1.4 The Components of Risk: Frequency and Severity
Key Takeaways
- The components of a risk are the peril (the cause of loss), the hazard (the condition that makes the peril more likely or more serious), and the subject matter of insurance together with the insured's interest in it
- Frequency is how often losses of a given type occur in a period; severity is how much each loss costs when it does occur
- Expected loss cost = frequency x severity, and this product — not either figure alone — is what an underwriter prices
- The frequency/severity matrix drives risk treatment: low/low is retained, high frequency/low severity is controlled and retained, low frequency/high severity is the classic insurance buy, and high/high is usually avoided or made uninsurable
- Increasing an excess removes small, high-frequency claims from the policy, so it cuts claims-handling cost and premium without changing the insurer's exposure to a large loss
Learning outcome 1 of the IF1 syllabus carries 9 questions, and two of its sub-topics deal with the anatomy of a risk rather than its classification: the components of risk, and the relationship between frequency and severity. Both are recall-and-reason topics that also underpin later chapters on underwriting, premium and the condition of average.
The Components of a Risk
Section 1.3 introduced peril and hazard. Put together with the thing being insured, they give the three components an underwriter needs before a risk can be assessed at all.
| Component | What it is | Example (a bakery) |
|---|---|---|
| Peril | The event or cause that produces the loss | Fire |
| Hazard | The condition that makes the peril more likely to occur, or the resulting loss more serious | Flour dust in the air, an ageing gas oven, no sprinkler system |
| Subject matter of insurance | The property, liability, life or interest exposed to the peril | The bakery building, its ovens, stock and the owner's liability to customers |
The insured's insurable interest in that subject matter (Chapter 5) is what turns the exposure into something that can lawfully be insured. Strip any component away and there is nothing to underwrite: a peril with no subject matter causes no loss to anyone, and subject matter with no peril faces no loss at all.
Hazard splits into the two familiar types:
- Physical hazard — a measurable, tangible feature of the risk itself: construction materials, the age of the wiring, the storage of flammable stock, a claimant's occupation.
- Moral hazard — an attitude or characteristic of the people involved: carelessness, a poor claims record, exaggeration, dishonesty, or an owner who is indifferent to loss because the property is fully insured.
Frequency and Severity — The Two Measurements
Once a risk has been identified, it is measured on two independent dimensions.
- Frequency (sometimes called the chance or probability of loss) is how often losses of that type occur over a defined period — for example, 12 windscreen claims per 1,000 vehicles per year.
- Severity is how much each loss costs when it does happen — for example, an average windscreen claim of £320.
The two are independent. A risk can be frequent and cheap (windscreen chips), rare and ruinous (a total factory fire), or any combination in between. Confusing the two is a classic exam trap: a question that says a peril "rarely happens" tells you about frequency and says nothing about how expensive it would be.
Why the product matters
The figure an underwriter actually prices is neither number alone but their product:
Expected loss cost = frequency x severity
Worked example. An insurer writes 5,000 identical delivery vans.
- Windscreen damage: frequency 120 claims a year, average severity £320. Expected cost = 120 x £320 = £38,400.
- Total vehicle fire: frequency 2 claims a year, average severity £24,000. Expected cost = 2 x £24,000 = £48,000.
The dramatic peril is 60 times rarer, but it costs the insurer more in total. That expected loss cost, plus expenses, reinsurance, capital cost and profit margin, is what the premium has to cover. It also explains why insurers gather data on both dimensions separately: a change in frequency (a bad winter) and a change in severity (parts inflation) require different pricing responses.
Attritional and large losses
Practitioners split the two dimensions into attritional losses — the steady stream of small, predictable, high-frequency claims that a portfolio produces every year — and large losses, which are rare, volatile and often reinsured. Attritional losses are largely a budgeting exercise; large losses are the reason insurers hold capital and buy reinsurance at all.
The Frequency/Severity Matrix
Because the two dimensions are independent, they combine into four quadrants, and each quadrant points to a different risk-treatment decision (Chapter 2).
| Low severity | High severity | |
|---|---|---|
| Low frequency | Retain. Not worth the cost of insuring — e.g. a broken office chair. | Insure. The classic insurance purchase: rare but potentially ruinous — e.g. factory fire, employers' liability. |
| High frequency | Control and retain. Predictable enough to budget for; reduce by loss control and absorb through an excess — e.g. minor motor damage in a fleet. | Avoid or reduce. Often uninsurable or unaffordable; the activity itself usually has to change — e.g. storing incompatible chemicals together. |
Two conclusions follow, and both are examinable:
- Insurance is at its most efficient in the low-frequency, high-severity quadrant. That is precisely where an individual cannot budget for the loss but a pool of similar exposures can (Section 1.5).
- High frequency and high severity generally fails the insurability tests. If a loss is close to certain and large, the premium approaches the loss itself, and the insurer is not spreading risk, merely pre-collecting it.
How Insurers Use the Two Dimensions
Excesses and deductibles attack frequency. Raising a motor excess from £100 to £500 removes most small claims from the policy: the insurer avoids both the payment and the disproportionate handling cost of a minor claim, and the premium falls. The insurer's exposure to a severe loss is unchanged — the excess is a small proportion of a £250,000 liability claim.
Policy limits, sums insured and reinsurance attack severity. A limit of indemnity caps how much a single loss can cost the insurer; excess-of-loss reinsurance (Section 2.7) transfers the layer above a retained amount.
The law of large numbers works on frequency, not severity. Adding more similar exposure units makes the number of losses more predictable. It does not make an individual loss smaller, which is why a portfolio of 100,000 homes in a single flood plain is still a dangerous accumulation however large the pool.
Applying It in the Exam
A typical scenario gives two perils with different profiles and asks which the insured should insure, retain or control. Work through it in order: identify the peril, the hazard that is aggravating it, and the subject matter; then estimate frequency and severity separately; then place the risk in the matrix. That sequence answers both the classification questions on learning outcome 1 and the applied risk-management questions on learning outcome 2.
An insurer's fleet portfolio produces 400 minor bodywork claims a year at an average cost of £480, and 3 total-loss fire claims a year at an average cost of £41,000. Which statement is correct?
A commercial fleet operator increases the policy excess on each vehicle from £100 to £750. What is the main effect on the risk transferred to the insurer?