Why a Link Ratio Can Change for Reasons Other Than Development
A link ratio compares cumulative losses at two successive development ages. Its value reflects the development pattern, but it can also reflect the calendar years in which those losses were recorded. Inflation, a change in claim-handling procedures, or a settlement initiative may affect one calendar period disproportionately. If the resulting link ratio is treated as an ordinary, repeatable development factor, projections may carry a temporary distortion into other accident periods.
The key diagnostic question is not simply whether a link ratio is high or low. It is whether the change is consistent with the development age, or whether it aligns more closely with a calendar-year event or a change in operations. A triangle and a plot of link ratios can help locate the signal, but they cannot establish its cause by themselves.
For example, the 12-to-24-month interval for accident year 2022 is associated with calendar year 2023. The 24-to-36-month interval for that accident year is associated with calendar year 2024. Conventions for labeling accident and calendar years should be checked against the triangle being analyzed, especially where periods are not annual.
How Calendar-Year and Operational Effects Appear
A calendar-year effect applies to losses emerging or being recorded in a particular calendar period, potentially across accident years and development ages. General claim-cost inflation can raise the nominal amount of later payments relative to earlier payments. If that increase affects the incremental losses in a link interval, the observed ratio may rise even when the underlying real-dollar development pattern has not changed.
Operational effects can also change the timing or measurement of reported losses. A settlement initiative may accelerate payments: more is paid in one period, with less remaining to be paid later. A new claim system, revised coding, a change in reporting practices, or different claim-handling capacity may affect the amounts or timing recorded in a triangle. These effects need not alter ultimate losses by the same amount as they alter a particular link ratio.
These mechanisms are clues, not automatic explanations. A high ratio in a calendar-year diagonal does not prove inflation or an operational change caused it. Accident-year mix, claim severity, coverage changes, and random variation can also matter. Diagnosis should combine the triangle pattern with information about claim costs, procedures, and the timing of known changes.
Why Inflation Can Affect a Cumulative Ratio
Consider a simplified model in which losses emerge in two successive increments. Suppose the first increment is 100 in base-price dollars and the second is also 100 in real terms. Without inflation, cumulative losses move from 100 to 200, giving a link ratio of 2.00. If the second increment is paid at a price level 10% higher, it is recorded as 110 in nominal dollars. The observed cumulative amount becomes 210 and the nominal link ratio is 2.10.
The ratio rises because the numerator includes the later, inflated increment while the denominator contains the earlier cumulative amount. This example isolates the effect; actual cumulative losses usually contain multiple increments with different emergence dates and price levels.
When the relevant inflation adjustment is known, a diagnostic comparison can restate incremental losses at a common price level. Deflate the affected incremental amount, then add it to the cumulative amount already expressed at the chosen base level. Deflating only the later cumulative total by one factor is generally incorrect when that total combines losses from multiple calendar years.
Worked Example: Separating a Calendar-Year Price Effect
A hypothetical liability portfolio has cumulative paid losses of $100,000 at 12 months for one accident year. By 24 months, cumulative paid losses are $210,000. The additional $110,000 emerged in a calendar year with a 10% price-level increase relative to the chosen base. A comparable real-dollar development pattern has a first increment of $100,000 and a second increment of $100,000. Find the observed ratio and the ratio after restating the second increment at the base price level.
Method and assumptions. Treat the amounts as deterministic paid losses, use a 10% price-level factor for the second increment, and do not discount. No mortality assumption or probability distribution is used. The $110,000 increment is assumed to be fully affected by this price change.
Step 1: Calculate the nominal incremental loss. The observed increment is $210,000 − $100,000 = $110,000.
Step 2: Calculate the observed nominal link ratio. Divide the 24-month cumulative amount by the 12-month cumulative amount: $210,000 ÷ $100,000 = 2.10.
Step 3: Restate the incremental loss. At the base price level, the increment is $110,000 ÷ 1.10 = $100,000.
Step 4: Calculate the restated cumulative amount and ratio. The restated cumulative amount is $100,000 + $100,000 = $200,000. The restated ratio is $200,000 ÷ $100,000 = 2.00.
Check and trap avoided. The restated increment matches the stated real-dollar increment, and the ratio matches the no-inflation benchmark of 2.00. Dividing the entire $210,000 cumulative amount by 1.10 would give $190,909.09, incorrectly applying the later-period adjustment to the earlier $100,000 as well.
Settlement Initiatives and the Timing of Paid Losses
A settlement initiative may change the timing of paid development without changing the eventual total paid. A temporary increase in payments can raise a link ratio at one age, while the subsequent ratio may be lower because some payments have already been made. This pattern is different from a permanent increase in claim costs: the latter can raise ultimate amounts, while a pure acceleration shifts amounts between periods.
The comparison is strongest when the affected calendar period and the following emergence are both observable. If the triangle ends before later development is visible, an apparent acceleration may be difficult to distinguish from a change in ultimate severity. A documented change in settlement strategy supports the interpretation, but the analyst should still check whether the observed amounts and timing are consistent with the proposed mechanism.
Worked Example: Detecting a Temporary Payment Acceleration
A hypothetical portfolio normally has cumulative paid losses of $100,000 at 12 months, $160,000 at 24 months, and $230,000 at 36 months. A settlement initiative moves $30,000 of payments from the 24-to-36-month period into the 12-to-24-month period. Calculate the link ratios before and after the initiative, and assess whether the ultimate paid amount changes.
Method and assumptions. Use deterministic cumulative paid losses and assume the initiative only changes payment timing. There is no interest or discounting, mortality assumption, or probability distribution. The full $30,000 shifted earlier would otherwise have been paid by 36 months.
Step 1: Calculate the baseline link ratios. From 12 to 24 months, the ratio is $160,000 ÷ $100,000 = 1.60. From 24 to 36 months, it is $230,000 ÷ $160,000 = 1.4375.
Step 2: Recalculate cumulative paid losses after the shift. At 24 months, cumulative paid losses increase by $30,000, from $160,000 to $190,000. At 36 months, the same $30,000 has merely been paid earlier, so the cumulative amount remains $230,000.
Step 3: Calculate the affected link ratios. The 12-to-24-month ratio is $190,000 ÷ $100,000 = 1.90. The 24-to-36-month ratio is $230,000 ÷ $190,000 = 1.210526, or 1.2105 to four decimal places.
Step 4: Check the implied increments and ultimate. Before the initiative, increments are $60,000 and $70,000 after the first $100,000. Afterward, they are $90,000 and $40,000. Both sets of increments total $130,000 after month 12, so 36-month cumulative paid losses remain $230,000.
Check and trap avoided. The first ratio rises and the next falls, while the cumulative amount at 36 months is unchanged. Treating 1.90 as a permanent development pattern would miss the reduced later emergence and could overstate projected paid losses.
Diagnosing a Suspicious Pattern
A practical diagnosis starts by locating the affected cells in both dimensions of the triangle. Ask whether unusual ratios cluster by development age, by calendar year, or around a documented operational change. A distortion that follows a calendar-year diagonal is a reason to investigate a calendar effect; it is not, by itself, a causal conclusion. A sharp change in ratios around a known handling initiative may be more consistent with an operational effect, particularly if the subsequent emergence has the expected direction.
Identify the accident year, development interval, and calendar year associated with the incremental emergence. Compare the observation with other ratios at the same age and with other cells in the same calendar-year diagonal.
Check whether a known inflation change, settlement effort, claim-system conversion, or process revision occurred in the relevant period. State what pattern that explanation predicts.
For inflation, assess whether the affected increments are higher in nominal terms than comparable amounts at a common price level. For accelerated payments, look for an earlier increase followed, where observable, by reduced later emergence.
Determine whether including the observation would treat a temporary effect as repeatable development. Document the evidence and uncertainty before deciding whether it belongs in a factor selection.
Paid and incurred link ratios can provide complementary evidence because they measure different cumulative quantities. A timing change in payments may be more visible in paid development than in incurred development, whereas changes to recorded claim estimates may affect incurred development. The pattern should be interpreted using the measure actually being analyzed; an explanation established for paid losses should not automatically be applied to another measure.
Worked Example: A Calendar-Year Signal in a Small Triangle
The following hypothetical cumulative paid-loss triangle is stated in thousands of dollars. Accident years are 2021, 2022, and 2023; development ages are 12, 24, and 36 months. A settlement initiative is known to have occurred during calendar year 2023. Calculate the observed 12-to-24-month volume-weighted link ratio, recalculate it without accident year 2022, and assess what the evidence does and does not establish.
| Accident year | 12 months | 24 months | 36 months | 12-to-24-month ratio |
|---|---|---|---|---|
| 2021 | 100 | 150 | 195 | 1.50 |
| 2022 | 120 | 216 | 270 | 1.80 |
| 2023 | 140 | 210 | — | 1.50 |
Method and assumptions. Use the ratio of summed cumulative amounts for the volume-weighted calculation. All amounts are deterministic and in thousands of dollars; no interest, mortality assumption, or probability distribution is used. Treat calendar year 2023 as the period associated with the 12-to-24-month increments for accident year 2022 and the 24-to-36-month increment for accident year 2021.
Step 1: Verify the age-to-age ratios. For accident year 2021, 150 ÷ 100 = 1.50. For accident year 2022, 216 ÷ 120 = 1.80. For accident year 2023, 210 ÷ 140 = 1.50.
Step 2: Calculate the volume-weighted ratio using all three observations. The summed 12-month amount is 100 + 120 + 140 = 360. The summed 24-month amount is 150 + 216 + 210 = 576. Therefore, the ratio is 576 ÷ 360 = 1.60.
Step 3: Recalculate without accident year 2022. The remaining 12-month total is 100 + 140 = 240, and the remaining 24-month total is 150 + 210 = 360. The ratio is 360 ÷ 240 = 1.50.
Step 4: Examine the calendar-year and operational evidence. The 1.80 ratio for accident year 2022’s 12-to-24-month interval is associated with calendar year 2023 and is higher than the 1.50 ratios for the other two accident years. The known settlement initiative is consistent with a temporary increase in paid emergence. For additional context, accident year 2021’s 24-to-36-month ratio is 195 ÷ 150 = 1.30, and that interval is also associated with calendar year 2023. These cells involve different development ages, so their ratios are not directly interchangeable, but they help identify the calendar-year diagonal to investigate.
Check and trap avoided. Including the 1.80 observation raises the weighted ratio from 1.50 to 1.60. This arithmetic shows its influence; it does not prove that the settlement initiative caused the difference. Automatically excluding the observation would also be unjustified without supporting evidence. The appropriate conclusion is to investigate the event and its subsequent emergence before deciding whether the 1.60 ratio represents repeatable development.
Common Mistakes and Practical Conclusions
- Confusing accident year with calendar year. An accident-year row contains multiple calendar-year emergence periods. Map each development interval to its calendar year before attributing a ratio change to a period event.
- Applying a price adjustment to the wrong amount. Cumulative losses combine increments from different periods. Adjust the affected increment or increments rather than mechanically multiplying the entire cumulative total by one inflation factor.
- Assuming an operational change changes ultimate losses. A settlement initiative can accelerate payments without changing the eventual total. Check the subsequent emergence when it is available.
- Treating a coincident event as proof. A calendar-year pattern or a documented process change supports a hypothesis, but other differences may explain the observations. State the evidence and uncertainty.
- Using a distorted observation without considering its effect. Calculate how much it changes the selected ratio, then assess whether the changed ratio is suitable for projecting other periods.
Check Your Understanding
Work each problem before checking the answer at the end of its statement.
- A hypothetical accident year is 2024, and age 1 represents the first 12 months. To which calendar year is its 12-to-24-month incremental emergence assigned? (Answer: 2025)
- Cumulative paid losses are $250,000 at 12 months and $300,000 at 24 months. Assume the $50,000 increment is fully affected by a 10% price increase relative to the base period. What is the 12-to-24-month ratio after restating the increment at the base price level? (Answer: 1.1818)
- A portfolio has cumulative paid losses of $500,000, $700,000, and $850,000 at 12, 24, and 36 months under its ordinary payment pattern. A settlement initiative shifts $60,000 from the 24-to-36-month period into the 12-to-24-month period, with no change in the 36-month total. What are the two affected link ratios after the shift? (Answer: 1.52 and 1.1184)
- Three hypothetical accident years have 12-month cumulative amounts of $80,000, $100,000, and $120,000, and corresponding 24-month amounts of $120,000, $170,000, and $180,000. The middle observation is suspected of reflecting a one-time operational effect. Calculate the all-year volume-weighted ratio and the ratio excluding the middle observation. (Answer: 1.5667 and 1.5000)
- A hypothetical prior cumulative amount is $600,000, and the next cumulative amount is $732,000. Assume the entire increment is affected by a 10% calendar-year price increase relative to the base level. Calculate the observed ratio and the ratio after deflating only the incremental amount. (Answer: 1.2200 and 1.2000)