Converting Cumulative And Incremental Payments

Why Convert Between Triangle Forms?

A paid-loss triangle can show either the amount paid during each development interval or the total paid up to each development age. These are two views of the same payments. Incremental values show when payments occur; cumulative values show the amount paid to date. Converting correctly matters because a calculation that expects one form will not generally work with the other.

The conversion is performed separately for each origin period. Within a row, the development ages are ordered from earliest to latest, and each observed cell refers to the same loss and payment scope throughout that row. The conversion does not change the underlying payments; it changes how those payments are grouped.

Definition: Let $I_{i,j}$ be the incremental paid loss for origin period $i$ during development interval $j$. Let $C_{i,j}$ be cumulative paid loss for origin period $i$, measured through development age $j$. If the first observed development interval is indexed by $j=1$, then $$ C_{i,j}=\sum_{k=1}^{j} I_{i,k} \qquad\text{and}\qquad I_{i,j}=C_{i,j}-C_{i,j-1}\quad (j\ge 2), $$ with $C_{i,0}=0$, so $I_{i,1}=C_{i,1}$.

The index $i$ identifies an origin period, while $j$ identifies development age or interval. For example, $I_{i,2}$ is the payment during the second development interval for origin period $i$; $C_{i,2}$ is the total paid in the first and second intervals combined. In an actual triangle, an origin period may have fewer observed development ages than an older origin period. Cells beyond the available valuation are unobserved, not automatically zero.

Converting Incremental Payments to Cumulative Payments

For each origin period, start with the first incremental amount and add each later interval's amount to the running total. The first cumulative value equals the first incremental value, because no earlier interval is included in the row. Every subsequent cumulative value is the preceding cumulative value plus the current increment.

C_{i,1}=I_{i,1},\qquad C_{i,j}=C_{i,j-1}+I_{i,j}\quad (j\ge 2)

This recurrence follows directly from the definition: the payments through interval $j$ consist of all payments through interval $j-1$, plus the payment in interval $j$. Apply it independently to every origin period. Do not carry a running total from one origin-period row into another.

Worked Example: Building Cumulative Rows From Three Origin Periods

A hypothetical insurer has incremental paid losses, in thousands of dollars, for three origin periods. Convert the observed amounts to cumulative form.

Origin period	12 months	24 months	36 months
2022	120	80	30
2023	150	60	—
2024	90	—	—

Method: running sums by row. For 2022, the 12-month cumulative amount is $120$. At 24 months, add the second interval: $120+80=200$. At 36 months, add the third: $200+30=230$.

For 2023, the 12-month cumulative amount is $150$, and the 24-month amount is $150+60=210$. There is no observed 36-month value for this origin period. For 2024, the only observed cumulative value is $90$ at 12 months; the later cells are unobserved.

Origin period	12 months	24 months	36 months
2022 cumulative	120	200	230
2023 cumulative	150	210	—
2024 cumulative	90	—	—

Check. The latest observed cumulative value in each row equals the sum of its observed increments: $120+80+30=230$, $150+60=210$, and $90=90$. The dashes remain unobserved; they have not been converted to zero.

Common mistake avoided: Adding the 2023 payments to the 2022 running total would mix two origin periods. Each cumulative sequence restarts at the beginning of its own row.

Converting Cumulative Payments to Incremental Payments

To recover payments by interval, subtract the cumulative value at the preceding development age from the value at the current age. The first incremental value is the first cumulative value because the row begins with no earlier payments included. For a given origin period, all differences must be taken within that row and in chronological order.

Conversion rule: If $C_{i,0}=0$, then for every observed interval $j$, $$ I_{i,j}=C_{i,j}-C_{i,j-1}. $$ For $j=1$, this gives $I_{i,1}=C_{i,1}-0=C_{i,1}$. For later intervals, the difference isolates the amount added between the two cumulative ages.

Worked Example: Recovering Interval Payments From Cumulative Values

A hypothetical specialty insurer records cumulative paid losses, in thousands of dollars, at 12-month development steps. Convert each observed row to incremental form.

Origin period	12 months	24 months	36 months	48 months
2021	10	40	55	70
2022	12	38	50	—
2023	8	25	—	—

Method: successive differences. For 2021, the first increment is $10$. The 24-month increment is $40-10=30$. The 36-month increment is $55-40=15$. The 48-month increment is $70-55=15$.

For 2022, the observed increments are $12$, $38-12=26$, and $50-38=12$. For 2023, they are $8$ and $25-8=17$. The resulting triangle is:

Origin period	12-month interval	12–24-month interval	24–36-month interval	36–48-month interval
2021	10	30	15	15
2022	12	26	12	—
2023	8	17	—	—

Check. Summing the 2021 increments gives $10+30+15+15=70$, its 48-month cumulative amount. The 2022 increments sum to $12+26+12=50$, and the 2023 increments sum to $8+17=25$. Each total matches the latest observed cumulative value in its row.

Common mistake avoided: Subtracting the current cumulative value from the prior one reverses the difference and changes the sign. The correct order is current minus prior: $C_{i,j}-C_{i,j-1}$.

Why Successive Differences Work

The conversion is not an approximation. It is an exact consequence of how cumulative amounts are defined, provided the values refer to consistent payments and valuation conventions. Write the cumulative values at two successive ages as sums of increments:

C_{i,j}=\sum_{k=1}^{j} I_{i,k}, \qquad C_{i,j-1}=\sum_{k=1}^{j-1} I_{i,k}.

Subtract the second expression from the first. Every increment from the first through the $(j-1)$th appears in both sums and cancels. Only the payment in interval $j$ remains:

C_{i,j}-C_{i,j-1} = \left(\sum_{k=1}^{j-1} I_{i,k}+I_{i,j}\right) -\sum_{k=1}^{j-1} I_{i,k} =I_{i,j}.

The same identity also provides a check in the other direction: summing the increments from $1$ through $j$ reconstructs $C_{i,j}$. Thus a converted row should satisfy both the successive-difference rule and the running-sum rule.

Interpreting Adjustments and Unobserved Cells

An incremental amount need not always be positive. A correction, recovery, or other accounting adjustment can reduce cumulative paid losses. In that situation, a negative increment may be valid, and the cumulative values may temporarily decrease. Do not replace a negative difference with zero merely because it seems unusual; first check the data definition and confirm that the adjacent cumulative values are comparable.

For example, suppose a row has cumulative amounts of $75$, $105$, $98$, and $120$, in thousands of dollars, at successive ages. The increments are $75$, $105-75=30$, $98-105=-7$, and $120-98=22$. Their sum is $75+30-7+22=120$, so the row reconciles. The negative value represents a net reduction during that interval under the stated cumulative data, not a failed subtraction.

A blank or unavailable triangle cell is different from a zero amount. A zero is an observed amount of no payment for an interval or through an age, depending on the data form. A blank indicates that the value is not observed in the triangle. The conversion rules apply only where the required values are observed. In particular, do not calculate an increment by subtracting a blank cell, and do not infer that an unobserved later cumulative amount equals the last observed amount.

Identify the data form.
Determine whether the cells are payments during separate intervals or totals paid through successive ages.

Work within one origin period.
Keep each row separate and order its observed development ages from earliest to latest.

Apply the matching operation.
Use running sums for incremental-to-cumulative conversion and current-minus-prior differences for cumulative-to-incremental conversion.

Reconcile the row.
Check that cumulative values are rebuilt by summing increments and that the final observed cumulative amount matches the sum of observed increments.

Common Exam Traps

Mixing rows. Each origin period has its own payment history. A running sum must restart for every row.
Reversing a difference. The increment at a later age is current cumulative minus prior cumulative, not the reverse.
Treating the first age like a later age. With $C_{i,0}=0$, the first increment equals the first cumulative value; there is no observed prior-age cell to subtract.
Confusing an interval amount with a total-to-date amount. An incremental cell covers one interval. A cumulative cell includes that interval and all earlier intervals in the row.
Filling blanks with zero. A missing or unobserved cell is not evidence of zero payment. Preserve the observed triangle boundary.
Assuming every increment must be nonnegative. A negative increment can arise from a valid adjustment. Check consistency and data meaning before changing it.

Key takeaway: For a fixed origin period, cumulative paid loss is the running sum of incremental paid loss, and incremental paid loss is the difference between successive cumulative values. A reliable conversion preserves the row total and does not turn unobserved cells into zeros.

Check Your Understanding

For each problem, convert only the observed cells and check the row totals where possible.

A hypothetical origin period has incremental payments of $42$, $18$, and $7$ thousand dollars at successive ages. Give its cumulative values. (Answer: $42,\ 60,\ 67$ thousand dollars.)
A cumulative row, in thousands of dollars, is $95,\ 140,\ 152$ at 12, 24, and 36 months. Find the three incremental amounts and verify their total. (Answer: $95,\ 45,\ 12$ thousand dollars; total $152$ thousand dollars.)
Two hypothetical origin periods have incremental rows $30, 25, 10$ and $45, 15$, respectively, in thousands of dollars. Convert each to cumulative form without combining the rows. (Answer: first row $30,\ 55,\ 65$; second row $45,\ 60$, all in thousands of dollars.)
A cumulative row is $80,\ 124,\ 119,\ 150$ thousand dollars at successive ages. Find the incremental values and state whether any increment is negative. (Answer: $80,\ 44,\ -5,\ 31$ thousand dollars; the third increment is negative.)
Three origin periods have observed incremental rows $20, 12, 8$; $35, 15$; and $40$, in thousands of dollars. Convert to cumulative form and calculate the sum of the latest observed cumulative amounts across the three rows. (Answer: cumulative rows $20, 32, 40$; $35, 50$; and $40$, all in thousands of dollars; latest-observed cumulative amounts sum to $40+50+40=130$ thousand dollars.)

What You'll Learn