CAS Exam 5 study materials
CAS Exam 5 study resources

Paid vs. Incurred Development After a Case-Reserving System Change

CAS Exam 5 · Put your knowledge to work in context.

All case studies

Paid vs. Incurred Development After a Case-Reserving System Change

A commercial liability carrier is setting reserves at December 31, 2025. Its new claims system went live January 1, 2025 and sets every newly reported claim at a default $1,000 case reserve until an adjuster reviews it; the prior system required an adjuster estimate within 24 hours. Cumulative paid losses ($000) at 12, 24, 36 and 48 months: AY2022 1,200/2,100/2,700/3,000; AY2023 1,350/2,450/3,100; AY2024 1,500/2,800; AY2025 1,700. Cumulative incurred: AY2022 1,500/2,300/2,850/3,050; AY2023 1,800/2,700/3,450; AY2024 1,900/2,600; AY2025 1,750. Volume-weighted development factors should be computed from all available years; selected tail factors are 1.10 on paid and 1.03 on incurred. There is no reinsurance. The CFO has asked for one net IBNR number and wants to know why the paid and incurred methods disagree by roughly $2.4M.

1. Compute the volume-weighted development factors, the ultimates and the total IBNR under both the paid and the incurred methods.

Read the worked answer

Paid factors

  • 12-24: (2,100 + 2,450 + 2,800) / (1,200 + 1,350 + 1,500) = 7,350 / 4,050 = 1.8148
  • 24-36: (2,700 + 3,100) / (2,100 + 2,450) = 5,800 / 4,550 = 1.2747
  • 36-48: 3,000 / 2,700 = 1.1111; tail 1.10

Paid ultimates ($000):

  • AY2022: 3,000 x 1.10 = 3,300.0
  • AY2023: 3,100 x 1.1111 x 1.10 = 3,788.9
  • AY2024: 2,800 x 1.2747 x 1.1111 x 1.10 = 4,361.9
  • AY2025: 1,700 x 1.8148 x 1.2747 x 1.1111 x 1.10 = 4,806.6

Paid IBNR: 300.0 + 688.9 + 1,561.9 + 3,106.6 = 5,657.4

Incurred factors

  • 12-24: (2,300 + 2,700 + 2,600) / (1,500 + 1,800 + 1,900) = 7,600 / 5,200 = 1.4615
  • 24-36: (2,850 + 3,450) / (2,300 + 2,700) = 6,300 / 5,000 = 1.2600
  • 36-48: 3,050 / 2,850 = 1.0702; tail 1.03

Incurred ultimates ($000):

  • AY2022: 3,050 x 1.03 = 3,141.5
  • AY2023: 3,450 x 1.0702 x 1.03 = 3,802.9
  • AY2024: 2,600 x 1.2600 x 1.0702 x 1.03 = 3,611.1
  • AY2025: 1,750 x 1.4615 x 1.2600 x 1.0702 x 1.03 = 3,552.3

Incurred IBNR: 91.5 + 352.9 + 1,011.1 + 1,802.3 = 3,257.8

The gap is 5,657.4 - 3,257.8 = 2,399.6, matching the CFO's $2.4M.

2. Diagnose the discrepancy and select a net IBNR with supporting reasoning.

Read the worked answer

Diagnosis. Compute case outstanding at the first evaluation (incurred minus paid at 12 months): AY2022 300; AY2023 450; AY2024 400; AY2025 50. Twelve-month paid losses rise steadily every year (1,200 to 1,350 to 1,500 to 1,700, about 12% per year), consistent with a growing book, while twelve-month incurred fall (1,900 to 1,750). The only mechanism that produces falling incurred with rising paid is a collapse in initial case reserves, and that is exactly what a $1,000 default reserve would do. The 12-24 incurred factor of 1.4615 is therefore a blend of genuine development and a one-time case deficiency, and applying it to AY2025 produces an ultimate far too low.

Which method to trust. Paid development is unaffected by case-reserving philosophy, so it carries the signal here, provided the system change has not also altered payment speed. So far there is no evidence of that: 12-month paid growth is in line with prior years and the 12-24 paid factor has been stable.

Selection. Start from paid ultimates.

  • AY2022 is essentially mature; paid gives 3,300 and incurred gives 3,142. Select about 3,250, IBNR about 250.
  • AY2023 and AY2024: take the paid ultimates, but watch that AY2024's 4,362 sits above AY2023's 3,789 on a growing book, which is plausible but worth checking against exposure growth.
  • AY2025: the paid ultimate of 4,807 is the least reliable number in the table, because it multiplies four factors. Cross-check with a Bornhuetter-Ferguson indication on an expected loss ratio.

Reasonable range for total net IBNR: $5.0M to $6.0M, point near $5.5M. Explicitly disclose that the incurred-based $3.26M is rejected as an artifact.

3. What monitoring would you put in place, and what else in the reserve analysis does this system change affect?

Read the worked answer

Monitoring.

  • Track 12-month case outstanding (both count and average) for AY2025 onward. When adjusters catch up, incurred development will first look unusually low and then show a catch-up; the paid-to-incurred ratio and the closure rate will confirm it.
  • Track the age at first adjuster estimate and the share of claims still at the $1,000 default at 30, 60, and 90 days. If most claims are reviewed within a month, the distortion is confined to the earliest evaluation and the 12-24 incurred factor is the only contaminated one.
  • Do not recalibrate the incurred factors from AY2025; either exclude it from the factor selection or apply an explicit adjustment until a clean year exists.

Other affected items.

  • ULAE. A default reserve with no adjuster entry reduces claim handling work in the short run. If ULAE is loaded as a percentage of case reserves or as a function of claim counts, the ratio may shift; check that the ULAE provision is driven by counts and payments, not by case outstanding.
  • Case reserve adequacy for prior years. If the default is also being applied on reopen or on transferred claims, older years could be understated too; test AY2023 and AY2024 for case deficiency using paid development.
  • Actuarial central estimate and disclosures. The change in case philosophy is a change in estimation basis, not in underlying liability. Say so explicitly, and flag that the incurred-based reserve indication this quarter is not comparable to last quarter's.