Introduction
A warehouse fire, a severe storm damaging insured buildings, a vehicle accident, and an injury claim from a customer can all threaten an organization’s objectives through loss. These events differ in cause and consequence, but each can create hazard risk: potential loss arising from an event that causes physical damage, bodily injury, or legal liability. For insurers, such risks may arise from policies they cover; for other businesses, they may arise from their own property, activities, or interactions with the public.
The classification is useful because it gives an organization a consistent way to recognize exposures before assessing them. The focus here is on identifying hazard risks and understanding simple calculations for hypothetical events. The treatment is aligned with the CAS Exam 9 (Risk Management for Actuaries) syllabus.
Recognizing Hazard Risk
Hazard risk is not limited to catastrophic events. A small kitchen fire, a worker injury, or a single customer’s slip-and-fall claim may also be hazard-related. The key question is whether a possible adverse event can cause physical damage, bodily injury, or liability. The loss may be paid by an insurer, retained by a business, or shared through a contractual arrangement; who ultimately pays does not by itself change the event’s hazard character.
For an insurer, examples include potential property claims from fire or wind, vehicle claims following collisions, and liability claims alleging injury or property damage. For a non-insurance business, examples include damage to its own equipment, injury to an employee, or a claim from a visitor injured on its premises. A business can face hazard risk even when it has purchased insurance: coverage may not eliminate every loss, and an insured event can still disrupt operations or affect customers.
Not every adverse change is a hazard risk. For example, a change in the value of an investment is not, by itself, a fire, accident, injury, or liability event. The distinction here is about the source of potential loss, not whether the consequence is financially important. A claim’s dollar amount is a financial measure of its impact; the event that produced the claim may still be hazard-related.
Terminology can vary by context. In enterprise risk classification, “hazard risk” commonly names a broad category of potential losses from events like those above. In insurance usage, “peril” may refer to the cause of loss, while a “hazard” can also mean a condition that increases the likelihood or severity of loss. In this tutorial, hazard risk refers to the enterprise category defined above; the surrounding facts should make the intended meaning clear.
Classifying Representative Exposures
Classification starts with the event and its possible impact, not merely the name of a department, product, or policy. A property insurer may have hazard exposure because a covered building could burn. A retailer may have hazard exposure because a customer could be injured in its store. In both cases, the organization’s objective could be affected through damage, injury, or a resulting claim.
A product name alone is not enough to classify a risk. A commercial auto policy can involve hazard risk when a collision causes injury or vehicle damage. In contrast, a change in investment value is not a hazard event just because it affects an insurer. When a scenario contains several possible causes, identify which specific event and consequence you are classifying rather than assigning one label to the whole organization.
| Hypothetical scenario | Classification | Reason |
|---|---|---|
| A storm damages a business’s roof | Hazard risk | A physical event may damage property |
| A customer is injured in a shop and makes a claim | Hazard risk | An accident may cause bodily injury and liability |
| A delivery vehicle collides with another vehicle | Hazard risk | An accident may cause injury or property damage |
| The market value of an investment changes | Not hazard risk in this classification | The potential loss does not arise from the hazard events defined here |
Quantifying Simple Hazard Scenarios
The earlier material on frequency, severity, and variability provides a useful starting point: describe how often an event may occur and how large its loss may be. For a simple one-period model, let \(I_j\) equal 1 if event \(j\) occurs during the period and 0 otherwise, and let \(c_j\) be its fixed loss if it occurs. The gross loss from that event is \(c_j I_j\). If the total loss is the sum of the event losses, its expected value is the sum of the expected event losses.
This result follows from linearity of expectation. Since \(I_j\) is 1 with probability \(P(I_j=1)\) and 0 otherwise, \(E[I_j]=1\cdot P(I_j=1)+0\cdot(1-P(I_j=1))=P(I_j=1)\). Therefore \(E[c_j I_j]=c_jP(I_j=1)\), and adding the expected values gives the stated expression. Independence is not required to add expected losses. Independence is needed for some joint probability calculations, such as multiplying probabilities to find that none of several events occurs.
These formulas assume the probabilities and loss amounts refer to the same defined period, such as one year, and that the stated losses are gross amounts before any insurance recovery, deductible, or other adjustment. If event severities vary, the calculation must also account for the possible severities conditional on an event. The following examples state their assumptions explicitly.
Worked Example: Fire and Liability Losses at a Hypothetical Business
A hypothetical business assesses two possible losses over the coming year. A fire occurs with probability \(0.020\) and, if it occurs, causes a fixed gross loss of \(\$150{,}000\). A liability claim occurs with probability \(0.005\) and, if it occurs, causes a fixed gross loss of \(\$80{,}000\). Assume the two events are independent, either or both can occur, and losses add if both occur. Find the probability of no loss and the expected gross loss.
Method and assumptions. Treat each event as a Bernoulli indicator for this one-year period. The fixed loss amounts apply whenever their corresponding events occur. Independence is used to calculate the probability of no loss; linearity of expectation is used for expected loss.
Probability of no loss. The probability of no fire is \(1-0.020=0.980\). The probability of no liability claim is \(1-0.005=0.995\). By independence, the probability that neither occurs is
Thus the probability of no loss is \(0.9751\), or \(97.51\%\).
Expected gross loss. The expected fire loss is \(0.020(\$150{,}000)=\$3{,}000\). The expected liability loss is \(0.005(\$80{,}000)=\$400\). Adding them gives
Check and common mistake. The liability probability is \(0.005\), or \(0.5\%\), not \(0.05\), or \(5\%\). Using \(0.05\) would give the incorrect calculation \(0.020(\$150{,}000)+0.05(\$80{,}000)=\$7{,}000\). Using the stated probability gives \(\$3{,}400\). The no-loss probability is less than 1, as expected when either event can cause a loss.
Several Similar Exposures
When multiple locations or assets face the same kind of hazard, keep the event probability and loss amount tied to each exposure. If the locations are independent and have identical probabilities, the probability that all avoid an event can be found by multiplying the individual no-event probabilities. The expected aggregate loss can be found by summing the expected losses location by location. A common probability does not mean that the locations necessarily suffer one shared loss; that depends on the assumptions about dependence and the event definition.
Worked Example: Storm Damage Across Hypothetical Shops
A hypothetical retailer operates three shops. In one year, each shop has probability \(0.040\) of a storm causing a fixed gross loss of \(\$25{,}000\). Assume storm events at the three shops are independent, and each shop can have at most one such loss during the year. Find the probability of at least one loss and the expected aggregate gross loss.
Method and assumptions. Use the complement of the event that all shops avoid storm damage to find the probability of at least one loss. Sum the expected losses for the three shops to find the expected aggregate.
For any one shop, the no-loss probability is \(1-0.040=0.960\). Independence gives the probability of no losses at all three shops:
Therefore, the probability of at least one loss is
For expected loss, one shop has expected gross loss \(0.040(\$25{,}000)=\$1{,}000\). For three shops, the expected aggregate is \(3(\$1{,}000)=\$3{,}000\).
Check and common mistake. The result is a probability of \(0.1153\), or \(11.53\%\), to four decimal places. Simply adding the three event probabilities gives \(0.1200\), which is not the exact probability of at least one event because outcomes with losses at multiple shops would be counted more than once. The expected aggregate loss is \(\$3{,}000\); this sum of expectations does not require independence.
When Severity Varies
A hazard event need not produce the same loss every time. If the loss amount depends on the event’s severity, first calculate the average loss conditional on the event, then account for the probability that the event occurs. The conditional severity outcomes must cover the possibilities assumed in the model and their probabilities must sum to 1. The resulting expected amount is still a gross loss if no recoveries or policy terms have been applied.
Worked Example: Variable Loss From a Hypothetical Wildfire Event
A hypothetical property owner models one year of wildfire exposure. The probability of a wildfire causing damage to the insured property is \(0.010\). Conditional on such damage, the gross loss is \(\$100{,}000\) with probability \(0.60\), \(\$400{,}000\) with probability \(0.30\), or \(\$1{,}000{,}000\) with probability \(0.10\). Assume there can be no more than one such event in the year, and there is no loss if the event does not occur. Find the expected gross loss for the year.
Method and assumptions. First find the expected severity conditional on the event. Then multiply by the probability of the event. The conditional severity probabilities sum to \(0.60+0.30+0.10=1.00\), so they describe all modeled outcomes given damage.
The expected gross severity given the event is
The event occurs with probability \(0.010\); otherwise the loss is zero. Thus
Check and common mistake. The expected annual loss, \(\$2{,}800\), is lower than the conditional expected severity, \(\$280{,}000\), because the event occurs only \(1\%\) of the time in this model. Multiplying the event probability by only one of the possible severities would ignore the other conditional outcomes.
Practical Classification and Common Traps
A sound classification should name the event and the exposure in context. “Property risk” by itself may be too vague: a building could be exposed to fire, severe weather, or another physical event, and each may have a different probability and consequence. Likewise, “liability” is clearer when the possible event is described, such as a visitor being injured and making a claim. The detailed construction of risk statements and taxonomy boundaries is a separate topic; for present purposes, the event and its hazard-related consequence are enough to identify the category.
- Confusing the event probability with the severity. A 2% probability is \(0.020\), not \(0.20\). Keep the probability and the dollar loss in their proper roles.
- Using independence without stating it. Multiplying separate no-event probabilities is justified only when the events are independent under the model.
- Assuming expected losses require independence. Adding expected losses is valid by linearity of expectation even if the events are dependent, provided the event-specific expected losses are correctly specified.
- Classifying by who pays rather than what happens. Insurance may transfer part of a loss, but a fire or accident can remain a hazard event for the organization exposed to it.
- Treating every adverse financial consequence as a hazard event. First identify the source of potential loss. A dollar-valued impact alone does not make the underlying event hazard-related.
Check Your Understanding
For each hypothetical situation, identify the hazard classification or calculate the requested quantity using the assumptions given. Report probabilities to four decimal places where applicable and dollar amounts to the nearest cent.
- A small bakery faces a possible oven fire that could damage equipment, and its owner also faces a possible change in the value of an investment account. Which exposure is hazard-related, and why? (Answer: The oven fire exposure is hazard-related because it may physically damage property; the investment-value change is not a hazard event in this classification.)
- Over one year, a hypothetical workshop has an independent probability \(0.030\) of a fire causing a fixed gross loss of \(\$40{,}000\), and probability \(0.010\) of an accident claim causing a fixed gross loss of \(\$10{,}000\). Find the probability of no loss and the expected gross loss. (Answer: No-loss probability \(=0.97(0.99)=0.9603\); expected gross loss \(=0.030(\$40{,}000)+0.010(\$10{,}000)=\$1{,}300\).)
- A hypothetical company has four independent storefronts. Each has probability \(0.020\) of severe-weather damage in a year, with a fixed gross loss of \(\$60{,}000\) if damaged. Find the probability of at least one damaged storefront and the expected aggregate gross loss. (Answer: Probability \(=1-0.98^4=0.0776\); expected aggregate gross loss \(=4(0.020)(\$60{,}000)=\$4{,}800\).)
- A hypothetical insurer models a liability event with annual probability \(0.015\). Conditional on the event, gross claim severity is \(\$50{,}000\) with probability \(0.50\), \(\$200{,}000\) with probability \(0.35\), and \(\$500{,}000\) with probability \(0.15\). There is no loss if no event occurs. Find the conditional expected severity and the expected annual gross loss. (Answer: Conditional expected severity \(=0.50(\$50{,}000)+0.35(\$200{,}000)+0.15(\$500{,}000)=\$170{,}000\); expected annual gross loss \(=0.015(\$170{,}000)=\$2{,}550\).)