Why Sparse and Zero Exposure Require Care
A period with little exposure can have a very high or low observed pure premium simply because a small number of claims is divided by a small denominator. A period with zero exposure presents a different problem: its period pure premium is undefined. Neither issue is solved by taking an unweighted average of the period ratios. For a portfolio-level indication, the basic safeguard is to check that the losses and exposures are on a consistent basis, then calculate the ratio of pooled losses to pooled exposure.
This tutorial focuses on that calculation and on the data checks needed before using it. The result is useful only when losses and exposures refer to the same coverage, population, and experience period, and when any needed adjustments, such as loss trend, have been applied consistently.
Why Pooling Works
For every period with positive exposure, its losses can be written as \(L_i=E_i p_i\). Substituting this expression into the pooled calculation gives
provided that periods with zero exposure have no losses assigned to them on the selected, matched basis. The period pure premiums are therefore averaged using exposure shares as weights, not equal weights. A period with a small exposure receives a small weight in the pooled result, even if its own ratio is unusually high or low.
This identity also explains the zero-exposure issue. A period with \(E_i=0\) cannot supply a period ratio \(p_i\), and so it cannot be inserted as an ordinary weighted-average term. When the period is truly outside the experience population and has no corresponding losses, it contributes zero to both sums and may be omitted. If losses are recorded against zero exposure, the data must be investigated before choosing how the losses enter the calculation.
Worked Example: Pooling a Sparse Period
Worked Example: Pooling a Sparse Period
A hypothetical commercial equipment portfolio has the following selected losses and earned exposures over four periods. Assume the losses and exposures are on the same coverage and accident-period basis. Calculate the pooled pure premium, and explain how the period with only 10 exposure units affects it.
| Period | Exposure \(E_i\) | Losses \(L_i\) |
|---|---|---|
| 1 | 1,000 | $80,000 |
| 2 | 10 | $4,000 |
| 3 | 0 | $0 |
| 4 | 500 | $45,000 |
Step 1: Calculate period ratios only where exposure is positive. Period 1 has \(80{,}000/1{,}000=\$80.00\) per exposure unit. Period 2 has \(4{,}000/10=\$400.00\). Period 4 has \(45{,}000/500=\$90.00\). Period 3 has no defined period ratio because its denominator is zero.
Step 2: Pool losses and exposure. Total losses are \(80{,}000+4{,}000+0+45{,}000=\$129{,}000\). Total exposure is \(1{,}000+10+0+500=1{,}510\) units.
Step 3: Divide pooled losses by pooled exposure.
Rounded to the nearest cent, the pooled pure premium is $85.43 per exposure unit. Period 2’s ratio of $400.00 is based on only 10 units, so it can be unstable. Its exposure weight is \(10/1{,}510\), or approximately 0.0066, meaning it has only a small influence on the pooled result. The zero-exposure period contributes neither losses nor exposure under the stated assumptions.
Check. The exposure-weighted average gives \((1{,}000\times80+10\times400+500\times90)/1{,}510=(80{,}000+4{,}000+45{,}000)/1{,}510=\$85.43\), agreeing with the pooled calculation. The method is pooling, and it avoids the common mistake of averaging period ratios equally.
When Zero Exposure Has Recorded Losses
A positive loss amount paired with zero exposure is a warning that the selected numerator and denominator may not be aligned. Possible explanations include missing exposure records, a claim assigned to a different period from the exposure that generated it, or a difference in the population or coverage represented by the two fields. The correct treatment depends on the source of the mismatch; the ratio itself cannot diagnose it.
If investigation confirms that exposure is missing, repair the exposure data and recalculate. If a loss belongs to another period under the selected basis, align the loss with that period. If the loss is genuinely relevant to the overall portfolio but cannot be matched to the chosen exposure basis, its treatment requires a defensible selection decision; it must not be made to look harmless by dividing by zero or dropped without explanation.
Confirm that the losses and exposure cover the same population, coverage, and experience periods.
Determine whether the exposure is missing, the loss is assigned to a different period, or the selected data require another documented treatment.
Use aligned data where possible. Do not invent a positive denominator or suppress a loss merely to obtain a finite ratio.
Once the data are suitable, divide the selected total losses by the selected total exposure.
Worked Example: Resolving a Loss With Zero Recorded Exposure
A hypothetical specialty-insurance data extract shows losses of $64,000, $6,000, and $108,000 for three periods, and exposures of 800, 0, and 1,200 units. A review confirms that the $6,000 loss belongs to the second period and that 50 units of earned exposure for that period were omitted from the extract. Calculate the corrected pooled pure premium.
Step 1: Do not use the uncorrected zero-exposure ratio. The second period’s displayed ratio would be \(6{,}000/0\), which is undefined. This is not evidence that the period’s pure premium is infinite; it is evidence that the recorded denominator is incomplete for this loss.
Step 2: Correct the exposure. The second-period exposure is 50 units. Total corrected exposure is \(800+50+1{,}200=2{,}050\) units.
Step 3: Pool the losses. Total losses are \(64{,}000+6{,}000+108{,}000=\$178{,}000\).
Step 4: Calculate and check.
Rounded to the nearest cent, the corrected pooled pure premium is $86.83 per exposure unit. As a check, the corrected period ratios are $80.00, $120.00, and $90.00. Their exposure-weighted average is \((800\times80+50\times120+1{,}200\times90)/2{,}050=(64{,}000+6{,}000+108{,}000)/2{,}050=\$86.83\). The method is data reconciliation followed by pooling; it avoids both discarding the $6,000 loss and treating zero as a usable denominator.
Sparse Periods and Adjusted Losses
Small, positive exposure is not the same as zero exposure. A period ratio can be calculated when \(E_i>0\), but it may be highly variable when the denominator is small. For an overall indication, pooling usually keeps that period’s influence proportional to its exposure. If period losses need adjustment to a common level, apply the appropriate adjustment to each selected loss amount before pooling; do not average adjusted period ratios equally.
Worked Example: Pooling Losses After Trend Adjustment
A hypothetical repair-cost portfolio has the following exposures, losses, and specified loss trend factors to a common level. Assume the trend factors are appropriate for these periods and that the exposure basis is unchanged. Calculate the trend-adjusted pooled pure premium.
| Period | Exposure | Selected losses | Trend factor |
|---|---|---|---|
| 1 | 400 | $32,000 | 1.10 |
| 2 | 5 | $1,000 | 1.05 |
| 3 | 0 | $0 | 1.00 |
| 4 | 600 | $60,000 | 1.00 |
Step 1: Trend each period’s losses. Period 1 becomes \(32{,}000\times1.10=\$35{,}200\). Period 2 becomes \(1{,}000\times1.05=\$1{,}050\). Period 3 remains $0, and Period 4 remains $60,000.
Step 2: Add the adjusted losses and exposures. Adjusted losses total \(35{,}200+1{,}050+0+60{,}000=\$96{,}250\). Exposure totals \(400+5+0+600=1{,}005\) units.
Step 3: Calculate the pooled result.
Rounded to the nearest cent, the trend-adjusted pooled pure premium is $95.77 per exposure unit. The unrounded trend-adjusted period ratios for periods 1, 2, and 4 are $88.00, $210.00, and $100.00. Averaging them equally would give \((88+210+100)/3=\$132.67\), which is not the pooled indication: it gives the five-unit period the same weight as each much larger period. Pooling gives the sparse period only \(5/1{,}005\) of the exposure weight.
Check. Weighting the adjusted ratios by exposures gives \((400\times88+5\times210+600\times100)/1{,}005=(35{,}200+1{,}050+60{,}000)/1{,}005=\$95.77\). The method is trend adjustment followed by exposure-weighted pooling; it avoids using an unweighted average of period indications.
Common Mistakes and Practical Checks
The central distinction is between a period statistic and an aggregate statistic. A very high pure premium for a sparse period may be a valid calculation for that period, but it should not automatically receive the same influence as a period with much more exposure in a pooled indication. Conversely, a low or zero total exposure does not justify ignoring losses without checking what those losses represent.
- Dividing by zero. A period pure premium is undefined when exposure is zero. Do not label it zero, infinite, or any other calculated value.
- Averaging period ratios equally. This gives a sparse period the same weight as a much larger one. Use total selected losses divided by total selected exposure for the pooled result.
- Adding an arbitrary denominator. Replacing zero exposure with a small positive number creates a result determined by an invented value, not the observed experience.
- Discarding losses assigned to zero exposure. Investigate the mismatch and document the treatment. A zero denominator alone does not establish that the loss is irrelevant.
- Mixing incompatible bases. Pooling does not fix differences in coverage, population, or period assignment. Align the data before interpreting the result.
Check Your Understanding
For each problem, assume losses and exposure are on a consistent basis unless the question identifies a data issue. Report pure premiums to the nearest cent.
- A hypothetical policy group has losses of $15,000, $0, and $60,000 with exposures of 250, 0, and 750 units, respectively. Calculate the pooled pure premium and state whether the zero-exposure period has a period pure premium. (Answer: $75.00 per exposure unit; no, it is undefined.)
- A set of periods has exposures of 400, 8, and 592 units and losses of $28,000, $1,200, and $41,440. Calculate the pooled pure premium, then explain why the second period should not receive equal weight in an average of period ratios. (Answer: $70.64 per exposure unit; its exposure is only 8 of 1,000 units.)
- A data review confirms that a $4,800 loss belongs to a period whose recorded exposure was zero because 40 units were missing. The other period has 960 units of exposure and $72,000 of losses. Calculate the corrected pooled pure premium. (Answer: $76.80 per exposure unit.)
- A portfolio has exposures of 300, 0, and 700 units, losses of $27,000, $0, and $77,000, and trend factors of 1.20, 1.00, and 1.00. Calculate the trend-adjusted pooled pure premium. (Answer: $109.40 per exposure unit.)
- A portfolio has exposures of 900, 3, and 97 units, with losses of $72,000, $900, and $7,760. Calculate the pooled pure premium and compare it with the unweighted average of the three period ratios. (Answer: pooled $80.66 per exposure unit; unweighted average $153.33 per exposure unit.)