Concentration Limit Monitoring Workshop

Why Concentration Limits Need Careful Aggregation

A concentration limit is useful only when the exposure being measured is defined consistently and grouped at the levels the limit was designed to constrain. A portfolio can be within its overall limit and still exceed a limit for one location; it can also appear to exceed its total limit if location totals and product totals are incorrectly added together. These are different questions and require different calculations.

This workshop uses hypothetical exposures stated in thousands of dollars. Each example specifies a single valuation date and an exposure basis, such as gross insured value. In practice, a monitoring process should keep the basis, valuation date, currency, and population of covered risks aligned with the approved limit. The calculations do not determine how a limit should be set; they show how to test exposures against limits once those choices have been made. This approach is aligned with the CAS Exam 9 (Risk Management for Actuaries) syllabus.

Definition: An exposure is the measured amount of risk included in a specified monitoring population, using a stated basis and date. A concentration limit is a maximum permitted exposure for a defined group, such as a location, product segment, or parent group. A limit is breached when the measured exposure is greater than the applicable maximum. An exposure exactly equal to its limit is at the limit, but is not a breach under this convention.

Define the Exposure and the Grouping

Before adding numbers, decide what each number represents. For example, a hypothetical property portfolio might use gross insured values in force at the close of a specified date. Another monitoring measure might instead use net exposure after a stated risk transfer. Those measures are not interchangeable: a limit calibrated to one basis cannot be tested by silently substituting the other.

For a portfolio divided into locations \(i\) and product segments \(j\), let \(x_{ij}\) be the exposure in the location-product cell. If every exposure belongs to exactly one location and exactly one product segment, the total exposure \(T\), location total \(L_i\), and product total \(P_j\) are:

L_i=\sum_j x_{ij}, \qquad P_j=\sum_i x_{ij}, \qquad T=\sum_i\sum_j x_{ij}=\sum_i L_i=\sum_j P_j.

The location totals and product totals are two alternative summaries of the same portfolio. They are not two separate portfolios to add together. Each cell contributes once to the total, even though it appears in both a location subtotal and a product subtotal.

Limits can apply at several levels. For instance, there may be a total portfolio limit, a limit for each region, a limit for each site within a region, and separate limits for product segments. Each constraint must be tested against its own corresponding exposure total. Passing a parent limit does not guarantee that every child limit passes, and passing all site limits does not guarantee that a regional limit passes.

Definition: A nested limit structure applies limits to groups that contain, or are contained within, other groups. A regional exposure is the sum of the exposures assigned to its child locations. When location and product groupings cross one another, each dimension defines its own set of totals; the totals from different dimensions must not be combined as if they were disjoint exposure groups.

A Reliable Monitoring Sequence

A repeatable calculation helps prevent errors in both the exposure totals and the decision. The following sequence separates those tasks.

Fix the measurement basis.
State the date, currency, exposure measure, and included population. Confirm that these match the basis of the limits.

Assign exposure to cells.
Place each exposure in one location-product cell under the approved classification rules. Do not count the same underlying exposure twice within the same partition.

Calculate totals independently by dimension.
Sum cells to obtain location totals, product totals, and any parent-group totals. Reconcile each complete partition to the same portfolio total.

Test every applicable limit.
Compare each exposure only with its corresponding limit, using the stated breach rule. Record the calculation and result for each level.

Worked Examples

Worked Example: Testing Location and Product Limits

A hypothetical insurer measures gross insured values in force on one reporting date, in thousands of dollars. Its portfolio has three locations and three product segments. The cell exposures are:

Location	Auto	Property	Liability	Location total
East	18	12	10	?
West	8	20	12	?
Central	14	6	10	?
Product total	?	?	?	?

The total portfolio limit is 120. Location limits are East 38, West 42, and Central 35. Product limits are Auto 42, Property 40, and Liability 35. Determine whether any limits are breached.

Method: calculate cell sums, then test each margin against its own limit. First calculate the location totals:

L_{\text{East}}=18+12+10=40,\qquad L_{\text{West}}=8+20+12=40,\qquad L_{\text{Central}}=14+6+10=30.

Next calculate the product totals by summing down each column:

P_{\text{Auto}}=18+8+14=40,\qquad P_{\text{Property}}=12+20+6=38,\qquad P_{\text{Liability}}=10+12+10=32.

The portfolio total from the location margins is \(40+40+30=110\). The check from the product margins is \(40+38+32=110\), so both partitions reconcile. The total portfolio exposure, 110, is below its limit of 120. East is over its location limit because \(40>38\). West is below its limit because \(40\leq42\), and Central is below its limit because \(30\leq35\). All product totals are within their limits: \(40\leq42\), \(38\leq40\), and \(32\leq35\).

Decision: The East location limit is breached; the total, other location, and product limits are not. Check: both independent margin calculations give a portfolio total of 110. A common mistake is to conclude that the portfolio is compliant because the overall total passes, or because each product passes. Neither conclusion tests the East location constraint.

Worked Example: Testing Nested Regional and Site Limits

A hypothetical portfolio records gross insured values in thousands of dollars for two regions, each containing two sites, and two product segments. The exposures on the reporting date are:

Region	Site	Product A	Product B	Site total
North	N1	16	12	?
North	N2	14	10	?
South	S1	18	8	?
South	S2	12	10	?

The total portfolio limit is 105. The North and South regional limits are 50 and 50. Each site has a limit of 30. Product A and Product B have limits of 65 and 45, respectively. Determine whether any limit is breached.

Method: aggregate the child sites to regions, and calculate product totals separately. The site totals are:

N1=16+12=28,\quad N2=14+10=24,\quad S1=18+8=26,\quad S2=12+10=22.

The regional totals are \(28+24=52\) for North and \(26+22=48\) for South. The product totals are \(16+14+18+12=60\) for Product A and \(12+10+8+10=40\) for Product B. Thus the portfolio total is \(52+48=100\); the product-side reconciliation is \(60+40=100\).

The total portfolio passes because \(100\leq105\). North breaches its regional limit because \(52>50\), while South passes because \(48\leq50\). Every site passes: the largest site exposure is 28, which is below the site limit of 30. Both product segments pass because \(60\leq65\) and \(40\leq45\).

Decision: The North regional limit is breached, although the portfolio, all sites, South, and both product segments are within their limits. Check: the regional totals and product totals each reconcile to 100. A common mistake is to test only the site limits and infer that the North region passes; the regional limit applies to the sum of its sites, not to each site individually.

Worked Example: Avoiding a False Total Breach From Overlapping Views

A hypothetical portfolio has gross exposures in thousands of dollars on a single reporting date. It is divided into Coastal and Inland locations and Property and Auto products:

Location	Property	Auto	Location total
Coastal	24	6	?
Inland	16	14	?
Product total	?	?	?

The portfolio limit is 65. Coastal and Inland limits are 28 and 35. Property and Auto limits are 42 and 24. Determine the results and assess a proposed calculation that adds all location totals to all product totals.

Method: reconcile each partition to the portfolio once, then test its limits. Coastal totals \(24+6=30\), and Inland totals \(16+14=30\), giving a portfolio total of \(30+30=60\). Property totals \(24+16=40\), and Auto totals \(6+14=20\), also giving \(40+20=60\). The portfolio total passes because \(60\leq65\). Coastal breaches because \(30>28\); Inland passes because \(30\leq35\). Property passes because \(40\leq42\), and Auto passes because \(20\leq24\).

The proposed calculation gives \(30+30+40+20=120\). This does not measure portfolio exposure: each cell has been counted once in a location total and again in a product total. The correction is to use either complete partition to calculate the portfolio total, giving 60, and then test location and product limits independently.

Decision: Coastal breaches its limit; the portfolio, Inland, Property, and Auto limits pass. Check: the two valid portfolio reconciliations both equal 60. The invalid combined sum, 120, is not evidence of a total portfolio breach because its components overlap.

Why the Reconciliation Works

The equality between the sum of location totals and the portfolio total follows from how the cells are assigned. Suppose each cell belongs to one and only one location. Summing all location totals expands into the sum of every cell once:

\sum_i L_i =\sum_i\left(\sum_j x_{ij}\right) =\sum_i\sum_j x_{ij} =T.

If every cell also belongs to one and only one product segment, summing product totals gives the same result:

\sum_j P_j =\sum_j\left(\sum_i x_{ij}\right) =\sum_i\sum_j x_{ij} =T.

The equality does not mean that \(\sum_i L_i+\sum_j P_j\) equals \(T\). That sum counts each cell twice: once through its location and once through its product. This is why reconciliations should compare each full partition with the portfolio total rather than add overlapping partitions together.

Absolute Limits and Percentage Limits

An absolute limit specifies a maximum amount, such as 30 million of exposure in a location. A percentage limit specifies a maximum share of a defined denominator. If a group has exposure \(G\), the portfolio denominator is \(T\), and the permitted share is \(c\), the share test is:

\frac{G}{T}\leq c \qquad\text{or, equivalently,}\qquad G\leq cT,

The denominator must be the one defined by the limit. If a limit is stated as a percentage of total gross exposure, a net exposure denominator is not a valid substitute without an explicit rule authorizing it. Likewise, a group share is not calculated by adding shares across overlapping location and product dimensions.

Common Monitoring Traps

Testing only the portfolio total. A passing total does not establish that every location, product, or nested parent group passes.
Testing children but not parents. Several individually compliant sites can sum to a region that breaches its own limit.
Adding overlapping summaries. Location totals and product totals each represent the same underlying portfolio and must not be added together.
Changing the exposure basis. Gross and net exposures, different valuation dates, or different covered populations can produce incomparable results.
Using an unspecified denominator. A percentage limit is not testable until its portfolio base and measurement date are clear.
Assuming equality is a breach. Under the convention used here, a breach requires exposure to be strictly greater than the limit. Follow a different documented rule if the governing limit explicitly defines one.

Key takeaway: Define the exposure basis, aggregate each complete grouping separately, reconcile each partition to the same portfolio total, and compare every group only with its own limit. A portfolio-wide pass and a subgroup breach can both be true at the same time.

Check Your Understanding

For each hypothetical scenario, calculate the relevant totals and apply the stated breach convention: exposure breaches a limit only when it is greater than the limit.

A portfolio has Harbor exposures of 9 and 7 in Products X and Y, and Valley exposures of 5 and 11 in those products. The total limit is 35, Harbor's limit is 15, and Product X and Product Y limits are 15 and 20. Which limits are breached? (Answer: Harbor only; total is 32, Product X is 14, Product Y is 18, and Harbor is 16.)
In Region A, Site A1 has Product X exposure 12 and Product Y exposure 8; Site A2 has 9 and 6. Region B has one site with exposures 10 and 7. The total limit is 55, Region A's limit is 34, each site's limit is 22, and the Product X and Y limits are 32 and 23. Which limits are breached? (Answer: Region A only; its total is 35, the portfolio total is 52, all site totals are at most 20, Product X is 31, and Product Y is 21.)
A portfolio has Location 1 exposures of 13 in Product A and 7 in Product B, and Location 2 exposures of 8 and 12. The portfolio limit is 42, location limits are 19 and 21, and product limits are 22 and 20. Identify all breaches and reconcile the portfolio. (Answer: Location 1 only; location totals are 20 and 20, product totals are 21 and 19, and both partitions reconcile to 40.)
A portfolio has total exposure 80, of which Zone Q represents 22 and Product R represents 18. Each group is subject to a limit of 25% of total portfolio exposure. Which group breaches? (Answer: Zone Q only; each limit is \(0.25(80)=20\), so Zone Q's 22 breaches and Product R's 18 passes.)
A portfolio has Location A exposures of 14 in Product M and 28 in Product N, and Location B exposures of 16 and 17. The total limit is 80, location limits are 40 and 35, and product limits are 35 and 46. A report adds the location totals and product totals before comparing with the total limit. Correct the report and identify breaches. (Answer: Location A breaches at 42; the portfolio total is 75, Location B is 33, Product M is 30, and Product N is 45, so all other stated limits pass. Adding the two partitions gives 150 and double-counts every exposure.)

What You'll Learn