ACST1052: Introduction to Actuarial Studies WEEK 11 – GENERAL INSURANCE Slightly less morbid insurance Your feedback is important! • A reminder to please fill the feedback survey on iLearn – it is on the right of your iLearn homepage at https://ilearn.mq.edu.au/my/ • • I value and consider carefully every comment received. I am looking for ideas to make the unit better, so please don’t be shy if you have suggestions about content, resources, assignments, tutorials, etc. Anything that can improve your learning experience… let me know! ACST1052: Week 11 2 Overview • • • • • • • This week, we look at the General Insurance (GI) industry, which comprises a large variety of insurance products (everything that is “not life” and “not private health”). It is one of the most common areas of practice for actuaries in Australia (if not the most common). 29% of the Actuaries Institute members quoted GI as their “primary” area of practice in 2023. We first give an overview of the main GI products sold, we describe some insurance “principles” and we list some ways in which GI differs from Life Insurance. We then present some bits of mathematics essential to GI. Next week, we will provide some comments more specifically about the Australian GI industry. ACST1052: Week 11 3 General insurance generalities • “General Insurance” is a bit of a lazy term. In the USA and Canada, GI is called “Property and Casualty Insurance”, which is a bit more descriptive: ― Property insurance refers to insurance for “stuff”: buildings and their contents, motor vehicles, commercial aircraft or cargo, etc. ― Casualty insurance refers to insurance in the case you are found legally liable to provide compensation to another party (e.g., negligent acts causing injury). • There are significantly more lines of business in the general insurance industry compared to life insurance, because so many more types of events can be insured. • Usually, the GI industry is further split into two parts ― Personal lines: sold to individuals. ― Commercial lines: sold to businesses. ACST1052: Week 11 4 Some of the main Personal Lines • • • • Motor Insurance ― Compulsory Third Party (CTP, call “Greenslip” in NSW) ▪ Covers death or injury compensation caused to a third party in a car accident, when you are at fault. ▪ “Compulsory” because you cannot drive a car that does not have CTP insurance. ― Third Party Property Damage ▪ Covers your liability for damaging other people’s cars and property. ― Comprehensive ▪ Third Party Property Damage + covers your car being stolen or damaged (e.g., by weather). Home and Contents ― Covers all kinds of damage to your home (and its contents). ― Often covers burglary as well. Travel Insurance ― Covers losses related to travel, e.g., loss of baggage, trip cancellations, overseas medical costs. Pet Insurance ACST1052: Week 11 5 Some of the main Commercial Lines • Commercial Motor ― Like personal lines but for business vehicles like taxis and trucks. • Commercial Property ― Like “home and contents” but for business premises, such as shops, offices and factories. ― Often includes a “Business Interruption” component. • Marine and Aviation • Worker’s Compensation ― Employers purchase this to cover the cost of supporting workers who get injured at work. • Public Liability ― Covers liability to members of the public on business property. • Product Liability ― Covers products from businesses that may cause injury or death to customers. • Professional Indemnity ― Covers liability for professionals who are sued for doing a “bad job” (e.g. negligent acts or advice). (a bit more info here) ACST1052: Week 11 6 Principle of Indemnity • In general insurance, the amount payable to you for a claim is usually equal to the financial loss you have suffered (subject to sum insured limits and deductibles, which we explain later). ― E.g. for comprehensive motor, a no-fault crash will mean the insurer pays to fix your car to its original condition. • The idea is to put the insured person in the same financial position as if the unexpected insured event did not occur. The principle of indemnity is necessary to avoid moral hazard, which we mentioned last week. • There have been famous examples of moral hazard when over-insurance was allowed – consider the following historical example of marine insurance. ACST1052: Week 11 7 Marine Insurance and Samuel Plimsoll • • • • In the late 1800’s, UK ships were constantly being shipwrecked and hundreds of sailors died at sea every year. Samuel Plimsoll was a politician who campaigned heavily against “coffin ships” – these are ships that were unseaworthy and often over-loaded, with significant insurance taken on them, usually above their actual worth. ― This meant that if the ship was lost at sea, the ship owners would earn a profit! ― Plimsoll had quite a struggle getting bills passed on these issues – it turns out that many parliamentary MPs were also ship-owners… Even in the case of ship-owners not deliberately sinking their ships for the insurance money, the moral hazard still existed because the ship-owners would be careless about maintenance, repairs and crewing. Bottom line: when the policyholder has control over the insured event, this generates moral hazard, especially if over-insurance is allowed. ACST1052: Week 11 8 The principle of “Insurable Interest” • There are also limitations on what sort of events you can buy insurance on. For general insurance, an insurable interest must exist at the time of the claim – this means that you can only make a claim if you have suffered some financial or economic loss due to the insured event occurring. • For example, you cannot buy home and contents insurance on your neighbour’s house. • The principle of insurable interest also applies to life insurance (but you only have to show insurable interest at the time of purchasing the policy). This might be reasonable for the life of a spouse, parent, or even a business partner/employee. However, this would not apply to a random stranger (or celebrity?), as their death would not cause you a financial hardship. ACST1052: Week 11 9 Insurable Risks Not all risks are insurable – general insurers can be pickier about their products and so will only insure things they deem likely to be profitable. The following principles are good “rules of thumb” for what constitutes insurable risks: • • • • • • The loss must be fortuitous (i.e., a matter of chance). The insurer must be able to define the risk. The insurer must be able to assess the risk. The premium must be affordable. The insurance must not be contrary to the public interest. There must not be excessive risk / accumulation of risk (although you might get around this with reinsurance). • Can you think of an insurance product that would be “contrary to the public interest”? ACST1052: Week 11 10 Example of “being contrary to the public interest” • It usually isn’t possible to insure yourself against actions you take that are against the public interest (e.g., criminal activity). • For example, you couldn’t buy insurance against getting a speeding ticket – the moral hazard generated may result in the temptation to drive fast and endanger the lives of others. • Another example comes from the USA where “Carry Guard” policies were sold by the National Rifle Association (NRA). These have been labelled as “murder insurance” because the policies provided an upfront payment to cover legal fees if the policyholder is involved in a shooting – even if the policyholder ends up pleading guilty or is convicted of a crime. But, ― “After two years of state investigations and numerous lawsuits, the National Rifle Association has stopped offering its Carry Guard self-defense insurance products”. ACST1052: Week 11 11 Examples of “Accumulation of Risk” • • • Insurers are often reluctant to cover events that might cause a large number of claims all at the same time, as these claims could potentially ruin the insurance company. Some examples include: ― Flood and Cyclone insurance in Australia ― Earthquake insurance in NZ ― Terrorism insurance Insurers MAY be willing to cover these risks if they are able to purchase reinsurance – but reinsurance premiums will be high and reinsurers may also be reluctant to provide such reinsurance arrangements. When insurance is not available on the market but would be societally beneficial, governments sometimes step in to provide insurance themselves. For example, ― New Zealand has the Earthquake Commission (EQC) which covers the costs of certain natural disasters. Premiums are paid into the scheme whenever someone buys a home insurance policy. ― Australia has a Cyclone Reinsurance Pool: “The Government has implemented a reinsurance pool for insurance companies to transfer their risk for cyclones and cyclone-related flood damage, called the Cyclone Reinsurance Pool (cyclone pool), which is backed by a $10 billion Government guarantee.” ACST1052: Week 11 12 Claims Development Process Claims Development Process • The development process of a claim is generally much more variable for general insurance than for life insurance. • Some important features: • Reporting Delay – time between Time of Occurrence and Time of Report • Settlement Delay – time between Time of Report and Time of Settlement ACST1052: Week 11 14 Reporting and Settlement Delays • Reporting delay can be very long ― Example: Medical malpractice (professional indemnity) ― Example: Worker’s compensation ▪ Famous case: Asbestos (banned in 2003, discovered to be carcinogen). • Settlement delay can be very long ― Example: CTP may require lengthy legal procedures ― Example: Worker’s compensation may require long rehabilitation ACST1052: Week 11 15 Short vs Long Tail Lines of Business Often in practice, you will see lines of business divided into short vs long tail • Short tail lines are those where claims are reported and settled quickly ― Because the whole process doesn’t take very long for most of the claims, the risk of having lots of large claim payments in the future is not very high ― Remember that insurers must hold capital as buffer for future uncertainty. ▪ Because short-tailed lines have less risk and uncertainty (usually), they will have to hold lower amounts of capital to maintain appropriate levels of solvency. • Long tail lines may take years or even decades before most of the claims are settled ― Lots of risk and uncertainty (e.g., Asbestos case) ― You must hold lots of capital for these LoBs ― It is difficult to assess the value of outstanding claims liabilities… this valuation of liabilities is one of the important tasks actuaries do! ACST1052: Week 11 16 Key differences between LI and GI • • • • • • • Life insurance insures specific mortality / disability events whereas general insurance products insure a much larger variety of products. Life insurance products generally have a shorter reporting and settlement delays (except for IDII). Life insurance products have a defined Sum Insured (and a claim payment is usually for the full sum insured), this is not always the case for general insurance (so the magnitude of claims is not known ahead of time). ― For example, CTP has no SI amount – the claim cost is however much money is required to compensate the claimant for damages, and this can be HUGE (medical costs and rehabilitation, pain and suffering, future economic loss). There can be multiple claims on a GI policy. GI policies are generally short term, e.g. you buy your car insurance every year, covering the next year. Regulations in Life Insurance mean that life insurance pricing is generally much more regulated, and you can only take a small number of factors into account (e.g., age, sex, medical history). This is less the case for GI. GI companies can adjust premiums selectively and cancel unprofitable policies. ACST1052: Week 11 17 General Insurance Claims Modelling Motivation • When we did calculations for life insurance contracts, the claim “amount” was not random (only the timing of payments was random). • A common feature of general insurance products is that claim amounts are not known ahead of time (e.g., if you claim on your comprehensive motor insurance, it could be for a few 100’s dollars of repairs up to the total value of the car). • Also, contrary to LI, on GI policies an insured is usually allowed to make multiple claims on a single policy. E.g., if you are very unlucky (or maybe a very bad driver), you could have 1,2,3, etc. car accidents per year (while you can only die once). • So, we are now in a context where it is relevant to model both frequency (how many claims) and severity (how big are the claims) as random variables. ACST1052: Week 11 19 Modelling general insurance claims Claim Frequency • First, let 𝑁 be a random variable representing the number of claims. ― For example, this could represent the yearly number of claims on a single policy or on a whole portfolio of policies. Claim Severity • Let 𝑋1 , 𝑋2 , … , 𝑋𝑁 be the sizes of the 𝑁 claims. • We make the simplifying assumption that the claim amounts are i.i.d. So, we can without confusion call “𝑋” the generic severity random variable. • The total claim amount 𝑆 is then 𝑁 𝑆 = 𝑋𝑖 𝑖=1 ACST1052: Week 11 20 Claim frequency distributions • It makes sense to have 𝑁 be a discrete random variable. In Week 9, we had the “number of deaths” 𝐷 be Binomially distributed. This makes less sense in the context of general insurance claims modelling (what would 𝑛 and 𝑝 represent?). • Instead, a common assumption for frequency is the Poisson distribution. Recall the PMF for the Poisson distribution is Pr 𝑁 = 𝑛 𝑒 −λ λ𝑛 = 𝑛! for 𝑛 = 0,1,2,3 … Properties you (should) already know: • σ∞ 𝑛=0 Pr(𝑁 = 𝑛) = 1. • E 𝑁 = 𝜆. • Var 𝑁 = 𝜆. ACST1052: Week 11 21 Additive property of independent Poisson rvs • The Poisson has a cool property: if 𝑁1 ∼ Poisson 𝜆1 , 𝑁2 ∼ Poisson 𝜆2 , … , 𝑁𝑛 ∼ Poisson(𝜆𝑛 ) are independent Poisson random variables (with possibly different lambdas). Then, the sum: 𝑛 𝑁 ∗ = 𝑁𝑖 𝑖=1 follows a Poisson(𝜆∗ ) distribution! • Question for a Champion: what is 𝜆∗ ? • This property is useful as it allows us to model the total claims at the insurer portfolio level (under simplifying assumptions). ACST1052: Week 11 22 Additive property: Josephine’s love life • • • On any given day, the number of “Likes” Josephine gets on a dating app is a Poisson random variable. The average number of likes on weekdays is 1, while it is 2 on Saturday and Sunday. The numbers of likes received on different days are independent. What is the probability she gets at least 4 likes in any given week? ACST1052: Week 11 23 Solution • Call 𝑁 the weekly number of likes. We have 𝑁 = 𝑁1 + 𝑁2 + ⋯ + 𝑁7 , hence 𝑁 ∼ Poisson 1 + 1 + 1 + 1 + 1 + 2 + 2 = Poisson(9) • So, Pr 𝑁 ≥ 4 = 1 − Pr 𝑁 < 4 = 1 − (Pr 𝑁 = 0 + Pr 𝑁 = 1 +Pr 𝑁 = 2 + Pr 𝑁 = 3 ) = 1 − 𝑒 −9 92 93 1+9+ + 2! 3! = 1 − 𝑒 −9 ⋅ 172 = 0.978774 ACST1052: Week 11 24 Additive property: Insurance Claims Example • • • There are 20,000 policies in a portfolio. The number of claims on any one policy is Poisson distributed with parameter 0.01, and each claim costs $5,000. What are the expected value and standard deviation of the total claim cost? Solution: Because the claim cost is fixed at $5,000, the expression for total claim cost is simply 𝑆 = 5000𝑁 where 𝑁 is known to be a Poisson(200). Hence, E 𝑆 = 5000 ⋅ E 𝑁 = 1,000,000 S. D 𝑆 = ACST1052: Week 11 50002 ⋅ Var 𝑁 = 70,711 25 Claim severity distributions • • • • The previous example was oversimplified. As argued before, for GI the claim size would usually be random as well. There are many options to model such random claim sizes, and we usually prefer to use continuous distributions. There are many choices available – we often consider the shape and tail of the distribution when examining the optimal distribution to fit to the data. Common choices are ― Lognormal ― Gamma ― Pareto ― Burr Unfortunately, in this course we don’t really have the mathematical tools yet to derive the full distribution of 𝑆… unless we are in the simple case of a discrete claim size 𝑋. ACST1052: Week 11 26 Distribution of 𝑺: Example • • Assume that both 𝑵 and 𝑿 are discrete, with PMFs given below. Question: what is Pr 𝑺 = 200 ? ACST1052: Week 11 𝑛 0 1 Pr(𝑁 = 𝑛) 0.90 0.07 𝑥 100 200 Pr(𝑋 = 𝑥) 0.5 0.3 2 3 0.02 0.01 300 0.2 27 What is Pr 𝑺 = 𝟐𝟎𝟎 ? • We need to list all the possibilities which make the total claim size exactly 200. 1. One claim of size 200 2. Two claims of size 100 • Thus, Pr 𝑆 = 200 = 0.07 × 0.3 + 0.02 × 0.52 = 0.026. • To get the full distribution of 𝑆, you would need to do this for every possible value of 𝑆 0,100,200,300, … , 900 . It’s tedious to do by hand, although not terrible to do via Excel or R. • Even though we can’t (yet) derive the full distribution of 𝑆 in the case 𝑋 is continuous, there is one more thing we can do, which is to derive a general expression for the expected value of 𝑆. ACST1052: Week 11 28 Expected value of 𝑺 • • As before, denote the (random) number of claims by 𝑁 and the (random) claim amounts by 𝑋1 , … , 𝑋𝑁 (assumed to be i.i.d. and independent of 𝑁). Then, the excepted value of the total claim amount 𝑆 = σ𝑁 𝑖=1 𝑋𝑖 is simply: E 𝑆 = E 𝑁 ⋅ E[𝑋] law of total expectation • Proof. Recall the law of total expectation: E 𝑆 = E[E[𝑆|𝑁]]. This is the main ingredient of the proof, as we then have 𝑁 E 𝑆 = E[E[𝑆|𝑁]] = E E 𝑋𝑖 |𝑁 = E 𝑁 ⋅ E[𝑋] = E 𝑁 ⋅ E 𝑋 . 𝑖=1 • What is the variance of 𝑆 ? We don’t cover it here, but there is a “simple” formula for it, which is the topic of a tutorial exercise. ACST1052: Week 11 29 Example • Assume that an insurer’s total number of claims for a certain line of business is a Poisson(200) random variable. In addition, the size of any individual claim is modelled with the following density function: 𝑥 1 − 𝑓 𝑥 = 𝑒 5000 , 5000 • 𝑥≥0 What is the expected value of the total claim amount? ACST1052: Week 11 30 Deductibles Deductibles in general insurance • • • In general insurance, policies often require the policyholder to cover the first portion “$𝑑” of a claim. This is called a deductible or excess. Example: if you have a motor policy with a deductible of 𝑑 = $700 and the repairs after an accident cost $3,000, then you must pay $700, and the insurer pays the other $2,300. Of course, this means that if you are in a minor accident and the repairs cost only $500, then the insurer doesn’t pay anything. Question: What are some advantages of deductibles? ACST1052: Week 11 32 Advantages of a deductible 1. Reduces moral hazard ― People are incentivized to be careful if they know that they will have to bear some part of any loss. 2. Eliminates a lot of small claims ― Saves administration costs since people won’t submit claims that are less than the deductible. 3. Reduces the premium rates ― From the insurer’s PoV, a deductible means a lower expected value for the claims, hence they can charge less for the cover. This effect is compounded by the efficiency gained from not having to handle lots of small claims. ACST1052: Week 11 33 Calculating a risk premium allowing for a deductible • • Consider the previous example but we now allow for a deductible of $150 per claim. 𝑛 0 1 Pr(𝑁 = 𝑛) 0.90 0.07 Loss 100 200 Net 𝑥 0 50 Pr(𝐍𝐞𝐭 𝑋 = 𝑥) 0.5 0.3 2 3 0.02 0.01 300 150 0.2 To determine the expected value of the claim costs, we just need to calculate the average frequencies multiplied by average net claim size (allowing for the deductible): E 𝑆 = E 𝑁 × E[Net 𝑋] = 0.14 × 45 = 6.30 ACST1052: Week 11 34 Claim size caps and large losses Setting claim caps • Sometimes, general insurers can limit their exposure to large losses by capping the amount payable. • For example, if you own a $50M house, the insurer can limit the amount you are able to insure with them on a home and contents insurance policy. • But in many cases, this can contradict the purpose of insurance ― Worker’s compensation – the point is to pay compensation and costs for rehabilitation until the person can get back to work (however long this takes). ― Motor CTP – similar to the above. • So, if an insurer does not put a cap on the amount payable, how can they limit their risk? ACST1052: Week 11 36 Reinsurance • A common way for insurers to manage their risks is through reinsurance. The two main types of reinsurance contracts are: ― Quota Share ― Excess of Loss • Note that even if an insurer sells policies with only small sums insured, they can still be exposed to high total claim costs if there are catastrophe losses. • Let’s do a quick example involving Excess of Loss reinsurance. Excess of loss reinsurance is essentially an insurance company buying insurance with a deductible (from a reinsurer). ACST1052: Week 11 37 Insurer’s expected claims cost with XOL reinsurance • • Consider the same example as before, but the reinsurer pays the excess of claims cost over $180 on individual claims. To calculate the expected value of claims for the insurer, we have that E 𝑆 = E 𝑁 × E[Net 𝑋] = 0.14 × 140 = 19.60. ACST1052: Week 11 𝑛 0 Pr(𝑁 = 𝑛) 0.90 Loss 100 Net 𝑥 100 Pr(Net 𝑋 = 𝑥) 0.5 1 0.07 200 180 0.3 2 0.02 300 180 0.2 3 0.01 38 Reinsurer’s expected claims cost From the reinsurer’s perspective, we have that 𝑛 0 1 Pr(𝑁 = 𝑛) 0.90 0.07 Loss 100 200 𝑦 0 20 Pr(𝑌 = 𝑦) 0.5 0.3 2 3 0.02 0.01 300 120 0.2 In the above, we have let 𝑌 represent the amount paid by the reinsurer for each claim. E 𝑆 = E 𝑁 × E 𝑌 = 0.14 × 30 = 4.20 ACST1052: Week 11 39 Gross Premiums Terminology • We call “risk premium” the expected present value of the insurance benefits (only) on a given insurance policy (which typically covers a 1-year period). • We call “pure premium” the expected present value of insurance benefits and all other costs and expenses associated with selling the policy. • We call “gross premium” (or “office premium”) the premium actually charged to the customer. • Often, for GI, we won’t worry about “present value” because all costs are assumed to arise in a short period of time, so we may ignore discounting. If nothing is specified in a question, you may ignore discounting (i.e., assume the interest rate is 0%). • Note those terms are (unfortunately) not universal (you may see elsewhere people calling “pure premium” what we call “risk premium”). ACST1052: Week 11 41 Gross Premiums • As we had previously discussed, insurers need to charge more than just the expected value of the claims, otherwise they would almost certainly become insolvent. • We previously discussed that “loadings” are applied on top of the risk premium to get the final price charged: ― Expense loading ― Profit loading ― Safety loading • Additionally, insurance policies are often sold through insurance brokers, who are paid a commission (which we must include in the calculation of the final “gross premium”). Usually, commissions are stated as a percentage of the gross premium. ACST1052: Week 11 42 Risk Premium Example • Going back to our example (with no deductible or reinsurance)… what would be the risk premium? 𝑛 0 1 2 3 • Pr(𝑁 = 𝑛) 0.90 0.07 0.02 0.01 𝑥 100 Pr(𝑋 = 𝑥) 0.5 200 300 0.3 0.2 Answer: $23.80. ACST1052: Week 11 43 Gross Premium Example • The risk premium for one policy is $23.80, but now let’s assume that: ― The broker commission is 5% of the gross premium ― Claims handling expenses are $10 per claim made ― Other expenses are $2 per policy ― The gross premium needs to provide $4 in safety/profit loading • What is the gross premium (denoted 𝐺)? • Gross Premium = Risk Premium + Expected CHE + Expenses + Profit Loading + Commission 𝐺 = 23.80 + 10 × 0.14 + 2 + 4 + 0.05𝐺 𝐺 = 32.85. ACST1052: Week 11 44
0
You can add this document to your study collection(s)
Sign in Available only to authorized usersYou can add this document to your saved list
Sign in Available only to authorized users(For complaints, use another form )