2007 17,650 1,265 9,575 1.520
(iii) State the key assumption being made when the increase in the average
premium per policy is used as the assumed premium rate increase. [1]
(iv) Calculate the incurred B-F ultimate for the 2007 accident year with the ELR
based on the weighted average (by premium) of the last five accident years.
[5]
When looking at the summary list of reserves by country and class, you notice that
there is a book of business called MisMass. You have not been provided with any
data for this class; however from the summary list of reserves you can derive that the
reserves have been estimated to be £142m and the discounted reserve figure for this
book is £86m.
(v) Determine the implied Discounted Mean Term (DMT) using a discount rate of
5% per annum and hence comment on what this book of business may contain.
[2]
You have requested additional data from the company on MisMass to show how the
reserve figures were derived.
(vi) State what reserving methods would be applicable for projecting the claims
data for this book of business, and comment on any limitations of these
methods. [3]
[Total 26]
END OF PAPER
ST3 A2008—5

EXAMINERS’ REPORT
April 2008
Introduction
M A Stocker
June 2008
Comments
Individual comments are shown after the solutions to each part question that follows.
1 London Market Brokers:

Act as specialist intermediaries e.g. Lloyd’s broker, reinsurance broker
Are agents of the insured
Organisational aspects of their work are important as they are often international
organisations and clients and the insured are often in different countries/time zones
Place business using the slip system
Prepare the appropriate slip, which shows the main features of the risk to be insured,
in a standard format
Prepare additional information relating to the risk at the request of the underwriter and
further technical assistance
Present the slip to a lead underwriter, who on the basis of the slip and additional
information will set the rate and terms
Present the lead underwriter’s rate and terms to the insured for approval
The lead underwriter with the broker indicates the proportion of the risk (line) he is
prepared to accept (by signing and stamping the slip)
If the insured agrees to the rate and terms, the broker will approach the follow market
to place the remainder of the risk on the lead rate and terms
Follow underwriters will sign their lines and stamp the slip in a similar fashion to the
lead (– they act as co-insurers)
The broker continues until the risk has been over-placed
The broker, in agreement with the insured, signs down the lines of the underwriters so
they total 100%
If the broker fails to place 100% of the risk, the insured may retain the remaining
proportion or more typically the broker will seek to renegotiate better terms
Collecting premiums from the insured and paying claims
Advising and negotiating the best terms for the insured
The broker also provides an admin role e.g. preparing policy documentation
Comments on Q1: A bookwork question on which many candidates scored well.
2 (i) Increasing advertising spend will increase overall expenses

However, if it is a successful campaign with high sales volumes then it could
decrease expenses per policy as the existing fixed expenses are spread more
thinly over the larger book of business.
Increasing profit loading will increase profit per policy.
However this will result in higher premiums and may reduce volumes of
business to a level that results in lower total profit
The overall effect on total profit will depend on the effect on the volume of
business and hence an assumption around volume of business is needed
The overall effect on profit will also be impacted by changes in the mix of
business resulting from varying these two items
The overall effect on profit will also be different depending on the mix of new
business and renewals
(ii) In terms of the model total profit is premiums less expenses less claims
Also any valid alternative profit formula e.g. (per policy profit load × policy
volume)
Set up the model to give total profit as an output
Review past company data on rate changes versus policy volumes
Page 2
In addition, obtain data from comparison web sites to help determine where
the premium currently sits in relation to the rest of the market in order to
determine sales volumes
Other expenses will be fixed or overheads that are independent of sales
volume and this will also complicate the total profit calculation
Analyse effect on volumes in different groups e.g. age, sales channel or social
economic group
Consider correlations between the profit loading and marketing spend
Produce a set of one-way or two-way analysis tables for each selected profit
loading showing how total profit varies by marketing spend, i.e. elasticity
and produce a set of one-way or two-way analysis tables for each selected
marketing spend showing how total profit varies by profit loading
Select the combination of factors which maximises total profit
Alternatively given that we have an output which is correlated to two variables
we could fit a Generalised Linear Model (GLM) to the model outputs
The GLM could then be used to maximise the total profit
Set up the model such that the factors can be easily interpreted and monitored
Comments on Q2: Most candidates acknowledged that changes in costs and profit
loading would affect volumes and many picked up full, or nearly full, marks on part
(i). However, in (ii), students tended to trot out the points given in the core reading
without thinking through and applying to the specifics of this particular question.
3 (i) + Promotes confidence in actuaries’ work, by ensuring that actuaries maintain

a high level of training and expertise
+ Being able to demonstrate compliance protects an actuary against litigious
criticism
+ Protects the public by making actuaries accountable
+ Promotes consistency and greater scrutiny of professional judgments
+ This in turn can reduce work loads e.g. consistent processes make it easier to
review/pick up a predecessor’s work
+ Enables fellow actuaries to be challenged
+ Provides actuaries with a point of reference for clarification
+ Consistent with some other professions
– Above advantages are only applicable if the guidance is well written and not
open to interpretation
- Guidance will never be fully comprehensive
- Guidance can become out of date
– Additional bureaucracy and time-consuming
– Infrastructure needed if not already in place, e.g. overseeing boards
– Additional costs
– Potentially restricts use of individual judgment
– Danger of setting standards that are too high/unrealistic, e.g. listing and
justifying every incremental development factor assumption in a reserving
exercise
Comments on Q3(i): ) A few candidates misinterpreted “guidance” to mean

“regulation”. Many students failed to generate many distinct points and so tended to
score badly on this part.
Page 3
(ii) State the methods used when selecting ultimate for each class of business, type
of reserves and year of account
Detailed description of the methods used
A detailed explanation of the methods underlying assumptions
A list and discussion of the key assumptions and judgments used
A description of the process by which the assumptions have been identified
Highlighting any illustrations based on assumptions that the are not regarded
as appropriate
The rationale for selecting the methods used in producing results and for each
method a statement of key assumptions
Where the results of different methods or assumptions presented in the report
differ significantly, comments on the likely reasons for the differences and an
explanation of the basis for the choice of results
Sufficient data and other information to understand the key assumptions made
and the process by which the assumptions have been identified
Comments on Q3(ii): Many candidates failed to restrict their answer to the

“methodology and assumptions” section as required and therefore wasted time
writing unnecessary points.
(iii) Introduction
Definition of terms
Purpose and scope
Information and data used
Analysis of emerging experience (comparison with last year)
Business conditions over the year
Results
Sensitivity testing of assumptions used
Uncertainty
Comments on Q3(iii): There was a mixed performance for this part of the question
with some candidates scoring well and others answering poorly.
4 Claim payments are almost always the most significant item of the liability outgo
Ultimate claims and payment patterns can be taken from the reserving calculations
Alternatively , project claim amounts forward using standard techniques (e.g. chain
ladder) to derive payment patterns
This will depend on the basis required for the projection e.g. claim amount are
required on a best estimate or prudent basis
Perform any adjustments to the data as necessary e.g. large claims, trends, inflation
etc.
Projecting forward for the full expected run-off period for the book Individual period
by period projections are important
Model the motor property damage separately from the bodily injury claims
As the bodily injury claims will be much longer tailed and may involve periodic
payments
Page 4
Calculations should be done on a monthly or quarterly basis

to take account of any seasonal effects
Split the business into cohorts written in the same period
Care needs to be taken to ensure that all types of claim amounts are projected e.g.
outstanding, IBNR, URR
For example triangles projected using an accident year cohort will not include URR
Split the business by currency where appropriate
Reinsurance and other recoveries should be included separately as an offset to the
liability outgo taking into the reinsurance programme
Reinsurance recoveries do not usually coincide with the claim payments and hence a
timing difference exists
Future known reinstatement premium, and other reinsurance premium, payments will
be modelled
If large gross claims have been modelled then it may be necessary to calculate the
appropriate reinstatement premium with the necessary delay
Model expense outgo

Fixed expenses can be modelled adjusting for inflation as appropriate
Some items may occur at specific times of year e.g. rates or local taxes
Allow for any one off expenses
Claim handling expenses can be projected with claims
Alternatively, claims handling expenses can be projected separately in a triangle
Taxes must be calculated on estimated profits and the timing will be known
Contributions to the MIB fund for uninsured drivers, or to the Lloyds central fund if
relevant should also be allowed for
Comments on Q4: Virtually all candidates took note that this question related to a
motor portfolio and mentioned the need to split out BI and property damage claims.
However, even though the question specified liability outgo, many students wasted
time writing about premiums and investment income. Similarly, we are interested in
existing liability outgo but some students talked about commission on business which
would be written in the future.
Some poorer answers failed to discuss expenses, concentrating only on claims outgo,
and therefore missed out on some of the easier marks available.
5 (i) BCP = Average Claim Amount x Claims Incident Rate

= (Total claim cost / number of claims) /
(Total Exposed to Risk / number of claims)
Number of claims cancel out to give the alternative form of BCP, namely
= total claim cost / total exposed to risk = BCP
Comments on Q5(i): Answered well by most candidates, although a small minority

didn’t seem to know where to start with the definition of burning cost, which is a
fundamental concept in General Insurance. Also some candidates defined burning
cost as total incurred claims, as opposed to ultimate claims, over total exposed to
risk.
Page 5
(ii) Analyse the frequency and severity separately

Break into homogenous groups if necessary and possible
Check for errors in the data
Assuming that we have no outstanding and IBNR data project the number of
paid claims to ultimate
Estimate an average cost per claim and multiply this by the projected number
of claims to obtain the estimated claim ultimate amount
Because of changes over time may not want to use all 10 years of data,
depending on data volumes
Adjust for large claims
Truncate, remove, spread or add claims cost to bring claims to an average
level
Adjust for unusually heavy/light experience
Decision depends on the extent to which such claims (or lack of claims) are
expected to recur during the exposure period of the new rating series
Assume no change in target market or distribution approach
Adjust for trends in claims experience – bring all data to the mid point of the
exposure of the new rating series
Decide how historic trends may project into the future to allow for them in the
projection of the risk premium
Adjust for changes in risk

They may show up as trends or as step changes
May appear due to changes in the mix of business
May be different for different types of claim
Adjust for changes in cover

They may show up as trends or as step changes
The major changes are likely to involve the perils covered and limits or
excesses applied to any claim
Perils may be excluded hence claim cost reduced
or specific claims excluded
or new perils added so an estimate of claims cost added
Changes in limits or excesses will need the claim cost distribution to be used
Data will be incomplete as claims below the excess point are unlikely to be
reported
Changes in underwriting or claim settlement procedures will result in similar

changes to those for changes in cover
Adjust for past and future inflation
Break down time periods into periods of uniform inflation
Adjust for the original period of writing the business and time period of new
rating series
Page 6
Adjust for delays between claim notification and individual payments until
final settlement, both for claims occurring in the base period and for those
expected to occur during the exposure period of the new rates
Determine the period of exposure containing the base experience
Determine the full period of exposure covering the claims that can arise from
the policies written under the new rates
and adjust for the timing difference
Project exposure values

Ensuring that each claim amount corresponds to the premium data
To arrive at a risk premium rate, we need a rate of claims per unit of exposure
Here these exposure units are expressed in terms of monetary units so the base
exposure values need to be projected at an appropriate rate of premium
inflation
Premium rate increases can be estimated by looking at historic average
premium rates (if actual rate increases are not available)
Although this assumes that the mix of business, terms & conditions etc.
remain the same in all years
Comments on Q5(ii): Many candidates tried to answer this part by giving the same
answers as in Q4. (Q5 relates to analysing and adjusting past data to make it
relevant to the future for the purposes of premium rating whereas Q4 was concerned
about the timings of cash flow projections.) Most candidates scored badly because
they failed to generate sufficient points for the 13 marks available. The better
answers covered the inflation adjustments required for the future rates, which linked
into part (iii). Most candidates failed to go into any or enough detail on the
projection of exposure values.
(iii) Assuming policies are uniformly written over the period

Assuming no change in reporting and no settlement delays
Assuming claims occur on average mid way through the year
Assuming that inflation is uniform over the calendar year
For claims: Inflate from the mid point of historic reporting to expected future
reporting date
Historic claims occur on average on 30/06/06
Historic claims reported on average on 31/03/07
Future claims occur on average on 30/09/08
Future claims reporting date 30/06/09
= 1.040.75 × 1.05 × 1.060.5
= 1.113
Marks provided for alternative methods for example:

Deflate from reporting date to written date
= (1/1.04)0.25 × (1/1.03)
= 0.9614007
Inflation of claim cost between the mid points of writing is
= 1.03 × 1.04 × 1.050.25
= 1.0843460
Page 7
Inflation from mid point of writing to mid point of reporting is

= 1.050.75 × 1.060.5
= 1.0679352
So total adjustment = 0.9614007 × 1.0843460 × 1.0679352
= 1.113
Comments on Q5(iii): A fairly easy projection calculation which most candidates

attempted reasonably well. Marks were lost mainly because the projection dates used
were inconsistent e.g. projecting from the average incurred date in the past data to
the expected average reported date in the future. A surprising number of candidates
expected to pick up a mark for making the assumption that policies are annual, failing
to notice that this is specified in the question. Some candidates tried to project the
claim data from the average date the claims occurred historically to the average date
the claims are expected to occur in the future exposure period despite the fact that the
question states that the claim data being projected are for claims reported between 30
September 2006 and 30 September 2007.
6 (i) EBCT 1 assumptions

The distribution of each Xj depends on a parameter, denoted θ,
whose value is fixed (and the same for all the Xj s) but is unknown.
Given θ, the Xj s are independent and identically distributed.
Where θ could be a real number or it could be a more general quantity such as
a set of real numbers.
The random variables {Xj} are identically distributed.
The Xj s are not (necessarily) unconditionally independent.
EBCT 2 assumptions
The distribution of each Xj depends on the value of a parameter, θ, whose
value is the same for each j but is unknown.
Given θ, the Xj s are independent (but not necessarily identically distributed).
E[Xj│θ] does not depend on j.

PjV[Xj│θ] does not depend on j.
Comments on Q6(i): Bookwork, although many students had not learnt this.
(ii) Differentiate equation with respect to a0 and put the derivative equal to zero.
This gives the following equation:
Equation 1:
n
E[m(θ) − a0 − Σ aj Xj] = 0
j=1
Which from the definition of m(θ) gives:

n
a0 = E[m(θ)] (1 − Σ aj)
j=1
Page 8
Next differentiate with respect to ak where k ≠ 0, and put the derivative equal
to zero. This gives:
n
E[Xk(m(θ) − a0 − Σ aj Xj)] = 0
j=1
which gives:
n
V[m(θ)] + (E[m(θ)])2 − a0 E[m(θ)] − Σ aj(V[m(θ)] + (E[m(θ)])2) −
j=1
ak E[s2(θ)] =0
Rearranging this last equation gives:
Equation 2:
n
ak E[s2(θ)] = (1 − Σ aj) (V[m(θ)] + (E[m(θ)])2) − a0 E[m(θ)]
j=1
This equation holds for k = 1,2,3,……., n. An important point to notice is that

the value of ak does not depend on k. In other words:
a1 = a2 = ... = an.
Now denote the common value of a1, a2, …..... , an by Z/n so that:
n
Z = Σ aj
j=1
and the estimator can be written as:
Equation 3:
n
a0 + Σ aj Xj = a0 + ZX
j=1
where:
n
X = Σ Xj / n.
j=1
Equations 1 and 2 are now two linear equations in two unknowns, a0 and Z.
The solution to these two equations is:
Equations 4 & 5:
a0 = (1 − Z)E[m(θ)]
n
Z= .
n + E[ s (θ)] / V [m(θ)]
2
So, the solution to the problem of estimating m(θ) given is given by the right
hand side of equation 3 with a0 and Z given by Equations 4 & 5, respectively.
Page 9
In summary, the estimate of m(θ) given X given by EBCT Model 1 is:
(1 − Z) E[m(θ)] + ZX
where:
n
X = Σ Xj / n
j=1
and:
n
Z=
n + E[ s (θ)] / V [m(θ)]
2
Comments on Q6(ii): This also was bookwork. This part was not even attempted by
the vast majority of candidates. Of those who did attempt it, many gave up mid-way
through.
(iii) Company A:
Volume mean
= (100 × 30 + 90 × 23.1 + …. 78 × 3.1) / (100 + …. + 78)
= 1,400 / 433 = 3.233
∑Pij(Xij - Xi)2
= [100 × (3 – 3.233)2 + 90 × (3.1 – 3.233)2 + 70 × (4.4 – 3.233)2
+ 95 × (2.8 – 3.233)2 + 78 × (3.1 – 3.233)2] / (5 - 1) = 123.2
Company B:
Volume mean
= (110 × 2.7 + 110 × 33 + 90 × 2.9 + 105 × 31 + 105 × 2.9) / (110
+ …. + 105) = 1,500 / 52.0 = 2.885
∑Pij(Xij - Xi)2
= [110 × (2.7 – 2.885)2 +110 × (3 – 2.885)2 + 90 × (2.9 - 2.885)2
+ 105 × (3 - 2.885)2 + 105 × (2.9 - 2.885)2] / (5 - 1) = 4.8
Px = (2 × 5 - 1)-1 × [433 × (1 - 433/953) + 520 × (1 - 520/953)]

= 52.5032
E[m(θ)] = (1,400 + 1,500) / (433 + 520) = 3.043.
E[s2(θ)] = (123.2/4 + 4.8/4) / 2 = 15.999
V[m(θ)] = [100 × (3 – 3.043)2 + 90 × (3.1 - 3.043)2 + … 105 × (3 - 3.043)2

+ 105 × (2.9 - 3.043)2] / (9 × 52.50) - E[s2(θ)]
Page 10
= [(0.2 + 0.4 + 134.4 + 3.8 + 0.1 + 11 + 0.2

+ 2.1 + 0.9 + 3.6) / 9 – 15.999] / 52.50
= [ (138.9+17.8)/9 – 15.999] / 52.50
= (17.412 – 15.999) / 52.50 = 0.0269
Z = 433 / (433 + E[s2(θ)]/V[m(θ)])

= 433 / (433 + 15.999 / 0.0269))
= 0.4215
Hence expected claims = 0.4215 × 3.233 + (1 - 0.4215) × 3.043 × number of

policies in 2007
= 3.123 × 78.0 = £243.6K
Assuming the same number of policies as the latest year or any other
reasonable assumption e.g. average of last 5 years, trend in average
movements, downward linear trend
Comments on Q6(iii): Many candidates scored highly on this part but some failed to
realise that the formulae required are in the formulae book. Not giving the number of
policies for 2007 in the question may have presented difficulties for some candidates.
(iv) The expected claim amount for company A is lower as Model 2 allows for the
volume of business which has recently declined for company A .
or any other reasonable observation
Comments on Q6(iv): For candidates that attempted this part of the question most
stated that this was due to Model 2 taking into account the volume of business.
7 (i) Projecting incurred claims data only

Projecting incurred claims data only is preferable to only projecting paid data.
Incurred projections are useful as paid claims development will be less mature
than incurred claims.
Paid projections, however, can help identify changes or inconsistencies in the
strength of case reserves or possible redundancies.
Projecting both bases can therefore reveal features of claim reserves that
would otherwise be missed.
Need paid claims development if discounted reserves required
Would recommend that the client projects both paid and incurred data.
Projecting net of reinsurance only

Projections at a net level will be robust as long as the proportion of
reinsurance recoveries remains stable.
It would be preferable to project at a gross level and apply the actual
reinsurance program to the projected future claims
This may not be feasible in practice.
Alternatively project at gross level and analyse the trend in reinsurance to
gross ratios for premiums, paid, incurred, outstanding claims in order to select
reinsurance IBNR ratios.
Page 11
In addition, understand change to the reinsurance programme and how this

will impact the IBNR ratios selected.
ELR using 3 -year rolling average

3-year rolling average is good where the historical selected ultimate loss ratios
are volatile over time.
Using a weighted average by ultimate premium though would smooth out any
volatility over the accident/underwriting year (as the years with the largest
ultimate premiums will be more stable)
Gives credit to the account’s own unique experience.
Easy to apply
but a fairly mechanical approach, which doesn’t take into account any possible
trends showing in the data/unusual years etc.
Claims and premiums are calculated on the same basis.
No account has been made to allow for premium rate changes and claims
inflation.
Would recommend a weighted average where the number of prior year
historical selected ultimate loss ratios is based on the data.
Would recommend loss ratios are adjusted to allow for premium rate changes
and claims inflation.
Recommend use of market stats benchmarks, where available for a sense
check – although need to ensure that the data is on a consistent basis.
Exchange rate conversion

This exchange rate conversion method would work on paid data but not on
incurred.
The outstanding need to be treated in a different way depending on whether
they are
New claims in the quarter/year
Prior claims with no movement in the quarter/year
Re-stated claims with movement in the quarter (e.g. due to additional
information or hyper-inflation)
For the first two type of outstanding claims the method is fine, for the last type
of claim the entire outstanding amount would need to be converted into US
$’s.
The method assumes that the movement in exchange rates move in exactly
offsetting ways to movements in inflation – this is not always the case in
practice.
Dealing with exchange rates is always complicated. This approach does have
its disadvantages but it is not unreasonable therefore no recommendation to
change the current method.
Comments on Q7(i): Candidates failed to generate many points for the 12 marks
available and so tended to score badly. The better answers contained a sensible
structure, e.g. they took each of the 4 notes in turn and considered the advantages,
disadvantages and recommendations, although there were still an insufficient number
of points made. The poorer answers tended to be very muddled. Some candidates
were unable to comment fully on why both paid and incurred projections are
important and the disadvantages of projecting at a net level only. Hardly any
candidates scored marks for the discussion of hyper-inflation and exchange rates, but
Page 12
it was recognised by the examiners that this was a difficult concept under the time
pressure of an exam. Many candidates mistakenly considered the BF ELR to be
lacking in independence, some even stating that this made the BF method to be the
same as the chain ladder method. However, because the ELR related to the previous
three years, it was in fact suitably independent. Incorrect assumptions about the
independence of the ELR for the BF method have been commented on in previous
examiners’ reports.
(ii) + Useful in exceptional circumstances e.g. a very new book of business with
no prior data or extremely volatile own history – as is the case for this
company.
+ Underwriter has a good knowledge of the business
+ May reflect market rate changes and inflation effects as well as trends in
claims frequency and average cost
+ Provides independent estimates
– Underwriter’s estimates may be too optimistic and hence not representative
of the actual loss ratios.
– Need to check consistency of the basis used.
This approach is reasonable, would recommend the use of market stats data as
well, where available, as a sense check
Need to ensure consistency with the basis e.g. gross/net
commission/reinsurance.
Comments on Q7(ii): This part of the question was answered reasonably well.
(iii) The mix of business is the same for each year.

or other reasonable key assumption
Comments on Q7(iii): The question asks for one key assumption and so students
should only state one assumption. The examiners are not impressed by a scattergun
approach in which candidates hedge their bets by stating a number of possible
assumptions.
(iv)
Year Earned Earned Incurred Incurred Selected Premium Claims 2007 Selected
Premium Policy Claims Cumulative Ultimate Rate Cost On- Ultimate
$000’s Years $000’s Development Loss Increase Increase Level Losses
Factor Ratio for 2007 for 2007 Loss $000’s
Level Level Ratio
2002 11,750 1,150 8,765 1.000 75% 1.366 1.28 70.1% 8,236
2003 13,000 1,275 10,350 0.960 76% 1.368 1.22 67.5% 8,776
2004 12,500 1,125 9,235 0.940 69% 1.256 1.16 63.6% 7,951
2005 13,250 1,050 9,500 0.920 66% 1.106 1.10 65.8% 8,720
2006 15,250 1,125 11,250 0.975 72% 1.029 1.05 73.4% 11,201
2007 17,650 1,265 9,575 1.520 78% 68.3% 13,697
Where premium rate factor for 2002 is (17,650/1265)/(11750/1150) = 1.366,

etc.
Page 13
Claims cost increase for 2002 is 1.055 = 1.276, etc.

2007 on level loss ratio is 75% × (1.276/1.366) = 70.1%, etc.
2007 IER = (11750 × 70.1% + 13000 × 67.5% + … + 15250 × 73.4%) /

(11750 + 13000 + …. + 15250)
= 68.3%
2007 Selected IBF Ultimate = 9575 + (1-(1/1.520))×(68.3%×17650) = 13,697.
Comments on Q7(iv): Slightly different answers could be achieved depending on

whether candidates used the selected ultimate losses or the selected ultimate loss ratio
on which to base the calculation, because of rounding differences. Marks were
awarded for either basis and also if premiums and claims were adjusted to same-year
values as the question did not specify what weighted premiums to use. Many
candidates failed to even attempt to adjust either the premiums or the claims for
premium increases or inflation, even though information on these was given in the
question. Some candidates also calculated a straight average rather than a weighted
average, as required in the question, when calculating the ELR.
(v) 142 × 1/ (1.05)DMT = 86

(1.05)DMT = 142/86
DMT × ln(1.05) = ln(142/86)
DMT = ln(142/86) / ln(1.05)
DMT = 10 years.
The DMT is very high. This is likely to be a liability class of business with
latent claims e.g. asbestos.
Comments on Q7(v): This part of the question was well answered by the majority of
candidates. Whilst most candidates recognised that this would be data from a
liability class not all candidates realised that a tail length in excess of 10 years would
suggest latent claims like asbestos. Some candidates mistakenly tried to calculate the
discounted mean term using paid development patterns even though there are no data
to do this.
(vi) Exposure-based methods but amount and quality of data are often insufficient
Paid and incurred survival ratios and IBNR to outstanding ratios.
These ratios all use figures that are estimated by the company, e.g. incurred,
reserves, IBNR, OS and will differ depending on reserving strength.
This also makes benchmark comparisons difficult, as each company will have
different reserving basis.
Paid survival ratios do at least have paid figures in the numerator (which are
not open to interpretation).
Comments on Q7(vi): Even though most candidates realised that this was a long-tail
book of business, with a small number of claims which would be potentially very large
and variable, they were still suggesting the application of chain ladder methods.
Only a few candidates scored marks on this part.
Page 14
EXAMINATION

booklet.
supervisor.
question paper.
1 You are the actuary for a small general insurance company with a low solvency
margin. The insurer is considering purchasing excess of loss reinsurance to reduce its
claim volatility and hence reduce its probability of ruin.
(i) Explain whether the proposed course of action will have the desired effect and
state the implications on the insurer’s profit. [4]
You model the probability of ruin at the end of one year according to a compound
Poisson process. According to your calculations the company’s profit load is 0.1 ≤ θ
≤ 0.2, the reinsurer’s profit load is 0.15 ≤ γ ≤ 0.25 and the Poisson parameter is
1,000 ≤ λ ≤ 1,500.
(ii) State, with explanation, which combination of parameters would be worst for
the insurer. [2]
[Total 6]
2 A general insurance company writes only personal lines business. It would like to
build a financial model to provide checks on the business and has asked for your
advice.
(i) List the requirements that a financial model should meet. [2]
For a particular class of business, you have been told that the individual claim
amounts have a Gamma distribution with parameters α = 12 and λ = 0.02. The claim
frequency has a Poisson distribution with λ = 0.25.
(ii) Calculate a risk premium for this class of business. [2]
You are aware that there is some uncertainty surrounding the bases for calculating
both the frequency and the severity and would like to give a range of possible values
for the risk premium rather than the point estimate you have already calculated. You
would like to add a margin of 5% of the standard deviation for the amounts and 2% of
the standard deviation for the frequency.
(iii) Calculate the acceptable range. [3]

[Total 7]
ST3 S2008—2
3 You are an actuary in a general insurance company that applies an excess to its motor
insurance business.
(i) Define the term “excess”. [1]
(ii) Give reasons why the insurance company would use an excess. [3]
The motor pricing manager has estimated that the distribution of accidental damage
ground up losses for the coming year will be approximately as follows:
Loss (£) Number of claims
0 3,500
1 to 50 1,750
51 to 100 750
101 to 150 250
151 to 200 250
201 to 500 500
501 to 1,500 3,000
1,501 to 2,500 1,500
2,501 to 7,500 500
The excess is currently £100.
(iii) Calculate the amount that the insurance company expects to pay out for
accidental damage claims next year. [2]
(iv) Recalculate the amount that the insurance company would expect to pay if
inflation causes ground up losses to double but the excess remains at £100,
assuming that the relative distribution of claims stays the same. [2]
(v) Given that the losses have exactly doubled, explain why the insurance
company’s liability has not doubled as well. [1]
[Total 9]
4 Describe the following characteristics of marine insurance:
(i) Benefits [2]

(ii) Insured perils [3]
(iii) Exposure measures [1]
(iv) Claim characteristics [4]
(v) Risk factors [4]
[Total 14]

5 You are an actuary at a medium-sized general insurance company. One of the
business units writes two annually renewable binding authorities, whereby
underwriting authority is delegated to a third party and underwriting risk is 100%
retained by the insurer. Contract A incepted on 1 January 2004 and Contract B
incepted on 1 April 2004. The business unit wants to analyse its profitability over the
past three years.
You have been given the following information for policies written on the 2004
binding authorities:
Contract A Contract B
Average premium £265 £1,560

Acquisition costs 40.0% 32.5%
You have also been given the following information on successive rate changes that
were applied to the base rates at each given date between 2004 and 2007:
1 May 2005 +1.0% +2.0%

1 November 2005 +2.0% +3.0%
1 March 2006 +3.0% +5.0%
1 October 2006 -2.0% +1.5%
1 June 2007 -4.0% +1.5%
1 September 2007 -5.0% +1.5%
The numbers of policies written on the 2004 to 2007 binding authorities were:
2004 50,000 10,000

2005 52,000 11,100
2006 53,040 11,432
2007 57,824 12,004
(i) Calculate the earned premium (net of DAC) for 2005, 2006 and 2007, stating
any assumptions that you make. [12]
You have been given the following calendar year incurred claim movement
information:
Current Year Prior Years Current Year Prior Years
2004 £6,025,000 £0 £5,000,000 £0

2005 £6,625,000 £750,000 £5,937,500 £1,500,000
2006 £5,126,000 £1,593,750 £5,475,000 £2,700,000
2007 £6,100,000 £1,875,000 £5,375,000 £2,025,000
ST3 S2008—4
Claims paid in each year were 35% of GWP for Contract A and 31% of GWP for
Contract B. Investment return was 4.5% per annum. Overheads allocated to the
business unit were £1.5m in 2004. Internal per policy expenses in 2004 were £24 for
Contract A and £78 for Contract B. Overheads and per policy expenses increased at
4.0% per annum.
(ii) Calculate the pre-tax profit made by the business unit in 2005, 2006 and 2007,
stating any additional assumptions that you make. [10]
(iii) Comment on the results of your findings in (i) and (ii), paying particular
attention to the impact of the two contracts on the overall results. [8]
[Total 30]
6 You are the pricing actuary for a general insurance company that writes private motor
insurance. You have been contacted by a national cycling club, which is interested in
selling bicycle insurance to its members.
(i) List the two key insured perils that would be covered in a bicycle insurance
policy. [1]
The cycling club wants their members to be able to obtain quotes for this insurance
very quickly and easily and therefore wants the number of rating factors to be
restricted to two.
(ii) State the most important requirements of rating factors. [1]
(iii) State two rating factors which could be used for this class of insurance. [1]
(iv) List the non-standard add-on covers that could be included to give a
potentially very comprehensive policy. [8]
Your company has not written bicycle insurance before and therefore has no data
relating to this. It has been suggested that you could use industry-wide data to help
you to set the prices.
(v) Explain the difficulties you might encounter in obtaining and using industry-
wide data. [4]
You have been asked to specify the future data requirements for this class of business
so that a suitable management information system can be considered.
(vi) List the data items for which you would ask in order to help you to develop
sophisticated pricing for this business in the future. [8]
[Total 23]

7 For a number of years a reinsurer has written a working layer per event risk XL treaty
with unlimited reinstatements. The cedant places this treaty to protect the liability
element of a large book of private motor vehicle insurance. The reinsurer has recently
introduced a stability clause and an aggregate deductible to the layer.
(i) Define each of these new features and explain the impact of their introduction
on the expected cost of claims to the layer. [5]
(ii) State the advantages and disadvantages to both the reinsurer and the cedant of
the addition of each of these new features to the layer. [6]
[Total 11]
END OF PAPER
ST3 S2008—6

EXAMINERS’ REPORT
September 2008
Introduction
R D Muckart
December 2008
Comments
Individual comments are shown after the solutions to each part question that follow.
1 (i) Greater risk is associated with greater variability, and hence a larger
probability of ruin
Purchasing excess of loss reinsurance will reduce claim volatility and
therefore reduce the probability of ruin.
The effect on the insurer depends on both the retention level and the
reinsurer’s profit load/cost of reinsurance
Consideration of different types of reinsurance (Individual/ Aggregate/
Catastrophe).
Layers of reinsurance may be needed to give full coverage / other comment
about layers.
Buying reinsurance is a trade off between profit and volatility – reinsurance
will reduce the claims volatility but it will also reduce the company’s expected
profit.
Buying reinsurance will give a more stable profit year on year
Reduced profits will worsen the solvency position but reduced claims
volatility will strengthen it.
Above a certain profit load it will be better for the insurer to retain all the risk
and below a certain profit load it will be better for the insurer to cede all the
risk.
Between these two profit loads the insurer needs to fix the retention level in
order to minimise the probability of ruin.
Therefore the insurer can set a retention level such that the probability of ruin
is less than it would be without reinsurance thus achieving the desired effect.
Comments on Q1(i): Generally well answered.
(ii) θ = 0.1, γ = 0.25, λ = 1500
Insurer’s profit margin is smallest, making less profit

Reinsurer’s profit margin is biggest, taking more profit from the insurer
For a given claim severity distribution the higher the value of λ, the higher the
expected number of claims and the variance.
Comments on Q1(ii): Generally well answered, but some candidates believed

incorrectly that the Poisson parameter does not impact the probability of ruin and
some even incorrectly gave λ = 1000 as the answer.
2 (i) Requirements:
Valid/appropriate
Complete/comprehensive
Adequately documented
Reflection of the risk profile of the classes of business being modelled
Parameter values should be accurate for the classes being modelled / fits the
data
Outputs and their degree of uncertainty should be capable of independent
verification
Outputs should be readily communicable/explainable/easy to understand
Should not be overly complex/take too long to run
Should not be too expensive to run
Flexible for the purpose
Page 2
Comments on Q2(i): Bookwork; most candidates scored well.
(ii) Severity: E(X) = α/λ = 12/0.02 = 600

Frequency: Ε(Y) = λ = 0.25
Risk premium = 600 × 0.25 = £150
Comments on Q2(ii): Nearly all candidates calculated the risk premium correctly.
(iii) Severity: Var(X) = α/(λ2) = 30,000

σ(X) = 173.205
so E(X) = 600 ± (0.05 × 173.205) = (591.34 , 608.66)
Frequency: Var(Y) = λ = 0.25

σ(Y) = 0.5
so E(Y) = 0.25 ± (0.02 × 0.5) = (0.24 , 0.26)
Total Range (£141.92, £158.25)
Comments on Q2(iii): This part involved calculating the ranges for the frequency
and severity separately and then combining by multiplying. Some students incorrectly
tried to add the frequency margin (a number) to the severity margin (as an amount)
before applying to the risk premium. No marks were lost by students who calculated
a correct one-sided upper range.
3 (i) The excess is the sum, specified in the policy,

that the insured must bear before any liability falls upon the insurer.
They are widely used in personal lines insurance and may be compulsory
(applying to all claims of the type specified) or voluntary (to secure lower
premiums).
Comments on Q3(i): Many candidates failed to define this term correctly.
(ii) Reasons to use an excess

To reduce the number of claims
To reduce the amount of each claim
To eliminate small claims, leading to reduced expenses
To encourage policyholders to reduce moral hazard,
leading to lower claims
To reduce the customer premium (voluntary excess),
leading to higher volumes and better retention rates
As an endorsement imposed by the insurer to limit their exposure to certain
perils
To encourage better risk management by customers
Comments on Q3(ii): Many candidates failed to pick up easy marks by generating

too few examples.
Page 3
(iii) Assume an average claim is at the mid-point of each range/claims are

distributed evenly within the range
(125 -100) × 250 + (175 -100) × 250 + (350 - 100) × 500

+ (1,000 - 100) × 3,000 + (2,000 - 100) × 1,500 + (5,000 - 100) × 500
= 6,250 + 18,750 + 125,000 + 2,700,000 + 2,850,000 + 2,450,000
= £8,150,000
Comments on Q3(iii): The calculations were straightforward but some students

failed to deduct the excess from the average loss. Equal credit was given for
candidates using average losses starting at 25.0 or 25.5.
(iv)
New Loss Range New Average Loss Number
0 0 3,500
1 to 100 50 1,750
101 to 200 150 750
201 to 300 250 250
301 to 400 350 250
401 to 1,000 700 500
1,001 to 3,000 2,000 3,000
3,001 to 5,000 4,000 1,500
5,001 to 15,000 10,000 500
(150 - 100) × 750 + (250 - 100) × 250 + (350 - 100) × 250 + (700 - 100) ×
500 + (2,000 - 100) × 3,000 + (4,000 - 100) × 1,500 + (10,000 - 100) × 500
= 37,500 + 37,500 + 62,500 + 300,000 + 5,700,000 + 5,850,000 + 4,950,000
= £16,937,500
Comments on Q3(iv): A simple question but some candidates were unable to apply
the excess to obtain the correct result irrespective of having calculated part (iii)
correctly.
(v) The number of claims hitting the excess is affected as well as the severity.
This is a leverage/gearing effect caused by the excess being fixed in £ terms.
If X is the loss and Y is the amount the insurer actually pays out then
(Y = X – 100) becomes (Y = 2X – 100), which is not double
Comments on Q3(v): An easy mark picked up by most candidates, even those who
did not calculate parts (ii) and (iii) correctly
Page 4
4 (i) Benefits
Marine Property
To indemnify the insured against the value of the loss or damage to the marine
hull (subject to limits or excesses).
Cover can also be for marine cargo and specie and marine freight.
Marine Liability
To indemnify the insured against a financial loss (subject to limits or
excesses). Associated legal expenses may also be covered.
Comments on Q4(i): This bookwork question was answered well by many candidates
but some failed to mention third party liability.
(ii) Insured Perils

Marine Property
Perils of the sea/other navigable waters eg. storm, tsunami
Fire
Explosion
Jettison
Theft of cargo
Spoilage and contamination of cargo
Piracy
Capsizing
Stranding
Collision (iceberg or other)
Actions of the sea (e.g. waves damaging vessel)
Running aground
Specie (valuables)
Marine Liability
Damage to 3rd party property
Injury to 3rd parties (including death)
Injury to employees (including death)
Errors and omissions
Comments on Q4(ii): Most candidates were only able to reproduce a short

“standard” list and failed to go beyond that.
(iii) Exposure Measures

Insured value of the hull/ship
Tonnage of hull/ship
Value of cargo
Limits of liability
Comments on Q4(iii): Answered well by most candidates.
Page 5
(iv) Claim Characteristics

Reporting delays: claims usually reported when the vessel reaches a major
port.
(May be only a very small delay if claim takes place in the port.)
Settlement delays: could be long,

especially if there is a dispute over legal liability or the amount that should
be paid.
Claim Amounts: variable. Relatively small amounts for hull damage to small
vessels; very large amounts for complete loss of a large vessel and its cargo.
Liability claims very variable; legal expenses element can dominate.
Claim Frequency: infrequent for hull but more frequent for cargo
Accumulations of risk are possible
e.g. geographical concentration (storm/tidal wave); spillage of hazardous
material
Moral hazard – frequency increases in bad economic conditions
Salvage and subrogation are often employed
Currency issues
Comments on Q4(iv): Strong candidates were able to make sensible comments

covering a range of claim characteristics of the main marine insurances. Weaker
candidates generated few points and made too many unqualified generalisations.
(v) Risk Factors

Hull
Level of cover / excesses and limits
Size/tonnage of vessel
Type of vessel
Condition of vessel
Age of vessel
Type of industry
Classification society
Engine type/manufacturer
Country of build
Experience of captain and crew
Detention history
Areas sailed in (rough seas/war zones etc.) / locations visited
Tonnage of hull
Previous claims experience of ship
Previous claims experience of owner
Insured value / sum insured
Cargo
Level of cover
Value of cargo
Nature of cargo
How packaged
Where stored on ship (deck versus hold)
Page 6
Trade terms
Trade routes taken
Standard property insurance risk factors apply when warehoused at port
Liability
Number of passengers and crew
Type of work undertaken by the insured (e.g. shipbuilder, marina operator)
Limits of liability
Comments on Q4(v): Few candidates were able to produce a sufficiently wide range
of factors to score well.
5 (i) General Assumptions:

Policies attach evenly over any relevant period of time to both Contracts
Risk is uniform for each policy and each policy lasts 12 months
New policies do not change the risk profile on either Contract (i.e. mix
constant)
so the average premium is not affected by new business or lapses and remains
constant other than for a rate change
Endorsements and cancellations are ignored
Rate changes applied to policies written on the day of the change
Reinsurance is ignored
Commission levels constant from 2004 to 2007
In order to answer (iii) properly, best to keep A and B figures separate as far
as possible
Calculate the monthly average premiums from 1st January 2004 to 31st
December 2007 by adjusting for the rate changes.
1 May 2005 £ 267.65 £ 1,591.20

1 November 2005 £ 273.00 £ 1,638.94
1 March 2006 £ 281.19 £ 1,720.88
1 October 2006 £ 275.57 £ 1,746.70
1 June 2007 £ 264.55 £ 1,772.90
1 September 2007 £ 251.32 £ 1,799.49
Method selected based on limited time available

… and expected to give similar answers to the more accurate “tranche”
method
Assumption: annual average premium × policies written is an appropriate
measure of written premium in an underwriting year
Assumption: can use overall proportions for earned/unearned premiums for
each year
Page 7
Take a simple average of the monthly average premiums in each calendar year
to give annual average premiums for each of the underwriting years:
2004 £ 265.00 £ 1,560.00

2005 £ 267.66 £ 1,588.76
2006 £ 278.42 £ 1,713.68
2007 £ 264.73 £ 1,770.84
From the definition of the “annual accounting basis”, it follows that the
policies attaching to Contract B after 31st December 2004 should not be
included in the 2004 underwriting year and hence should not be included in
the UPR carried forward from 2004 to 2005
Adjust Contract B for policies attaching to the ’04 binder, but that are written
in the ’05 underwriting year (i.e. 75% written in 2004; 25% written in 2005).
Repeat for ’05 to ’07. No adjustment required for Contract A.
Adjusted Policy Numbers:
2004 50,000 7,500

2005 52,000 10,825
2006 53,040 11,349
2007 57,824 11,861
Average Premium × Number of Policies Written = Written Premium
Written Premiums (000’s):
Contract A Contract B Total
2004 £ 13,250 £ 11,700 £ 24,950

2005 £ 13,918 £ 17,198 £ 31,117
2006 £ 14,768 £ 19,449 £ 34,216
2007 £ 15,308 £ 21,004 £ 36,312
Calculate the Acquisition Costs as 40% for A, 32.5% for B.
Calculate the UPR b/f for 2005 as 50% of the Contract A premium written in
2004 and 62.5% of the Contract B premium written in 2004. The 62.5%
comes from assuming earning using the 24ths method (i.e. 9/24 earned in ’04
and 15/24 earned in ’05).
Page 8
For all the subsequent underwriting years the earnings are assumed to be
50%/50%.
Acq UPR
GWP NWP UPR c/f NEP
Cost b/f
2004 13,250 5,300 7,950 0 3,975 3,975

2005 13,918 5,567 8,351 3,975 4,175 8,150
2006 14,768 5,907 8,861 4,175 4,430 8,606
2007 15,308 6,123 9,185 4,430 4,592 9,023
2004 11,700 3,803 7,898 0 4,936 2,962

2005 17,198 5,589 11,609 4,936 5,804 10,740
2006 19,449 6,321 13,128 5,804 6,564 12,368
2007 21,004 6,826 14,178 6,564 7,089 13,653
2004 24,950 9,103 15,848 0 8,911 6,937

2005 31,117 11,157 19,960 8,911 9,980 18,891
2006 34,216 12,228 21,988 9,980 10,994 20,974
2007 36,312 12,949 23,362 10,994 11,681 22,675
Comments on Q5(i): Most students correctly calculated the premiums following the
rate changes for each contract, although a small number of students incorrectly
applied each rate change to the 2004 premium, thereby failing to compound the
changes. Most candidates attempted to do the calculations by looking at each tranche
of rates separately. This was an accurate and valid approach but candidates using
this approach were unlikely to be able to finish the calculations in the time available.
Because of the limited time available, it was expected that students would split the
calculation by year, taking an average premium for each year, rather than by tranche.
The examiners gave equal credit to either method. Also, while it is recognised that
candidates do not have a spreadsheet available to perform the calculations setting out
the workings in the form of tables would have made the calculations easier to perform
and to carry forward to later parts of the question. Not many candidates did this
thereby making their workings more complex than necessary. While some candidates
were able to make the correct calculations for Contract A, most failed to adjust
correctly for Contract B. Some candidates did not appear to understand the concept
of deferred acquisition costs.
(ii) Additional Assumptions:

Ignore profit commission, tax and investment return on free assets
Ignore reinsurance
Incurred claims include IBNR/IBNER allowance
No AURR required
Sum the calendar year incurred claim amounts for each contract = incurred
claims
Calculate the paid claims in each calendar year (e.g. 35% × GWP for Contract
A)
Page 9
Outstanding claims reserve c/fwd = reserve b/fwd + incurred claims – paid

claims with reserve b/fwd for 2004 = 0
Per policy expenses in ’04 given. For each contract for each calendar year,
# policies written × per policy expenses × per policy expense inflation =
Expenses (per policy)
Allocated Overheads in ’04 given. Apply 4% inflation p.a. to get ’05 to ’07.
Split for A, B e.g. pro-rata to incurred claims
Calculate total expenses as Expenses (per policy) + Expenses (allocated

overheads)
Underwriting Result = Earned Premium (net of DAC) – Claims Incurred –

Total Expenses
Investment Return is taken as the average UPR b/fwd + Claim Reserve b/fwd
+ ½ of cash flow times the annual investment return of 4.5% where cash flow
is NWP less paid claims less expenses, or other reasonable formula
Insurance Result = Underwriting Result + Investment Return
Inc Paid
Inc Cyr Inc Pyr Res b/f Res c/f
Clms Claims
A
2004 6,025 0 6,025 4,638 0 1,388
2005 6,625 750 7,375 4,871 1,388 3,891
2006 5,126 1,594 6,720 5,169 3,891 5,442
2007 6,100 1,875 7,975 5,358 5,442 8,060
B
2004 5,000 0 5,000 3,627 0 1,373
2005 5,938 1,500 7,438 5,331 1,373 3,479
2006 5,475 2,700 8,175 6,029 3,479 5,625
2007 5,375 2,025 7,400 6,511 5,625 6,514
Total
2004 11,025 0 11,025 8,265 0 2,761
2005 12,563 2,250 14,813 10,203 2,761 7,370
2006 10,601 4,294 14,895 11,198 7,370 11,067
2007 11,475 3,900 15,375 11,869 11,067 14,573
Page 10
Per Pol
Per Pol Total
Exps O'heads
Exps Expenses
Tot
A
2004 24.00 1,200 820 2,020
2005 24.96 1,298 777 2,075
2006 25.96 1,377 732 2,109
2007 27.00 1,561 875 2,436
B
2004 78.00 585 680 1,265
2005 81.12 878 783 1,661
2006 84.36 957 890 1,848
2007 87.74 1,041 812 1,853
Total
2004 1,785 1,500 3,285
2005 2,176 1,560 3,736
2006 2,334 1,622 3,957
2007 2,602 1,687 4,289
U'wtg Inv Insurance

Result income Result
A
2004 -4,070 29 -4,041
2005 -1,299 273 -1,026
2006 -223 399 176
2007 -1,389 476 -913
B
2004 -3,304 68 -3,236
2005 1,641 388 2,029
2006 2,345 536 2,881
2007 4,400 679 5,079
Total
2004 -7,373 197 -7,176
2005 342 661 1,003
2006 2,123 935 3,057
2007 3,011 1,155 4,166
Comments on Q5(ii): Even if students had not managed to complete part (i), it was
possible to pick up quite a lot of marks in part (ii), and some students did this. Even
without premium figures, claims and expense figures could be calculated and
candidates also gained marks for formulae and workings even where the final figures
were incorrect. Very few students calculated claims reserve figures or were able to
calculate profit.
Page 11
(iii) Comments:
Number of policies, average premiums and total premium
Contract A average premiums rose in 2005 and 2006, but fell back to 2004
levels in 2007 following a series of rate reductions.
Growth in Contract A policy numbers was most prolific as rates decreased
Average premium decreases are likely to have been part of a growth strategy
(because if the market had been decreasing rates to the same extent, volumes
would not have grown at that rate)
Contract B average premiums rose from 2005 to 2007 and growth in policy
numbers was also strong over the same period.
So the market may have been putting rates up at a faster pace than the
increases on Contract B or Contract B rates started too low
In 2004 the GWP of Contract A was slightly higher than that of Contract B.
By 2007 the GWP of Contract B was dominating the mix.
As the acquisition costs on Contract A policies (40%) are higher than on
Contract B policies (32.5%), Contract B’s contribution the total Earned
Premium (net of DAC) is greater than its contribution to the GWP
Claims
The total claims ratio (Incurred Claims/Earned Premium (net of DAC)) has
been improving slightly year on year from 2005 to 2007.
Contract A claims ratio has deteriorated over that period, due probably to
falling premium rates and rising claims costs.
Contract B claims ratio consistently improves from 2005 to 2007 due, in part,
to consistent rate increases over the period.
At 2007 the Contract B claims ratio may suggest that further upside potential
is limited.
So the total claims ratio improvement is being driven by the claims ratio
improvement in Contract B and its relative size in (net earned) premium terms
compared to Contract A
Expenses
The total expense ratio (Total Expenses/Earned Premium (net of DAC))
improved from 2005 to 2007.
Contract B has a much better expense ratio than Contract A.
The improvement in Contract B's expense ratio from 2005 to 2007 has been
cancelled out largely by the deterioration in Contract A's expense ratio.
Contract B total per policy expenses are lower than Contract A total per policy
expenses so Contract B has a better per policy expense ratio.
Overheads are falling as a proportion of premium as the annual premium
growth is well ahead of inflation in overheads.
Performance
The total combined ratio was less than 100% from 2005 to 2007, so the
business unit is making an underwriting profit
The profit margin (Insurance Result/Earned Premium (net of DAC)) has
improved from 2005 to 2007. The contribution to the insurance profit from
Page 12
investment income decreased from 2005 to 2007 as the underwriting results

improved.
Contract B produced most of the underwriting profit in 2006 and 2007 and
significantly outperformed Contract A, so Contract B is cross-subsidising
Contract A.
This cross-subsidy may be deliberate on the part of the business unit e.g. the
unit writes both with the same third party administrator, so losing one contract
means losing the other.
If the trend in Contract A's underwriting result it will keep upward pressure on
the total underwriting result and will increasingly draw on subsidy from
Contract B.
Contract A requires rate increases/performance review/renegotiation of
commission terms if it is to be brought into profitability.
Past profitability of Contract B suggests that rates could be dropped to
increase volumes while still remaining profitable.
However, may become increasingly difficult to grow Contract B volumes
while Contract A is underperforming as Contract B is cross-subsidising
Contract A.
Both contracts produce similar investment income ratios (Investment
Income/NEP(net of DAC)).
The investment income ratio improved from 2005 to 2007 as there was more
cash available for investment in each successive year.
If Contract A had not been written, then the total profit would have been
higher over the three year period. (Contract A only made a small positive
contribution to the insurance profit in 2006. In 2005 and 2007 it was a
significant drain.)
Comments on Q5(iii): Very few candidates recognised that marks could be gained
for sensible comments made without having completed part (i) or part (ii). Those who
did, tended to score best on this question.
6 (i) 2 key insured perils:

Accidental Damage
Theft (from insured location/vehicle/within territorial limits)
Comments on Q6(i): Most candidates recognised theft as being the key peril. Some
candidates did not read the question and listed more than two perils.
(ii) Rating factor characteristics:

Practical (objectively measurable)
Relate to the intensity of the risk/define the risk
Do not correlate too closely with the other rating factors
Comments on Q6(ii): Surprisingly, the characteristic most frequently ignored by

candidates was that a rating factor should reflect the intensity of the risk. Without
that, the factor could represent anything.
Page 13
(iii) Rating factors:

Value of bicycle
Postcode / Age of policyholder / Use (e.g. business/pleasure) / frequency of
use
Comments on Q6(iii): Again, some candidates did not read the question and offered
more than two rating factors.
(iv) Non-standard add-on covers:

Public liability – death or bodily injury to 3rd party
Public liability – damage to property belonging to 3rd party
Worldwide cover (rather than limited to home country)
Cover while bicycle is away from the home (not always included when a bike
is covered under a Household policy)
Damage while in transit e.g. on a bike rack on a car, on an aeroplane
Hire of a bicycle while being repaired/replaced
Personal accident
Roadside recovery
E.g. to repair a puncture/after falling off and damaging bike
Cover while racing
Replacement of helmet if you fall off, even if damage is not obvious
New for Old cover on bikes up to, say, 3 years old
Cover for spare parts & accessories e.g. spare wheels not attached to the bike
(if not covered by Household Contents policy)
Non-standard bicycles
E.g. tandem/unicycle/electric bike
Extension to other members of family / other named drivers
Cover while child in seat on back
Extended warranty
Legal protection & assistance following an accident
Malicious damage/vandalism
Comments on Q6(iv): Poorly answered, with only the strongest candidates offering a
range of sensible add-on covers that reflected the nature and cost of the standard
insurance policy. Some candidates suggested add-on covers that would not be
insurable or would have had disproportionately large premiums when compared to
those typical of the standard insurance policy.
(v) Difficulties with data:

The data supplied by different companies may not be comparable:
different geographical section of market
different socio-economic section of market
differences in cover (exclusions, excesses, policy conditions)
differences in underwriting practice
differences in claim settlement practice
differences in nature of data stored by different companies
different coding used for risk factors/rating factors
Industry-wide data may not even exist for this class
Data may not be detailed enough for pricing
Data may be out-of-date (takes time to collect, collate & distribute)
Page 14
Data quality may be poor (depends on quality of data of all contributors)

Not all companies contribute so may be unrepresentative
Comments on Q6(v): Although this was a fairly standard question, many candidates
failed to cover the range of points needed.
(vi) Data: Policy

Postcode
Make/Model of bike
Age of bike/year of make
Value of bike
Excess chosen (if there is an option)
Age of policyholder
Cover level (UK/Worldwide; leisure/racing; etc.) including use of bike
Estimated annual mileage
Add-ons chosen
Policy number
Dates – inception/renewal/mid-term change
Premium amount
Payment frequency (annual/monthly)
Payment method (DD/credit card)
Unique link between policy and claim, for matching
Unique link between this policy and other products e.g. customer number
Claims
Incident date, reported date, settlement date
Claim number
Cause of loss
Paid date for all claims payments
Claim amount paid, by peril
Claim amount outstanding, by peril
Claims handling expenses
Claim status
Comments on Q6(vi): Many candidates were able to suggest a wide range of data
items that could reasonably feature in the management information system.
7 (i) Aggregate Deductible

Introduction of the aggregate deductible means that now the sum of the claims
to the layer must exceed the deductible before the cedant can make a recovery
so for a given amount of exposure, expect the aggregate deductible to reduce
the cedant’s expected recovery and increase the cedant’s retention
The extent of the impact of the aggregate deductible depends on:

the size of the aggregate deductible (for a given exposure in vehicle years)
the expected number and severity of losses to the layer (for a given exposure
in vehicle years)
e.g. large aggregate deductible relative to expected number/size of losses
means lower recoveries for the cedant (and vice versa for a small aggregate
deductible)
Page 15
Stability Clause
Before the stability clause applied, the expected amount of total losses to the
layer would have increased annually (all else being equal) because of:
the effect of TPBI inflation on severity of individual losses to the layer (i.e. the
conditional expected value of a loss to the layer increases with inflation)
and the gearing effect of TPBI inflation increasing the frequency of losses to
the layer (i.e. probability of a loss to the layer increases with inflation)
A stability clause means the attachment point and layer limit are adjusted in
line with some specified index (e.g. fixed x% p.a. or a healthcare cost index)
with the intention of maintaining real values to the layer
so the layer widens with each application of the index
e.g. £1m xs £1m indexed by 2% is £1.02m xs £1.02m
Adding the stability clause has the following expected impact

The frequency of losses to the layer may drop over time e.g. a claim that starts
in the layer may settle below the layer
For a given loss, its actual attachment point depends on the settlement date
(i.e. the attachment point will increase in line with the stability clause index
until the loss settles)
If the deductible is small relative to the expected claims cost without the
deductible, the expected claims cost to the layer is simply the cost without
deductible less the deductible amount.
Whereas if the deductible is relatively large then a straight deduction is not
correct and claims to the layer can only be estimated using a distribution and
probabilities.
The actual impact of the stability clause depends on the cedant’s actual claims
experience and on the inflation in TPBI claims relative to the index applied to
the layer i.e. inflation could be different to the assumed indexation.
Comments on Q7(i): Most candidates were able to define a stability clause and
explain the effect this would have. Fewer were able to explain the aggregate
deductible.
(ii) Reinsurer
+ stability clause ensures alignment of interest by encouraging faster claims
settlement (as net retention increases with each year due to the indexation of
the attachment point and limit),
+ stability clause gives some protection against expected future inflation in the
claims to the layer
+ aggregate deductible reduces exposure to the cedant and allows the reinsurer
to use capital elsewhere
+ benefits if the sum of claims to the layer doesn’t breach the aggregate
deductible or claims settle below the indexed attachment point
– actual claims inflation may outstrip the indexation thereby eroding the
benefit of the stability clause over time (likely in practice)
– potential increase in expenses for setting up and managing more complex
contracts
– lower premium income with introduction of aggregate deductible
– more volatility in claims cost to the layer relative to the premium charged
Page 16
Cedant
+ the aggregate deductible reduces reinsurance spend (especially beneficial if
reinsurance rates are hard)
+ can use the aggregate deductible to manage risk appetite
+ the aggregate deductible means higher expected profit as ceding less to the
reinsurer generally means ceding less profit
+ cedant can manage total exposure to the reinsurer (reinsurer security impacts
capital requirement)
+ cedant may be able to negotiate a lower premium because of the stability
clause
– aggregate deductible delays recoveries (cashflow implications)

– greater loss retention, so alternative source(s) of capital required
(alternatives may be more costly).
– greater volatility in the retained losses
– retains some inflation risk i.e. if the TPBI inflation is lower than the
indexation, then more likely that a claim estimated to settle in the layer settles
below the layer
Comments on Q7(ii): Any advantages and disadvantages given in part (i) were given
credit under part (ii). In general, candidates failed to think of enough valid
advantages and disadvantages to score highly in part (ii).
Page 17
EXAMINATION
23 April 2009 (pm)

booklet.
2. You have 15 minutes before the start of the examination in which to read the
questions. You are strongly encouraged to use this time for reading only, but notes
may be made. You then have three hours to complete the paper.
supervisor.
5. Attempt all six questions, beginning your answer to each question on a separate sheet.
question paper.
1 List regulatory restrictions that could be applied to a general insurance market, briefly
explaining the reason for each. [12]
2 A general insurance company selling personal lines insurance policies is considering

removing the excess from future policies and replacing it with a deductible of the
same magnitude.
(i) Explain the terms “excess” and “deductible”, highlighting the main difference
between them, and giving an example to support your answer. [4]
(ii) Discuss the effects that this change could have on the company’s future
operations and profits. [4]
(iii) Describe how the company should model the potential impact of this change
on profit. [4]
[Total 12]
3 The following data are provided for a general insurance class of business:
• paid claim amounts by accident year and development period

• incurred claim amounts by accident year and development period
• settled claim numbers by accident year and development period
• reported claim numbers by accident year and development period
• ultimate premium data by accident year
(i) Give examples of the checks that you would carry out during a reserving
exercise, explaining what each of these would show. [5]
(ii) Suggest, with examples, reasons why the actual run-off of outstanding claims
may at times be higher than the estimated amounts when monitoring the
claims reserve figures for mortgage indemnity business. [7]
(iii) List other reasons why a general insurance company would analyse claims
data. [4]
[Total 16]
ST3 A2009—2
4 A general insurance company arranges a layer of excess of loss reinsurance for £4.0m
excess of £1.0m for individual claims. There is also an annual aggregate deductible of
£4.0m.
The reinsurer charges an initial premium of £1.0m with a first reinstatement premium
of 120% of initial premium, a second reinstatement premium of 150% of initial
premium, and no further reinstatements.
The actual claims in the year are shown below in the order that they occurred:
Claim Claim size

1 £2.5m
2 £3.5m
3 £6.0m
4 £4.2m
5 £1.8m
6 £3.0m
7 £5.0m
(i) Calculate the claim amounts paid by the reinsurer and the reinstatement
premiums payable after each claim. [6]
(ii) Define aggregate excess of loss reinsurance. [1]
(iii) Explain how stop loss reinsurance is different from aggregate excess of loss
reinsurance. [3]
(iv) Explain why stop loss cover is generally quoted in terms of claim ratios rather
than monetary amounts. [1]
(v) Suggest possible risks relating to the direct writer’s operations that a reinsurer
would face if stop loss cover were made available. [3]
(vi) Describe the conditions that a reinsurer providing stop loss cover is likely to
impose on the business covered. [1]
[Total 15]

5 An individual claim amounts distribution, F(x), is a discrete distribution on the
positive integers, and the number of claims, N, has a binomial, Poisson or negative
binomial distribution. The probability functions for individual claim amounts (Xi) and
aggregate claim amounts (S), respectively, are:
fk = P(Xi = k) k = 1, 2, 3, ….
gk = P(S = k) k = 0, 1, 2, ….
Constants a and b are such that the distribution of N, P(N = r) = pr satisfies the
following equation:
pr = (a + b/r) pr −1 for r = 1, 2, 3, ...
and
⎡ n ⎤
E ⎢ X1
⎢⎣
∑ i ⎥⎥ = r / n
X = r
i =1 ⎦
E ⎡ X 1 ∑ X i = r ⎤ = ∑ jf j . f r(−nj−1)* / f r n*
m r
⎢⎣ i =1 ⎥⎦ j =1
r −1
pn f rn* = ∑ (a + bj / r ) f j pn−1 f r(−n−j 1)*
j =1
(i) Derive a recursion formula for the aggregate claim distribution G(x). [6]
The number of claims follows a Poisson distribution with mean 125. The claims are
assumed to be random variables, independent of each other and independent of N,
with a Pareto distribution with mean £750 and standard deviation £1,750.
The PDF for the Pareto distribution is:
αλ α
f ( x) = x>0
(λ + x)α+1
(ii) Show that E[S] = £82,523, given that S = X1+ X2+ X3 +…+ XN , if a policy
excess of £100 is introduced. [6]
(iii) Calculate suitable values for the parameters of a translated gamma distribution
to approximate the distribution of S given that the standard deviation is
£23,480 and coefficient of skewness of S is 0.6838. [3]
[Total 15]
ST3 A2009—4
6 A country has experienced a significant increase in insurance premium rates in recent
years. The regulators believe that the major factors contributing to this are a
significant increase in the level of bodily injury court awards as well as the proportion
of claims cost spent on legal expenses. Although solicitors only charge a fee if the
case is successful, fees can account for up to 40% of damages awarded.
As a result, the regulators are planning to introduce a Bodily Injuries Compensation

Authority (BICA), with the principal aim of reducing premium rates, which will
operate as follows:
• Disputed bodily injury claims less than two years old will automatically be
referred to the BICA before they are allowed to be processed through the court
system.
• Any disputed bodily injury claims not notified within two years of injury will be
referred to the court system.
• The claimant will be charged a small fixed fee, regardless of the outcome of the
claim.
• General insurance companies will be charged a fee in proportion to their market

share based on total GWPI to cover the additional running expenses of the BICA.
• The initial application will be made by post on a standard form completed by the
claimant’s doctor.
• The claimant will not meet the BICA staff in person and therefore legal
representation will not be possible.
• Damages awarded by the BICA will be based on standard amounts by type of

injury. A single amount will cover pain and suffering, medical expenses and loss
of earnings.
• Any associated property damage claims will not be processed by the BICA.
• The claimant will be notified of the damages awarded by post.
• The BICA will aim to settle claims within 90 days of notification.
• If the claimants are dissatisfied with the level of compensation, they may reject
the award and continue with the claim through the court system.
(i) Discuss the advantages and disadvantages to the claimant of this system of
claims resolution. [7]
(ii) Describe the adjustments that a private motor pricing actuary would need to
make to the existing premium calculation assumptions and methodology as a
result of the introduction of the BICA. [13]
(iii) Suggest improvements that could be made to the proposed system, giving
reasons for these. [10]
[Total 30]
END OF PAPER
ST3 A2009—5

EXAMINERS’ REPORT
April 2009
Introduction
R D Muckart
July 2009
Comments
Individual comments are shown after the solutions to each part question that follows.
GL 3.7.09 © Institute of Actuaries
1
• Restrictions on the type/ amount of business a general insurance company can
write / classes of business it is authorised to write.
Ensures companies have appropriate expertise/ sufficient capital to write the
business classes
• Initial authorisation of new insurance companies..
Ensures companies have appropriate expertise / sufficient capital to write the
business classes
• Limits on premium rates that can be charged.
Ensures premium rates are sufficient to meet future claims/ ensure
policyholders not overcharged
• Restrictions on information that may be used in underwriting and premium rating.
For ethical / anti-discrimination reasons.
• The requirement to deposit assets to back claims reserves.
To ensure the company has sufficient funds to pay claims.
• The requirement to maintain a minimum level of solvency.
To ensure if claims are significantly worse than expected the company will
still remain solvent.
• Restriction on the type or amount of certain assets allowed to demonstrate
solvency.
To prevent high-risk assets from backing liabilities.
• Restrictions on the currency, domicile and duration of assets allowed to
demonstrate solvency (or mismatching reserves).
To ensure that assets match liabilities by term and currency so that short term
changes in exchange rates will not have an impact on solvency margins.
• The use of prescribed bases to calculate premiums, asset values and liabilities to
demonstrate solvency.
To ensure accurate estimates of liabilities and uncertainty.
• Licensing agents to sell insurance and requirements on the method of sale.
To ensure company has necessary expertise and that insured is well informed.
• The requirement for risk-based capital calculations & ICA analyses.
To ensure accurate estimates of liabilities and uncertainty.
• Requirement to pay levies to consumer protection bodies.
To protect policyholders and maintain faith in insurance market.
• Legislation to protect policyholders should general insurance companies fail, e.g.
Financial Services Compensation Scheme.
To protect policyholders and maintain faith in insurance market.
• Cooling off period, e.g. fourteen day cancellation rules on policies issued.
To protect policyholders and promote confidence in the industry.
• Regulations with respect to treating customers fairly.
To protect policyholders and promote confidence in the industry.
• Restriction on countries a general insurance company can write business in.
Prevents exposure to volatile risks and unfamiliar legal systems and
regulations.
• Restrictions with respect to anti-competitive behaviour
Prevents formation of cartels, concentration of risk, and protects
policyholders.
• Requirement to file / publish premium rates before they can be used.
Prevents anti-competitive practices and therefore protects policyholders.
Page 2
• Mandatory restrictions on cover e.g. no deductible on EL

To protect policyholders and claimants and to ensure consistency of cover.
• Requirements to offer cover e.g. even in high-risk flood areas / motor 3rd party
liability.
For social responsibility and helps economy as a whole.
• Statutory requirement to offer certain cover e.g. EL & Motor 3rd Party Liability.
For social responsibility and helps economy as a whole.
• Disclosure / transparency of reporting requirements
To help regulators, investors, capital providers and policyholders assess the
soundness of the company.
• Requirement for a Statement of Actuarial Opinion to be produced by an approved
actuary.
Promotes confidence in the level of reserves and helps to prevent the failure of
a general insurance company.
• Requirements for management to be fit and proper.
Promotes confidence in the industry and helps prevent fraud.
• Restriction on the type of reinsurance that may be used.
To prevent exposure to risky reinsurers or reinsurance products.
• Restriction on discounting of liabilities and discounting rates that can be used
To ensure consistency and that reserves are sufficient.
• Prohibiting illegal products from being sold, e.g. kidnap insurance.
To discourage illegal practices.
• Requirement for general insurance companies to be audited.
To give regulators and investors confidence in the company and to prevent
fraud.

Comments on Q1: This question was reasonably well answered by the majority of
candidates but some students did not give enough distinct points or specific enough
reasons.
2 (i) Deductible
• Amount deductible from claim amount, payable by policyholder.
• Sum insured = S. Deductible = D. Loss = L.
• If L > S insured pays L – S + D, insurer pays S – D.
• If L < S insured pays D and insurer pays L – D.
Excess
• Sum specified by policy which insured must bear before any liability falls
on insurer.
• Sum insured = S. Excess = E. Loss = L.
• If L > S + E, insured pays L – S and insurer pays S.
• If L > E but L < S + E insured pays E and insurer pays L – E.
• Therefore effect differs in the case where L > S.

• The primary difference is that the deductible eats into the sum insured
whereas the excess sits below the sum insured.
• Hence for a policy with a deductible the maximum the insurer will be
liable to pay is the sum insured less the deductible.
Page 3
Example (this has to be realistic).

• sum insured of £30,000.
• Policy A has an excess of £5,000, policy B has deductible of £5,000
• Each policy experiences loss of £40,000
• Policy A: loss to insurer = sum insured = £30,000, loss to insured =
£40,000 − £30,000 = £10,000
• Policy B: loss to insurer = sum insured – deductible = £30,000 – £5,000 =
£25,000, loss to insured = £40,000 − £25,000 = £15,000.
An explanation without using L, S, D etc. which is correct is acceptable.
Comments on Q2(i): This was well answered by the majority of candidates.

However, some candidates did not understand how a deductible worked with some
getting the excess and deductible the wrong way round. Some students did not give an
example even though this was specifically requested in the question.
(ii) Replace excess with deductible

• Effect of change will be small as this is personal lines business.
• Any valid reason explaining why change will be small
• All else being equal insurers profits will increase as claims cost will
decrease so profit per policy will increase
• Depends on change in policy volumes
• Option of reducing premium rates to compensate for potential loss of
volume.
• Policyholders may not realise that there is a difference
• Although will be obligation to clearly explain policy changes to clients
(treating customers fairly)
• Policyholders may realise that there is a difference and demand a reduced
premium.
• Possible change in mix of business as a result.
• Effect depends on mix of renewal / new business.
• Also depends on factors such as advertising and distribution channels.
• Significant increase in costs associated with communicating changes to
brokers and policyholders, and updating policy documentation.
• Increase in costs associated with system changes & retraining of staff.
• Expenses will most likely increase by more than any benefits gained.
Comments on Q2(ii): Better candidates picked up on the key points for this part.
Candidates that did not understand the term deductible failed to answer this part of
the question correctly. Also, many students did not comment that the claim cost
impact would be small, particularly since it was personal lines business, and some
students did not consider the implications for expenses.
(iii) Model
• Set up the model to give total profit as an output
• Review company data on past changes in excess and effect on policy
volumes and profit
• Review past company data on any deductibles (if available) otherwise
review industry data on effects of deductibles on policy volumes and
profitability
Page 4
• Allow for other factors such as distribution channels, advertising

• Allow for any changes to mix of business e.g. age, social economic group
• Allow for level of expenses, both fixed and variable
• Allow for possible changes in premium rates
• Allow for correlations between excess / deductible and other rating factor
• Fit Generalised Linear Model to model outputs.
• Alternatively , fit a stochastic / deterministic model.
• Use results to maximise profits
• Sensitivity test model - test the goodness of fit.
• Set up so that factors can be easily interpreted and monitored
Comments on Q2(iii): Most candidates struggled to answer this part well with many
candidates talking about “choosing a base period and modifying the data”, rather
than tailoring the answer to the question and discussing what specific modifications
to the data would be needed and what the actual output of the model should be.
3 (i)
• Check consistent with the data from the previous development period — to
check for any errors, e.g. compare diagonals.
• Check any anomalies / large movements in the data. (If large claims etc.
are mentioned it must be because they have seen anomalies in the data)
• Cumulative paid to incurred ratios — to test for a change in the case
reserving strength or claims settlement practise. Look at the ratios by
accident and development period.
• IBNR to outstanding ratios — to test for a change in the reporting process.
Look at the ratios by accident and development period.
• Average outstanding case estimate — again to test for a change in the
strength of case reserves. Look at the ratios by accident and development
period.
• Trends in ultimate loss ratios by accident year — for consideration against
market benchmarks and knowledge of the underlying market conditions.
• Settlement claims divided by reported claims — to check for changes in
the claims handling process.
• Average cost per claim — as a sense check for reasonableness.
• Claim frequency — again for consideration against market benchmarks
and knowledge of the underlying market conditions.
• Check development patterns by accident year — to check consistency
between accident years and compared to the prior analysis
• Sense checks e.g. the relationship between paid and incurred is as you
would expect
• Compare results of different projection methods to see if the results are
consistent
• Survival ratios — if class of business includes latent claims.
• Check actual versus expected for the latest development period to identify
any large movements / changes in reserving strength
• any other reasonable diagnostics, based on the available data given, with
explanation, are acceptable
Page 5
Comments on Q3(i): This was poorly answered by the majority of candidates. Better
candidates mentioned points including: checking for errors and unusual
developments but most candidates did not consider any diagnostic checks (looking at
specific ratios) or benchmarking results as a sense check. Some candidates gave
answers that required additional data although the question specifically stated what
data were provided.
(ii)
• Unusually heavy experience e.g. increase in central bank base rates leading
to increase in number of claims
• Random fluctuations / volatility.
• Large or exceptional claims e.g. factory closing down leading to local
unemployment with one regional mortgage lender leading to increase in
number of claims
• Trends in claims experience numbers due to economic factors e.g. rising
unemployment/falling house prices leading to increased number of claims.
• Trends in claims experience amounts due to economic factors e.g. increase
in the level of repossessions leading to higher costs for the insurer.
• Changes in risk. e.g. switch to endowment mortgages from repayment
mortgages implying higher claim severity.
• Changes in cover e.g. increase in amount by which mortgage exceeds
normal advance.
• Unexpected changes in law e.g. introduction of new government
legislation stating that insurers must still pay out when the lender has
failed to apply any required underwriting.
• Quality and amount of reinsurance cover may have varied e.g. reinsurance
review.
• Inadequate claims reserving process e.g. inappropriate assumptions tail
selection, future inflation may differ to that assumed
• Change in mix of business e.g. policies sold to people with low job
stability
• MIG business has a strange risk profile (uneven with a long tail) that will
need to be taken into account. Therefore very difficult to reserve this class.
• Systematic fraud e.g. developers artificially inflating official market prices
through the use of incentives, mis-selling through increasing prevalence of
self-certification loans.
• Payment patterns may differ e.g. due to system changes
• or settlement patterns may differ e.g. postal strike.
Comments on Q3(ii): Better candidates understood the impact of the economy on

this class of business but could have generated more points. Some candidates did not
appear to understand what the question was asking for.
(iii)
• monitoring the adequacy and use of reinsurance
• comparing the relative profitability of various parts of the account
• reviewing present premium rates
• pricing new or amended products
• determining rating factors
Page 6
• experience rating
• financial planning and management information
• Assessment and allocation of capital
• monitoring the insurer’s asset/liability position
• giving investment advice based on the liability profile
• monitoring actual versus expected for both claims and premiums
• for statutory accounts
• establishing the need for other reserves e.g. URR
• identification of unexpected claims sources to tighten up policy wording /
testing efficiency of the claims management teams and loss adjustors
• plus any other sensible suggestions
Comments on Q3(iii): This was bookwork and was well answered by the majority of
candidates.
4 (i)
Claim Claim size Claim size in layer Reinsurer pays RI Premium
1 £2.5m £1.5m £0m £0

2 £3.5m £2.5m £0m £0
3 £6.0m £4.0m £4.0m £1.2m
4 £4.2m £3.2m £3.2m £1.2m
5 £1.8m £0.8m £0.8m £0.3m
6 £3.0m £2.0m £2.0m £0m
7 £5.0m £4.0m £2.0m £0m
Total £26m £18m £12m £2.7m
Comments on Q4(i): Many candidates answered this completely correctly with other
candidates making minor mistakes when calculating the reinstatement premiums.
Some candidates however misunderstood the impact of the annual aggregate
deductible with a few mistakenly assuming this was aggregate excess of loss even
though individual claims cover is specified.
(ii) A form of excess of loss reinsurance that covers the aggregate of losses, above
an excess point and subject to an upper limit,
Sustained from a single event or from a defined peril (or perils) over a defined
period.
Comments on Q4(ii): This is bookwork and was reasonably well answered by the
majority of candidates.
(iii)
• Aggregate XL only provides cover for either one peril over the year or one
event. Stop loss extends cover to all claims, arising from all perils or all
events, in a class or classes over the defined period
• Stop loss is broader in that it covers not only catastrophe events but also
unforeseen accumulations of losses
Page 7
• Stop Loss reinsurance directly relates to the primary insurer’s underwriting

results
• In general, Stop Loss reinsurance is difficult to obtain and is usually
limited to only a few lines of insurance
Comments on Q4(iii): This is bookwork and most candidates did understand the
difference between the two types of reinsurance although many students did not give
enough differences between the two.
(iv) By using claims ratios the limits (and premiums charged) rise in proportion to
the amount of business written by the direct writer.
If this was not the case then the direct writer could, after taking out cover,
write lots more business by cutting premiums and hence trigger the stop loss
limits in that way.
Comments on Q4(iv): Most candidates got this point. However, some candidates did
not understand why a reinsurer would not use monetary amounts.
(v)
• Direct writer’s premium generally under-priced (e.g. a competitive market)
• Poor underwriting by direct writer
• Poor premium rating structure leading to adverse selection
• Unusually heavy claims experience (e.g. catastrophes and/or large claims)
• Generally adverse claims experience (i.e. random event)
• Poor claims handling control particularly after deductible reached i.e.
moral hazard
• Risk insurer targets more volatile but profitable business for higher
expected returns once downside risk is removed.
Comments on Q4(v): Most candidates mentioned poor underwriting and claims

handling with better candidates generating more points.
(vi)
• Only reinsure a proportion of the risk e.g. 90% so that the insured retains
an interest in the risk (can be described as a participation clause / co-
insurance / deductible if used in the correct context)
• Ensure stop loss cover has an upper limit
• Maintain some control over underwriting and claims
Comments on Q4(vi): This was reasonably well answered with most candidates
suggesting that the insurer retains some interest in the risk via some sort of
participation clause/deducible/co-insurance.
Page 8
5 (i) g0 = p0
This holds as the minimum claim size is 1
∞
gr = ∑ pn frn*
n =1
∞
= p1 f r + ∑ pn +1 f r( n +1)*
n =1
∞ r −1
= (a + b) p0 f r + ∑∑ (a + bj / r ) f j pn f rn−*j
n =1 j =1
r −1 ∞
= ( a + b) g 0 f r + ∑ (a + bj / r ) f j ∑ p n f rn−*j
j =1 n =1
r −1
= (a + b) g0 f r + ∑ (a + bj / r ) f j g r − j
j =1
r
= ∑ (a + bj / r ) f j gr − j
j =1
Above is slightly different from core reading where equations incorrectly contain
(a + bj) / r) not (a + bj / r)
Comments on Q5(i): This is bookwork but was poorly answered by the majority of
candidates.
(ii) E[S] = E[N] E[X’] where
X’= 0 X ≤ 100
= X − 100 X > 100
E(X) = λ / (α − 1) = 750
V(X) = αλ2 / (α − 1)2 (α − 2) = 17502
∴α / (α − 2) = V(X) / E(X)2 = 17502 / 7502 = 5.444
∴α = 5.444 (α − 2)
∴α = 10.889 / 4.444 = 2.45
∴λ = 750 × 1.45 = 1087.5
∞ ( x − 100)
E ( X ′) = ∫ α+1
αλ α dx
100 (λ + x)
∞ ⎛ 1 100 + λ ⎞⎟
= αλα ∫ ⎜ − dx
100 ⎜ (λ + x )α λ + x α +1 ⎟
⎝ ( ) ⎠
∞
α⎡ 100 + λ ⎤
−1
= αλ ⎢ α−1
+ α ⎥
⎣ (λ + x) (α − 1) (λ + x) α ⎦100
Page 9
⎛ 1 1 ⎞
= αλ α ⎜⎜ α−1
− α−1 ⎟
⎟
⎝ (λ + 100) (α − 1) (λ + 100) α ⎠
λα
=
(λ + 100)α−1 (α − 1)
= 660.18
Note 2nd line of derivation of E(X’) was to simplify integration by splitting

numerator in 1st line as (λ+x) – (100+λ). The alternative is to do integration
by parts which is an equally valid approach, as follows:
∞ ( x − 100)
E ( X ′) = ∫ α+1
αλ α dx
100 (λ + x)
⎛⎡ ∞ ∞
α⎜ x ⎤ 1 ∞ ⎡ 1 ⎤ ⎞⎟
∫
1 1 1
= αλ ⎜ ⎢− . ⎥ + dx − 100 ⎢− . ⎥ ⎟
⎜ ⎢⎣ α (λ + x) α ⎥⎦ 100 α 100 (λ + x) α
⎢⎣ α (λ + x) α ⎥⎦ 100 ⎟
⎝ ⎠
⎛ ∞
α⎜ 100 1⎡ 1 1 ⎤ 100 ⎛⎜ 1 ⎞ ⎞⎟
= αλ ⎜ + ⎢− ⎥ + − ⎟
⎟ ⎟⎟
.
⎜ α (λ + 100) α α ⎢⎣ (α − 1) (λ + x) α −1 ⎜
⎥⎦ 100 α ⎝ (λ + 100) α
⎝ ⎠⎠
⎛ 1 1 ⎞ λα
= αλα ⎜ . ⎟=
⎜ α (α − 1) (λ + 100) α −1 ⎟ α−1
⎝ ⎠ (λ + 100) (α − 1)
E[ S ′] = 125 × 660.18 = £82,523
Comments on Q5(ii): Most candidates were able to determine the values for α and λ
and wrote down the correct integral to be solved but not many were able to solve it.
(iii) Let Y + k be a gamma random variable with the same first three moments as S.
Then equating parameters:
Skewness = 0.6838 = 2 / α
Variance = 23,4802 = α / λ2
Mean = 82,523 = k + α / λ
So α = 8.555
λ = 0.0001246
k = 13,848
Comments on Q5(iii): This was correctly answered by the majority of candidates

who attempted it, although some who did badly on parts (i) and (ii) did not attempt
this part.
Page 10
6 (i) Advantages
• Possible reduction in premium rates if the claimant is the policyholder.
• Settlement delays should be significantly less.
• No legal fees (if they are deducted from damages).
• Simpler process.
• More consistent / predictable
• Can still use court system
• If the injury is not severe the overall payout should be higher.
• If no medical expenses or loss of earnings incurred then the overall payout
should be higher.
• If claimant has lower than average earnings the payout would probably be
higher.
• Awards may be inflated initially to promote confidence in the system.
Disadvantages
• Process will not be specific enough for certain more complicated cases.
• Fee that is non-refundable.
• If the injury is severe the overall payout will be too low.
• If no medical expenses or loss of earnings incurred then the overall payout
will be too low.
• No representation.
• Two separate claims will need to be processed if there are property
damages as well i.e. 1 through BICA and 1 through insurer.
• May have to go to court anyway so overall costs / delays longer.
• Postal strike will delay the process
• BICA aims to reduce premium rates so may give lower awards
• If the claimant has above average earnings the payout would probably be
lower.
Comments on Q6(i): Most candidates answered this part of the question well
although a few candidates incorrectly talked about the advantages and disadvantages
from the insurer’s perspective.
(ii) Adjustments to pricing model

• Divide injury claims by type of injury
• Base assumptions on claim amounts published and used by BICA rather
than on own experience.
• Therefore results have less reliance on rating factors
• However if claims go to court awards will not be based on BICA.
• Therefore need an assumption about the rate of claims to court and the
nature of the claims to court.
• Divide claims by severity of injury — need to make an assumption on
potential savings or losses as a result of average damages.
• Divide injury claims into general damages and special damages e.g.
medical expenses, loss of earnings — need to make an assumption on
potential savings or losses as a result of average damages.
Page 11
Possible changes expected to number of claims:

• Possible increase in number as may be simpler to claim
• Possible decrease as more difficult for solicitors to encourage claims with
no win no fee offers
• Possible decrease in fraudulent claims.
• Significant changes in average cost per claim due to standard damages
regardless of severity of injury or level of medical expenses etc.
• Possible initial increase in average cost as BICA will want to promote
confidence in system.
• In long run possible decrease in average cost as claims inflation decreases.
• Amend assumption about future claims inflation.
• Could be due to rejection of BICA damages.
• or a delay of longer than 2 years before notification.
• Amend expected notification delays.
• Include possibility of longer delays due to claimants waiting 2 years to
enter the court system.
• Amend settlement delays.
• e.g. 30 days for small claims — 90 days for larger claims.
• Include longer settlement delays for claims entering court system.
• Refer to industry data from any countries that have a similar system in
place to support your calculations.
• Adjust large claim loadings where these are no longer felt to be
appropriate due to the change in the data groupings.
• Adjust for any proposed changes in cover following introduction of new
system.
• Adjust for any proposed changes in mix of business / target market /
distribution channel following introduction of new system.
• Include assumption about the proportion of combined damage and injury
claims as this will result in an increase in claims handling expenses as
more than one claim will need to be set up.
• Include assumption about the reduction in claims handling expenses as a
result of most bodily injury claims being processed by the BICA.
• Include assumption about any longer term reduction in fixed costs that
may be possible as a result of this.
• Amend assumption about the level of legal expenses.
• Change reinsurance loadings where appropriate`
• Include assumption about the BICA fees based on forecast market share.
• Amend investment return assumptions to reflect the change in the expected
duration of the liabilities.
• Factor in pressure from regulators to reduce premium rates.
• Consider possible changes to competition in the market.
• Amend any profit loadings that may have changed following the
introduction of the new system.
Page 12
Comments on Q6(ii): Most candidates did not generate anywhere near the number of
points on the marking schedule. Better candidates had a more structured answer
considering the claims that would be settled by BICA and those that would go to court
separately.
(iii) Changes to system

• Add severity of injury as this will result in fairer awards.
Severely injured claimants will be less likely to reject the award.
• Employ a group of independent doctors to process the application forms.
This will reduce fraudulent claims and ensure consistency of diagnosis.
• Make the fee refundable if the claim is successful and the claimant accepts
the award.
This is fairer to legitimate claimants and will also discourage
fraudulent claims as there are different costs involved depending on the
outcome of the claim.
• Include claims older than two years e.g. less than five years old.
This could further increase the savings on legal expenses
• Introduce other injury compensation e.g. medical expenses, loss of
earnings.
This will result in fairer results and claimants who have incurred
significant costs will be less likely to reject the award.
• Introduce standard property damage awards calculated by a pricing
actuary.
This will reduce claims handling expenses for the insurers and make
the process simpler.
• Introduce a deterrent for going through the court system e.g. the BICA
award is no longer available regardless of the outcome of the court case.
This will reduce settlement delays and legal expenses.
• Introduce face-to-face meetings if required. Allow additional
documentation if necessary.
This will result in fairer diagnosis and ensure claimants are happier
with the outcome.
• Introduce an arbitration / appeal process.
This will ensure that claimants are happier with the outcome and make
it less likely that they will go to court.
• Allow claimants to opt out of BICA.
This will save costs relating to those claimants who have no intention
of accepting the BICA award.
• Add the option of legal representation if requested but with fees covered
by policyholder.
This ensures that less educated claimants (or any with other issues)
will still be well represented, but still deters the majority of claimants
from engaging a solicitor.
• Refer more complicated cases or those expected to settle above a certain
threshold to courts.
This will result in a more appropriate outcome for complex cases that
cannot be properly addressed by the BICA.
Page 13
• Ban no win no fee practices.

This will reduce fraudulent claims and will also reduce the number of
claims that are unlikely to be successful.
• Amend the insurer’s BICA fee so that it is based on market share by line
of business or on a claim-by-claim basis.
This will more accurately charge those insurers with a higher
proportion of bodily injury claims.
• Process applications by fax / internet / email..
This will speed up the process and avoid postal strikes..
• Be more specific with which claims the BICA will apply to i.e. only future
notifications or current IBNR claims.
This will help pricing actuaries to more accurately calculate their
prices and prevent any unnecessary delays in the process.
Comments on Q6(ii): Most candidates made a reasonable attempt at this question,

although some of the reasons given for suggested improvements were questionable
e.g. “removing the small fixed fee as some claimants may not be able to afford it”
would have the opposite effect on fraudulent claims. Some students did not give
specific enough reasons and some tended to focus on the insurance companies’ points
of view rather than considering the claimants as well.
Page 14
EXAMINATION
2 October 2009 (pm)

booklet.
2. You have 15 minutes before the start of the examination in which to read the
questions. You are strongly encouraged to use this time for reading only, but notes
may be made. You then have three hours to complete the paper.
supervisor.
5. Attempt all six questions, beginning your answer to each question on a separate sheet.
question paper.
1 A small upmarket supermarket has approached a large general insurance company
directly and has enquired about insuring a promotional product. They sell a range of
luxury food hampers that are delivered all over the world. They would like to offer
purchasers a guarantee that if a hamper is lost in transit then a replacement will be
dispatched within seven days of the loss being accepted. The supermarket sends the
hampers uninsured via national postal services and couriers depending on destination
country and the weight of the hamper. For every hamper lost the insurance company
will pay the supermarket the replacement cost.
Explain what data would be required in order to rate this business, including the likely
source of these. [6]
2 The following data have been extracted from the accounts of a medium-sized general
insurance company which writes mainly motor and household business. The
company prepares its accounts on an accident year basis with a financial year end of
31 December.
(All figures in £ millions) 2007 2006

Net written premium 40.0 26.0
Net earned premium 38.2 25.5
Net incurred claims 36.5 17.3
Increase in DAC 2.5 1.5
Expenses paid 4.5 3.5
Commission paid 7.0 5.5
Investment income on insurance funds 6.5 3.4
Other investment income 2.0 1.2
Total assets at 31 December 51.5 35.0
Total liabilities at 31 December 38.0 20.5
“Net” in the above table means after allowing for reinsurance.
(i) Calculate the following for each of 2007 and 2006
(a) Loss ratio

(b) Expense ratio
(c) Operating ratio
(d) Solvency ratio
[5]
(ii) Comment on the ratios calculated in (i), giving possible reasons for the
features observed. [8]
[Total 13]
ST3 S2009—2
3 A general insurance company has been asked to tender to provide insurance cover for
the organisers of a major sporting event to be held in 2012 at a number of venues
across the United Kingdom. At the moment approximately half the venues have been
completed while it is anticipated that those remaining will be complete by early in
2012. In addition to the sporting events the competition will be preceded by a series
of concerts featuring top musicians. The proposed cover will be a three year policy
commencing on 1 January 2010 with the premium payable in three equal annual
instalments. It is intended that the cover will be on an “all risks” basis that will meet
the organisers’ insurance needs.
(i) Describe the different types of cover that would be required, including the
claims characteristics of each. [14]
(ii) List the exposure measures that would be required to determine the
appropriate exposure for rating the cover in part (i). [4]
(iii) Discuss other risk considerations that would be considered when rating this
particular product. [7]
(iv) State reasons why a general insurance company may choose to model claims,
other than to calculate the risk premium. [4]
[Total 29]
4 (i) State the main objective regarding the investment of the free reserves of a
general insurance company. [1]
The following information has been extracted from the management accounts of two
proprietary non-US general insurance companies, A and B.
Company A Company B
Value of assets £600m £55m
Value of liabilities £400m £40m
Largest class of business written UK Motor Commercial Property (USA)
An actuary is advising on the best investment strategy to use for the assets of each
company.
(ii) Discuss the strategies that the actuary would recommend and the factors that
should be considered when giving advice. [10]
(iii) Describe the extra information that would be needed before a more informed
level of advice could be given, explaining why this information is important.
[3]
[Total 14]

5 The actuarial team of a large accountancy firm is currently working on the audit of a
medium-sized general insurance company. The company writes a number of different
lines of business. The following is the only information that has been provided.
XYZ Insurance Plc — Year Ending 31 December 2008
Summary of written premium

Class Amount Territories
Commercial Property £250m 50% USA, 25% EU, 25% Asia
Motor Reinsurance £70m 50% local, 50% Turkey
Household £65m 100% local
Product Liability £60m 75% North America, 25% EU
Marine £100m Mixed global coverage
Professional Indemnity £55m Large global law firms
Other £50m Various
TOTAL £650m
Summary by accident year from internal actuarial reports

Class 2005 ULR* 2006 ULR 2007 ULR 2008 ULR
Commercial Property 200% 45% 50% 75%
Motor Reinsurance 50% 45% 50% 55%
Household 75% 70% 86% 80%
Product Liability not written not written 55% 65%
Marine 125% 50% 55% 35%
Professional Indemnity 52% 43% 52% 65%
Other 80% 80% 80% 80%
*ULR = Ultimate Loss Ratio = ultimate claims/ultimate premiums
High level summary of reserves and methods adopted
Class IBNR Reserves** Method

Commercial Property £250m £520m Based on run-off triangles
Motor Reinsurance £160m £300m Based on contract by contract review
Household £20m £80m Based on delay tables
Product Liability £55m £60m Based on case estimates and pricing
loss ratios for IBNR
Marine £120m £300m Based on run-off triangles
Professional Indemnity £190m £320m Based on report from US parent
Other £27m £100m Run-off triangles if data allow
** Reserves include IBNR
(i) Outline with reasons the areas where the team should concentrate its review
providing justification for your choices. [7]
The team is invited to attend a meeting with the company’s chief actuary.
(ii) Describe the key information that the team would wish to obtain from this
discussion on the methodology and assumptions used by the company. [7]
[Total 14]
ST3 S2009—4
6 A general insurance company writes direct household insurance. For the last eight
years its reinsurance programme has consisted solely of catastrophe excess of loss
reinsurance renewing on 1 January each year. This gives cover of 0.5% in excess of
0.05% of the total sum insured for the book at the date of loss and one free
reinstatement. The cover has always been 100% placed with the same reinsurance
company. The insurance company has experienced the following results.
Accounting year (beginning 1 January)
11 12 13 14 15 16 17 18 19 20
Sums insured 1 Jan (€bn) 28 30 33 38 45 52 56 66 81* 99*
Gross written premium (€m) 177 181 198 234 267 284 290 342* 422*
Gross earned premium (€m) 175 178 184 211 256 277 287
Net earned premium (€m) 159 163 170 197 241 261 271
Gross claims incurred (€m) 120 142 142 143 166 192
Net claims incurred (€m) 120 142 142 143 166 192
* = forecast
€1bn = €1,000m
In Years 12 and 13 the company suffered an unusually high number of large

subsidence, fire and liability claims. In Year 17 a severe windstorm and a widespread
flood resulted in additional claims of €19.3m and €33.7m at the start of February and
October respectively. If the weather events had not occurred, the gross loss ratio for
Year 17 would have been 67.0%.
It is now halfway through Year 18 and you are deciding on the reinsurance
programme for Year 19. The Board of Directors has questioned whether the existing
reinsurance programme is still appropriate.
(i) List the potential benefits to the company of purchasing this type of
reinsurance cover. [3]
(ii) Calculate the total gross and net claims incurred and gross and net loss ratios
for Year 17, stating any assumptions made. [5]
(iii) Suggest reasons why the Board of Directors has challenged the reinsurance
programme. [6]
(iv) Outline the options available to the company regarding its reinsurance
programme, giving their relative merits. [10]
[Total 24]
END OF PAPER
ST3 S2009—5

September 2009
EXAMINERS’ REPORT
Introduction
R D Muckart
December 2009
Comments for individual questions are given with the solutions that follow.
1
Manufacturing cost,
retail price of the hampers

shipping costs
to set claim cost per hamper
Obtained from supermarket and retailers
Historic sales volumes of each type of hamper.

and sales forecasts
for exposure measure
from the supermarket
Past hamper loss frequency

to estimate loss frequency & hence loss cost
Data could be obtained from postal service
Hampers sent worldwide and so the weights to apply to different territories will have
to be estimated, using client past data if available
Data splits (e.g. territories)
Internal claims data unlikely unless you have written this type of business before
Costs of administering this business, policy issuance costs, claim administration costs.
Need to know by how much to load the policy to recover expenses.

Exclusions (e.g. territories/strikes) including deductibles
Length of contracts
Recoveries (e.g. from postal services) or subrogation on damage
Available internally.
Need company profit targets,

and any relevant taxes
to know how to load the policy.
Available internally
Competitor information and other market research
Page 2
to assist in setting the premium at a level appropriate for current market
While students were generally able to generate a reasonable number of points on data
sources required to estimate future claims, very few clearly explained why these data
would be required as per the command word of the question. Many candidates spoke as if
the product has been around for some time using phrases such as “full policy download”
or "full list of rating factors". Although these forms of data could be available, the
question is fairly clear that this a new product, and therefore other sources of data would
be needed, and a fairly simple approach to pricing was needed. Many candidates did not
clearly identify the sources of data as requested. Few candidates went beyond the claims
information to consider the wider data required such as expenses and commission in
order to complete the rating process.
2
(i) Loss ratio = Claims Incurred/Earned Premium
2006: 17.3/25.5 = 67.8%
2007: 36.5/38.2 = 95.5%
Can give expense ratio credit if commission excluded explicitly or calculated

in separate ratio
Expense ratio = Expenses Paid (including commission) /Written Premium

2006: (3.5+5.5)/26 = 13.5% + 21.1% = 34.6%
2007: (4.5+7)/40 = 11.3% +17.5% = 28.8%
Alternatively, if calculated using earned premium:

Expense ratio = Expenses Paid (including commission) less change in DAC
/Earned Premium
2006: (3.5+5.5-1.5)/25.5 = 29.4%
2007: (4.5+7-2.5)/38.2 = 23.6%
No credit for operating ratio if commission excluded
Operating ratio = claims ratio + expense ratio

2006: 102.5% (or 97.3% if expense ratio is on an earned basis)
2007: 124.3% (or 119.1% if expense ratio is on an earned basis)
Solvency ratio = (Assets – Liabilities) / written premium
Page 3
2006: 14.5/26 = 55.8%

2007 13.5/40 = 33.8%
Generally well answered although a few mistakes by either using incorrect formulae or in
calculation
(ii) General comments

Ideally we would want to consider more than two years’ worth of data
before reaching any firm conclusions about possible trends.
Ideally we would want to calculate the ratios for each class separately,
which would help to analyse performance at sub-account level.
Loss ratio
The loss ratio has worsened substantially.
This is most likely to be a result of a catastrophe in 2007.
However, premiums have also risen substantially at the same time
…which could indicate that the company is writing poorer quality
business.
Further causes could include:
Generally very poor claims experience, such as many large liability
claims.
Poor underwriting.
Inadequate premiums.
Severe deterioration in claims controls.
A strengthening of reserves for outstanding claims or IBNR.
A change to the level of reinsurance cover
Failure of one or more reinsurers.
Expense ratio
The expense ratio has improved…
…particularly the commission element.
Administrative expenses could be lower as a proportion of premium
because of:
An increased volume of business over which the expenses have been
spread.
Cost reduction initiatives.
Page 4
Commission rates may be lower due to:

Renegotiation of rates with the distributor.
Increased use of a different sales channel.
An expansion in one of the classes that has a lower commission rate.
Reduction in profit-related commissions due to deteriorating results.
Operating ratio
The operating ratio has worsened, primarily due to a worsening loss ratio.
The fact that the denominators for the loss ratio and expense ratio may be
different can lead to a distorted picture if the business is growing or
shrinking rapidly.
However, there does not appear to be significant distortion in this case.
Written and earned premium have grown similarly, which suggests that the
bulk of the growth took place in late 2006/early 2007.
Solvency ratio
Solvency has worsened significantly.
The level of free reserves may not support the rapid increase in business
going forward.
This is because net assets have reduced whilst written premiums have
increased
The poor claims experience has probably contributed to this
Generally well answered with the best answers being those that worked methodically
through the four ratios, commenting on each in turn and generating a wide range of ideas
for the reasons for the changes in ratios from 2006 to 2007. Some candidates did not
appear to have read the question thoroughly, and commented more generally on all of the
figures provided rather than just the four ratios specified. Some labelled their ratios by
year correctly in part (i) but then reversed the direction (i.e. seemed to assume that 2006
came after 2007!) in their comments in part (ii) although would have gained some credit
for valid observations.
3
(i)
Employers’ Liability
Very large number of employees/volunteers will be working on the event
Page 5
Bodily injury claims of various sizes depending on nature of accident

These could be very large e.g. in the event of the permanent disablement of a
high earning employee.
…and can take a long time to settle due to litigation/medical evidence
May be reporting delays as the injury may not deteriorate for some time
..therefore the claims cost will be impacted by level of inflation
In the extreme, could be latent claim issues such as exposure to toxic
substances
Possibility that liability claims are re-opened
Public Liability
Large number of attendees (or other people near venues) at the events so
possible claims for slips/trips or more significant injuries
Also claims for damage/theft of property from negligence of organisers; these
will be settled more quickly
Frequency may be expected to be higher than for EL as a very large number of
spectators expected
Accumulation of claims as same event will impact many people
Financial Protection
Losses could arise from non-performance/insolvency of subcontractors
Or the failure of a commercial sponsor of the event
Potential could be very large and lead to lengthy legal actions as contracts
likely to be complex
Directors & Officers

Could be significant claims against the organisers for maladministration of the
event
Likely to be large and potentially notified long after the event
Construction
If organiser is responsible for construction of venues, likely to be claims for
damage/delay to unfinished stadia
Commercial Property
Potential for catastrophe losses from weather event e.g. flood
Page 6
Significant potential for terrorist attack as high profile event – could give rise
to significant damage/injury claims especially if negligence proved
Variable cost distribution – extreme case: loss of stadium
Much shorter reporting and settlement delay as cause likely to be identifiable
quickly
..but still potential for disputes e.g. damage due to negligence of architect
rather than storm
If cover included for business interruption this will be longer tailed
Contingency
Non-appearance of pop stars at concert could lead to significant losses if event
cancelled
Sponsors losing revenue from events
Likely to be very short tailed as will know of loss very quickly and cost of
making refunds etc.
Motor
Both bodily injury and property damage claims could arise as the organisers
are likely to operate motor fleet
PD claims small, consistently distributed, injury claims subject to delays but
less so than EL
Likelihood of seasonality of claims as more accidents in wet weather
Competitors PA & Belongings Cover

Fixed benefit for athletes competing at event
Amounts high depending on event and extent of athlete’s earnings
Goods In Transit
Covers for delivery of merchandised items/equipment to venues around the
country
Possibility of moral hazard if economic conditions worsen
Product Liability
Indemnifies against loss caused by defect in event-branded merchandise
Computer Cover
Indemnifies against loss caused by virus/criminal hacking of the event website
Page 7
..or losses arising from failure of event-booking engine
Fidelity Guarantee
Loss caused by theft/criminal act of employee
The better answers were those which worked through a thorough list of relevant
insurance covers plus descriptions of their characteristics tailoring each carefully to the
specific scenario described. Students were expected to demonstrate understanding of the
relative importance of different types of cover in such a situation and some wasted time
giving a lot of detail on relatively minor covers. Many of the claims descriptions were too
vague.
(ii) Number of venues

Number of competitors
Number of employees
Number of concerts
Length of sporting event
Cost of rebuilding venues (including sum insured)
Size of motor fleet
Payroll
Capacity of venues
Number of attendees at the concert/event
Expected ticket price of concert
Turnover of organisers
Number of subcontractors
Cost of work undertaken by subcontractors
Total merchandising sales
Website traffic
Number of sponsors
Amount paid by each sponsor
The best answers were those that worked methodically through their list of covers in part
(i) realising that more than one exposure measure could be applicable for each. Some
candidates wrote the same exposure measure repeatedly for different covers rather than
identifying a range of factors.
(iii) Amount of risk that will be retained by organisers – cover might only be
required for really large losses
Page 8
Extent of support from government – they may provide cover for any losses
related to terrorism
Expert reports from surveyors or health & safety consultants
The projected value of completed venues against those still under construction
at 1/1/10
Reinsurance cover available
Capital availability/margins
Investment (in risk context)
Expense risk (in context)
Diversification/how the product fits in with rest of business written
While premium is paid over three years, the risk is much higher during the
period of the event itself
Experience of organisers in holding similar events though due to scale may not
be appropriate comparison
Could use data from previous events to determine historic claim costs
..but likely to be difficult as not publicly available or covered by state
Could consider previous reliability of performers at concert and if event has
enough so that one or two withdrawing would not matter
Location of buildings (including aggregations) would impact cost of
rebuilding/repair
Future economic growth projections could be used to judge likelihood of
supplier insolvencies
Fire prevention measures installed in venues
Number of security personnel/technology available to prevent possible
terrorist attacks
Size of event could lead to skilled labour shortage and therefore cost of
carrying out repairs could be much higher than anticipated
Any aggregation within sectors of the event sponsors – e.g. if they are all
banks will increase risk
A challenging question that required candidates to think widely round the problem. It was
generally poorly answered with many missing that the product is over three years which
would be expected to be a key consideration. Other simple things were missed, e.g. what
data are available to price.
(iv) To select rating factors

To determine premiums using experience rating procedures
To estimate the effect of changing the level of cover by changing the level of
deductibles
Page 9
To demonstrate the effect of reinsurance

To estimate likely variability of claims experience
For reserving (including estimating possible effect of industrial diseases on
reserves).
Statutory requirements
To assess the degree of solvency
To determine and allocate capital to different classes/categories of business
To value portfolios for purchase/sale

To estimate cash flow to determine investment strategy
Investigations to draw out trends that may impact profitability
Budgeting and business planning for future years, including staff planning
This was well answered by those who had learned the bookwork thoroughly. The
question is written generically (stating "a GI company”), not about the specific scenario
and it also clearly asks for the reasons to be stated. However, some students did not
appear to have read the wording carefully and instead gave a discussion of claims
modelling under the scenario described previously in the question.
4
(i) To maximise the long-term return for policyholders/shareholders
Subject to meeting future liabilities and solvency requirements
..subject to satisfying the company’s risk appetite
e.g. covering any shortfall in assets
A bookwork question that was well answered by those who knew the Core Reading but
very vaguely and imprecisely answered by many.
(ii) A
£200m free reserves is significant so more investment freedom to invest in
riskier assets
Could invest £400m in secure assets to match liabilities by term
Suitable assets include cash/short bonds/index-linked gilts
Free reserves could be placed in higher risk investments to maximize return
Some examples could include equities or property
Mixture of long and short term investments good match for liabilities arising
from motor business
Consider matching the nature of the liabilities
Page 10
Use longer term assets to match bodily injury claims

Short term assets appropriate match for property damage claims
B
Less investment freedom as low free reserves of £15m
in absolute terms and relative to liability
Commercial property claims can be volatile..
… and need value of assets > value of liabilities so avoid volatile assets
Writes cat exposed business so may need more liquid assets
Largest class of business is in US so consider matching by currency
Suggest mainly short-dated assets (to match liabilities by term)
Some longer delays arising from business interruption claims
Suitable examples are secure assets e.g. cash/short bonds
Possibly small proportion in equities/indirect property/longer bonds
For both companies need to consider:

Investment expertise available (may be less for B as small)
The company's risk appetite
Level of investment expenses of alternatives
The impact on each company’s tax liabilities
Availability of assets to purchase in the market
Level of reinsurance held by each company
Future growth plans (availability of premium income)
The level of non-investible assets that each company holds
The influence of the supervisory authority on investment policy
The need to invest the statutory minimum margin short
Economic Outlook
Diversification of assets
The better answers here were those that considered each of Company A and Company B
in turn, and then general factors that would impact both. Students were expected to tailor
their answers to the detailed information provided, so those that simply stated generically
e.g. that investment mix should depend on financial strength would not have gained any
credit. Many students wrote about the short- tail versus long-tail nature of different
liabilities written, but did not always clearly relate these to the need to invest in short and
longer term assets respectively. Despite the heavy hint in part (i) of the question,
Page 11
relatively few students clearly identified a strategy for the investment of the free reserves
separately from the investment of the assets backing the liabilities.
(iii) What basis has been used for calculating Assets and Liabilities
…if too cautious, could constrain investment policy too much
What premium income is expected in the future?
…if high then could use income stream to help pay out liabilities, meaning
greater investment freedom
Are companies A and B ‘stand-alone’ or part of a larger group?
If they have access to funds from a larger parent, this may give more
investment freedom
What is the required Statutory Minimum Solvency Margin?
What modelling has been done for the future liability outgo?
…gives assurance that taxes, dividends and timing issues have been
considered
What assets competitors are investing in
Other classes of business written
Split out the figures to provide more detail (e.g. split of Motor PD & BI)
Marks given were generally low but most candidates made some comments about the
additional information that may be desirable.
5
(i) Ensure a certain percentage of reserves are reviewed
Commercial Property class covers a significant proportion of the premium
written (38%) so important
Historic loss ratios show likely to be cat exposed – significant source of
uncertainty for class
And USA and Asia exposure that might be particularly prone to cats
Large proportion of the Motor RI account written in Turkey – possibly very
different claims profile
Also non-standard technique used (contract by contract approach) so worth
investigating
IBNR is significant part of reserves for both Motor RI, PI and PL classes so
value added from actuarial investigation
Marine 2008 ULR looks unusually low compared to recent years
Review classes of business that have large reserves
Look at PI class as potential concentration of risk by industry (all global law
firms)
Page 12
..and relying on someone else's report

Investigate inconsistency between reserve amounts and WP in PI class
Product Liability class has only been written for two years so uncertainty due
to little historic data
Review the long-tailed/liability classes such as Product Liability and PI as
claims may emerge over a long period
Suggest spending less time on Household business as relatively small class
with stable results
The results from the other class look unusual as the same loss ratio booked on
all years – worth further investigation especially as method is vague
General Points
Expertise of the team that has produced the results – if new actuary looking at
some classes may need to review more
Any issues with currency conversions
If an external actuarial consultant has reviewed reserves may give more
comfort in these areas
The time available and resources to complete audit in professional manner
Any issues that senior management have raised regarding individual classes or
the actuarial function in general
Any other significant inconsistency observed
The best answers were those that worked methodically through each of the classes of
business, identifying two or three specific areas of concern for each using the information
provided. Many candidates ignored virtually all the information in the question despite
many marks being available for pulling the key points from both the numerical data and
the descriptions. Very few queried why the ULR for Other was 80% for all years.
(ii) Basis that the reserves are calculated on – best estimate or prudent
Are the reserve figures quoted IBNR + outstanding or do they include UPR,
AURR or other contingency reserves?
Have ranges been calculated in addition to point estimates?
What risks are written in the other class of business?
Is the report from the US parent for the PI class of a reasonable quality?
..are the figures used reviewed in any way by the UK team?
Have inflation and premium rate changes been incorporated in methods?
Are appropriate checks carried out on the data before use?
What reserving methods are used for the classes based on CL techniques
Actual vs Expected analysis - including any changes in methodology
If the BF method is used, how are the IELR assumptions derived?
Page 13
Is the analysis split between large and attritional losses as the development of
these can differ?
The actuary’s opinion of any trends in the loss ratios
Are the results presented net or gross of reinsurance/retrocession?
Ask for reinsurance resumé (i.e. provide details of the programme)
Have any of the reserves been discounted for investment income?
How have changes in the mix of business or terms and conditions been
represented in the methods chosen
Are there any known large losses that are not yet reflected in the data?
Is any separate allowance made for IBNER in Household class as delay table
method does not automatically provide this
Are claims and other specialists within the company involved in determining
the reserves
Has GN12/appropriate professional guidance been followed
Any more detail considered necessary on data splits and data sources
Any other significant request
This should have been tackled by thinking "what could possibly have gone wrong within
the reserving?" and "what is not completely clear from the information provided
already?" Reasonably well answered by the better students but many concentrated their
answers on just claims issues, such as development patterns, when a broader approach
was needed. Few candidates managed to generate a sufficiently wide range of points to
score highly on this question. Writing a page about each of three points rather than
making a large number of much briefer points is not generally a successful approach in
ST level examinations.
6
(i)
Smooth financial experience over time by reducing claims fluctuations.
Particularly since the retention is not very high.
Reduce the capital requirement due to the reduction in claims variability.
Particularly since the upper limit is fairly high.
Alternatively, increase capacity to write more risks through better use of
capital.
Diversify or further stabilise the portfolio by writing more risks.
Protect the company’s solvency by truncating the effect of catastrophes.
Protect against accumulations if there are concentrations of risk in certain
geographical locations.
Technical assistance from reinsurance broker.
Page 14
This type of cover may be good value compared to other options

Cover grows with the business as directly related to Sum Insured
..therefore saves admin costs as don't need to renegotiate
Improves market standing (rating agencies/regulator/policyholders)
The free reinstatement gives protection against second and possibly
subsequent events
Although asked regularly, few candidates managed to score full marks.
(ii) Gross claims incurred = 0.67 × 287 + 19.3 + 33.7 = 245.3

Assume the sum insured at any point is the weighted mean of the sums insured
at the 1 Jan before and after that point (linear change in SI during each year).
Assume that the flood event triggered a reinsurance recovery under the terms
of the contract (no RI disputes or defaults).
Assume that the quoted figures for the windstorm and flood losses are ultimate
estimated costs and are gross of reinsurance.
Sum insured (€bn) at start of Feb = (11 × 56 + 66) ‚ 12 = 56.83
Sum insured (€bn) at start of Oct = (3 × 56 + 9 × 66) ‚ 12 = 63.50
Retention (€m) Feb = 0.0005 × 1000 × 56.83 = 28.42
Retention (€m) Oct = 0.0005 × 1000 × 63.50 = 31.75
Recovery (€m) = 33.7 – 31.75 = 1.95 (no recovery from Feb event)
Net claims incurred = 245.3 – 1.95 = 243.3
Gross loss ratio = 245 / 287 = 85.5%
Net loss ratio = 243 / 271 = 89.8%
Most candidates were able to determine the gross claims incurred and gross loss ratio
correctly, but relatively few performed a thorough calculation of the net claims. Almost
all assumed that the sum insured would be constant over the year, but given the
information provided this is not a realistic assumption and so would not have gained full
credit. Loss ratio calculations often had the wrong denominator. Some candidates did
not fully answer the question, calculating only the claims amounts and not also the loss
ratios. Some students were not able to interpret the reinsurance layer correctly. Very few
gave the assumptions they were making.
(iii)
The business has moved on over the years but is still using the same
programme, which may make it appear that it has not been re-evaluated
properly each year.
Written premiums are projected to grow rapidly over the next two years,
Page 15
which may mean the insurance risks need to be re-evaluated.

The Board’s risk appetite or performance targets may have changed with
the growth of the business.
The company’s capital position may have strengthened recently, lessening
the need for reinsurance.
The following are the net loss ratios achieved in years 11–14.
Year 11 12 13 14
Net loss ratio 75.5% 87.1% 83.5% 72.6%
The loss ratios in years 12 and 13 are poor compared with surrounding
years
...so there may not be enough protection against individual large claims.
Despite buying reinsurance protection, the net loss ratio in year 17 was
still poor compared with other years.
The company has spent around €15m each year on reinsurance but has
only recovered €2m over the 7 years.
With two weather events fresh in the mind, the Board may be worried
about continued poor experience if these events become more severe or
frequent.
The price of this form and level of reinsurance might rise following
industry-wide weather losses.
The company might not be getting the best price if cover is always with
the same reinsurer.
The company may be paying too high loading in the premium for the
reinstatement element.
There is a concentration of default risk with a single reinsurer.
Most could see that the programme was not good value, but then couldn’t explain further
problems with the programme. The fact the company was expanding rapidly should have
indicated that the risk profile was changing.
(iv)
Raise (lower) the attachment point.
Reduces (increases) the reinsurance premium.
Increases (reduces) the capital requirement.
Reduces (increases) the recovery from catastrophes above the
Page 16
attachment point.
Reduce (increase) the upper limit
Reduces (increases) the reinsurance premium.
Increases (reduces) the capital requirement.
Reduces (increases) the recovery from extreme catastrophes.
An alternative is to include an annual aggregate deductible in the
contract
A further alternative to the above is to place less than 100%
Similar impacts to the above.
Remove the reinstatement.
Reduces the reinsurance premium.
Could leave the company with insufficient cover remaining following
a severe event.
This would require replacement cover, which may be difficult or
expensive to obtain.
Pay a reinstatement premium rather than having it free.
Defers or removes part of the expected reinsurance premium because
the reinstatement premium is only paid if the cover is used.
Reduces the effective cover because the recovery will be net of the
reinstatement premium.
Purchase additional reinstatements.
Increases the reinsurance premium and may be overkill.
Avoids having to purchase replacement cover following severe losses.
Stop reinsurance altogether
– But leaves company open to very large catastrophe losses
Purchase risk excess of loss.
Increases the cost of the programme.
Increases the company’s capacity to accept larger risks.
Protects against a cluster of individual large claims that fall outside the
catastrophe treaty.
Purchase quota-share cover.
Eases expansion by increasing underwriting capacity from the same
capital base.
Cedes profit.
Consider stop-loss cover.
Protects against adverse experience, regardless of cause.
Page 17
May be unavailable or prohibitively expensive.

Negotiate a better price with the current reinsurer
Cuts costs.
May drive the reinsurer to stop offering as much cover.
Negotiate a better price with an alternative reinsurer
Cuts costs.
Breaks the relationship with the former reinsurer.
Use several reinsurers on the programme.
May be able to obtain a better price.
May weaken the relationship with the former reinsurer.
Use reinsurers of different strength (credit rating).
Enables the company to balance price with risk appetite.
Consider alternatives to reinsurance, such as raising more capital/parental
bail-out or capital market solutions (alternative risk transfers)
May cut some costs but ART likely to be expensive for a class this
size.
May not be consistent with the company’s risk appetite
Students did not appear to have left themselves sufficient time to complete this question
part, given the number of marks available, and often covered only a relatively narrow
range of different options. The examiners were looking for a wide range of possible
alternatives although many candidates opted for a scattergun approach of listing other
types of reinsurance, whether appropriate or not. Few considered options related to
adjusting or re-broking the existing arrangement.
Page 18

Subjectst32005 2009

Uploaded by

Copyright:

Available Formats

Subjectst32005 2009

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Subjectst32005 2009

Uploaded by

Copyright:

Available Formats

Faculty of Actuaries Institute of Actuaries

8 April 2005 (pm)

Subject ST3 General Insurance

Time allowed: Three hours

INSTRUCTIONS TO THE CANDIDATE

4. Mark allocations are shown in brackets.

6. Candidates should show calculations where this is appropriate.

AT THE END OF THE EXAMINATION

Describe the following characteristics of this class of business:

(i) purpose and benefits [2]

7 A global relief agency provides humanitarian assistance to victims of natural disasters

ST3 A2005 3 PLEASE TURN OVER

Class A 2000 2001 2002 2003

Class B 2000 2001 2002 2003

(i) Comment on the financial characteristics (including financial trends) of each

Subject ST3 General Insurance

For each policy

All rating factor details:

Trade / Industry / Occupation

Details of risk premiums charged, may need to be a proxy

Date claim reported

Dates and amounts of payments

Estimates, if they exist, of amounts outstanding

Link to policy information

Type of claim, i.e. injury or disease

Industry statistics to back up own data where it is sparse

Changes in underwriting / claims handling / policy conditions

This will depend on how busy the company is.

Lower theft claim frequency likely for coach fleet

If a coach fleet has a total loss this will be a large claim.

Hence settlement delays will be greater for coach fleets.

For a coach fleet there is a potential accumulation of risk

May be different legislation impacting upon coach passenger claimants compared

Policy, claims and expense data are essential for reserving

This is also likely to be required for insurance Regulatory purposes, accounts or

Relevant, accurate information may help to prevent fraudulent behaviour

Purpose is to indemnify the policyholder against loss or damage to their commercial

Exposure used is the sum insured year

Stock amounts may fluctuate considerably. Stock may be covered on a declaration

No standard way of allowing for inflation in the policy.

Claims arise from sudden and determinable events (except subsidence)

Notification delays are short apart from subsidence

Possible accumulations owing to location

Generally low claim frequency compared with other property insurance

Many claims likely to be impacted by reinsurance arrangements

Good estimates of claims amounts can be made

Exposed to moral hazard especially in an economic downturn.

Frequency and Average Cost trends

Probably only have statutory accounts and returns available

e.g. UPR, (A)URR, OSCR, IBNR

Also, investigate trends in

claims handling costs and associated reserve.

Funding of the takeover

Loss/Claims ratio, Expense/Combined ratio, Commission rates, Investment

reinsurance purchased, recoveries made, reinsurers security

Investigate similar trends in other potential companies

Investigate the credit standing of the company

Investigate the combined model office

Investigate the synergies that the combined operation will have.