Why Exposure Comparability Matters
A portfolio can change even when the expected loss cost for a comparable risk does not. It may insure more properties, cover higher values, offer higher limits, or contain a different mix of territories or other risk classes. Those changes can alter total expected losses without representing a change in the expected loss cost of a fixed risk.
The analytical task is to decide whether the exposure measure already reflects the change. If it does not, the analyst may need to adjust or standardize exposures before comparing loss costs. That adjustment is distinct from loss-cost trend: it makes the risks being compared more comparable; it does not itself assert that the cost of a comparable risk has risen or fallen. This distinction is aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.
The basis must suit the question. Policy-years may be appropriate when each policy-year represents a comparable risk. For property risks whose expected losses are assumed proportional to insured value, value-years may be more informative. When the exposure unit is unchanged but coverage or class characteristics differ, a suitable relativity can convert each exposure into equivalent base exposures.
Standardizing Exposure for Risk Characteristics
Suppose a reference class has expected loss cost \(\theta\) per exposure unit. Let class \(k\) have \(E_k\) exposure units and a loss relativity \(r_k\) to the reference class. Under the model that expected losses are proportional to exposure and to the stated relativity, expected losses for that class are \(\theta E_k r_k\). The portfolio’s equivalent base exposure is therefore \(\sum_k E_k r_k\), and its indicated loss cost per equivalent base exposure is total losses divided by that quantity.
Here \(L\) is the loss amount measured on a consistent basis, \(E^*\) is adjusted exposure, and \(r_k=1\) for the reference class. The relativity describes expected loss per raw exposure unit relative to the reference class; it is not automatically the ratio of class premiums or of limits. Its use assumes that the selected factors are appropriate for the loss measure and that the class differences are represented adequately by those factors.
The justification follows directly from the model. If class \(k\) has expected loss \(\theta E_k r_k\), adding across classes gives total expected loss \(\theta\sum_k E_k r_k\). Dividing expected total loss by adjusted exposure yields \(\theta\). Thus, dividing observed losses by adjusted exposure estimates the reference-class loss cost, subject to the model and ordinary variability in observed losses.
Specify the exposure unit and the reference risk. For example, decide whether the measure is policy-years or insured-value-years.
Use an exposure amount, a relativity, or both only when each captures a distinct feature. Do not apply two adjustments for the same change.
Multiply each raw exposure by the applicable relativity and sum the results.
Divide consistent losses by adjusted exposures for each period. Interpret any remaining difference as a difference in observed loss cost, not proof of its cause.
Worked Applications
The examples use hypothetical portfolios. In each, the chosen exposure model is an assumption for the calculation, not a claim that risk is always proportional to a single measure or factor.
Worked Example: Changing Insured Values
A hypothetical property portfolio reports 80 policy-years in Period A, with an average insured value of $200,000 and losses of $1,920,000. In Period B it reports 60 policy-years, with an average insured value of $250,000 and losses of $2,025,000. Assume expected losses are proportional to insured value, and measure exposure in thousands of dollars of insured value per year. Compare the loss costs.
Method and assumption. Use insured-value-years rather than policy-years alone. This is appropriate for this calculation only under the stated proportionality assumption. The unit is $1,000 of insured value for one year.
Period A exposure. The average insured value is \(200\) thousands of dollars, so adjusted exposure is \(80 \times 200=16{,}000\) thousand-dollar-years. The loss cost is \(1{,}920{,}000/16{,}000=\$120.00\) per thousand-dollar-year.
Period B exposure. The average insured value is \(250\) thousands of dollars, so exposure is \(60 \times 250=15{,}000\) thousand-dollar-years. The loss cost is \(2{,}025{,}000/15{,}000=\$135.00\) per thousand-dollar-year.
Comparison and check. The ratio is \(135/120=1.125\), so Period B’s observed loss cost is 12.5% higher on this exposure basis. At Period A’s loss cost, Period B’s expected losses would be \(15{,}000 \times \$120=\$1{,}800{,}000\); the observed amount is $225,000 higher. As a check, the Period A and B loss totals divided by their respective exposure amounts return $120.00 and $135.00.
Common mistake avoided. Dividing only by policy-years gives $24,000 and $33,750 per policy-year, a ratio of 1.40625. That comparison treats a policy with $250,000 insured value as equivalent to one with $200,000 insured value. It does not isolate loss cost per insured-value unit.
Worked Example: Adjusting for a Change in Coverage Limit
A hypothetical insurer uses a $100,000 limit as its reference coverage. Its indicated loss relativities are 1.00 for the $100,000 limit and 1.35 for a $200,000 limit. In Period A, it has 200 policy-years at the reference limit and losses of $216,000. In Period B, it has 120 policy-years at the reference limit and 80 at the higher limit, with losses of $273,600. Use the given relativities to compare loss costs per equivalent reference policy-year.
Method. Keep policy-years as the raw exposure unit and apply the stated loss relativities. Do not treat the limit ratio \(200{,}000/100{,}000=2\) as the loss relativity; the problem supplies 1.35.
Period A. All 200 policy-years are at the reference limit, so adjusted exposure is \(200 \times 1.00=200\). Its loss cost is \(\$216{,}000/200=\$1{,}080.00\) per equivalent reference policy-year.
Period B. The reference-limit policies contribute \(120 \times 1.00=120\) equivalent years. The higher-limit policies contribute \(80 \times 1.35=108\). Total adjusted exposure is \(120+108=228\) equivalent years. The loss cost is \(\$273{,}600/228=\$1{,}200.00\).
Comparison and check. The adjusted loss-cost ratio is \(1{,}200/1{,}080=1.1111\), or an increase of approximately 11.11%. At $1,200 per reference equivalent year, the higher-limit class has modeled expected loss cost \(1.35 \times \$1,200=\$1,620\) per policy-year. The portfolio total at that base cost is \(120 \times \$1,200+80 \times \$1,620=\$273{,}600\), matching the stated losses.
Common mistake avoided. Dividing by the 200 raw policy-years gives $1,368 per policy-year. Comparing that with Period A’s $1,080 mixes a change in limit composition with loss-cost change. Conversely, doubling the higher-limit exposure because the limit doubled would ignore the supplied loss relativity.
Worked Example: Standardizing a Territory Mix
A hypothetical book has two territories. Territory A is the reference class with relativity 1.00; Territory B has relativity 1.50. Period 1 has 100 exposure units in A and 50 in B, with losses of $210,000. Period 2 has 80 units in A and 80 in B, with losses of $260,000. Calculate the adjusted loss costs and distinguish the mix effect from the remaining difference.
Period 1 adjusted exposure. The total is \(100 \times 1.00+50 \times 1.50=100+75=175\) equivalent A units. The loss cost is \(\$210{,}000/175=\$1{,}200.00\) per equivalent A unit.
Period 2 adjusted exposure. The total is \(80 \times 1.00+80 \times 1.50=80+120=200\) equivalent A units. The loss cost is \(\$260{,}000/200=\$1{,}300.00\).
Comparison and check. The adjusted loss-cost ratio is \(1{,}300/1{,}200=1.08333\), an increase of approximately 8.33%. At Period 1’s $1,200 base cost, Period 2’s expected losses under the stated territory model are \(200 \times \$1{,}200=\$240{,}000\); the observed losses exceed that amount by $20,000. Period 1’s calculation also checks: \(175 \times \$1{,}200=\$210{,}000\).
Common mistake avoided. The raw averages are $1,400 per exposure unit in Period 1 and $1,625 in Period 2. Their comparison reflects both the higher share of Territory B in Period 2 and the difference in loss cost indicated after applying the territory factors. Standardizing makes the comparison conditional on the supplied factors; it does not establish that territory is the only cause of the residual difference.
Choosing the Adjustment Carefully
Insured value, coverage limit, and risk class are not interchangeable concepts. If expected losses are assumed proportional to insured value and the exposure is already measured in insured-value-years, do not also multiply by a factor intended to represent that same insured-value change. That would count the change twice. By contrast, if the raw measure is policy-years, a coverage relativity may be needed to reflect how expected covered losses vary by limit.
A limit change should generally be represented by a suitable loss relativity, not by assuming that a 20% increase in the limit creates a 20% increase in expected losses. The additional coverage may affect only a portion of claims. Likewise, an insured-value adjustment is not automatically exact: partial losses, policy terms, concentration of values, and other risk characteristics can make losses depart from simple proportionality. The chosen basis and factors should match the type of losses being analyzed.
Class factors also require care. A factor may describe expected loss differences relative to a base class, but it may not capture every relevant feature. If a portfolio changes in several ways, an adjustment for one characteristic does not automatically control for the others. Keep the loss measure, exposure definition, and factors consistent across periods, and describe what the adjusted comparison does and does not hold constant.
| Portfolio change | Possible treatment | Important check |
|---|---|---|
| More policy-years with otherwise comparable risks | Use the larger exposure total; no class adjustment is implied by quantity alone | Do not call more exposure a higher loss cost |
| Higher insured values | Consider insured-value exposure if proportionality is a reasonable model | Do not also apply a factor for the same value change |
| Different limits or deductibles | Use an appropriate coverage relativity when policy-years alone do not represent comparable covered risk | Do not assume the limit ratio is the loss relativity |
| Different territory or other class mix | Weight class exposures by factors relative to a stated base | Check that the factors match the loss measure and period |
The central distinction is between how much risk was insured and the expected loss cost per comparable unit of risk. An exposure adjustment addresses the first question when the raw units are not comparable. A remaining difference in adjusted loss cost is a finding about observed losses on the chosen basis; by itself, it does not identify the cause or establish a trend in the cost of a fixed risk.
Check Your Understanding
For each hypothetical problem, calculate on the stated basis and distinguish any exposure adjustment from a change in loss cost.
- A small property portfolio has 40 policy-years with average insured value $150,000 and losses of $360,000. A later period has 50 policy-years with average insured value $180,000 and losses of $540,000. Assume losses are proportional to insured value. Calculate loss cost per $1,000 insured-value-year for each period and compare. (Answer: The exposures are 6,000 and 9,000 thousand-dollar-years; both loss costs are $60.00, so there is no change on this basis.)
- A reference class has relativity 1.00 and another class has relativity 1.40. A portfolio has 100 exposure units in the reference class and 50 in the other class, with losses of $190,000. Calculate adjusted exposure and loss cost per reference equivalent unit. (Answer: Adjusted exposure is \(100+50(1.40)=170\); loss cost is \(\$190{,}000/170=\$1{,}117.65\), rounded to the nearest cent.)
- A hypothetical insurer uses the $100,000 limit as its base. Its $50,000 limit relativity is 0.80. It has 100 policy-years at each limit and losses of $180,000. Calculate adjusted exposure and loss cost per equivalent base policy-year. (Answer: Adjusted exposure is \(100(0.80)+100(1.00)=180\); loss cost is \(\$180{,}000/180=\$1{,}000.00\).)
- Territory A has relativity 1.00 and Territory B has relativity 2.00. In Period 1, exposures are 90 in A and 40 in B, with losses of $170,000. In Period 2, exposures are 50 in A and 70 in B, with losses of $209,000. Calculate each adjusted loss cost and its percentage change. (Answer: Adjusted exposures are 170 and 190; loss costs are $1,000.00 and $1,100.00, a 10.00% increase.)
- An analyst measures exposure in insured-value-years and also multiplies each exposure by a relativity specifically intended to represent the same insured-value differences. What is the issue, and what should be checked before using both adjustments? (Answer: The insured-value difference may be counted twice. Check whether the relativity captures a distinct risk characteristic; if it represents the same value change already reflected in the exposure measure, use only one adjustment for that change.)