Core Path · L4.6.07

Counterparty and underwriting modules

12 minCounterparty defaultPremium and reserve riskNat catLife underwritingHealth

By the end you will be able to

  1. 1Compute the counterparty default module: LGDs, PDs by credit quality, the type 1 variance and its band, the type 2 factors and their combination.
  2. 2Build non-life underwriting risk: premium and reserve from volumes and σ, lapse, and catastrophe with the nat cat scenarios in outline.
  3. 3Name the seven life sub-modules and their shocks, and the health sub-modules, and combine them with their matrices.
  4. 4Compare RiskAtoms' stated and modelled figures with the rules: counterparty €150m → €267.3m, non-life €2,801.2m → €3,002.0m.

Why it matters

Level 3 wrote RiskAtoms' counterparty charge as a round €150m and took its cat charge from its own event model. Both are reasonable guesses, and neither is what the standard formula asks for. The formula wants the reinsurers by name and rating, and the sums insured by region. Supply them and counterparty becomes €267.3m, cat €1,084.3m: 54% more than the insurer's own 1-in-200.

Core explanation

Who owes you, and what you insure

Illustrative dataRiskAtoms Insurance SE, under the rules

The same made-up composite insurer as M3.8: Insurance risk, in € millions at 31 December 2025: own funds of €9,300m against a Solvency Capital Requirement of €5,000m, a ratio of 186%. Those are its booked figures and they stay so. Level 3 built the standard formula in outline; this module reads it against Delegated Regulation (EU) 2015/35 as in force at that date, article by article. Where Level 3 stated a figure or used its own model, the “rules in outline” view computes the standard formula instead, on assumptions labelled illustrative, and shows the difference. Every policy, bond, treaty and counterparty is invented.

Counterparty exposures and nat cat sums insured are assumptions, stated below illustrative. Premium and reserve, life and health are Level 3's calculations, checked against their articles.

Counterparty default: two kinds of exposure

Art 189 splits exposures in two. Type 1 are few and rated: here six reinsurers holding €3,000m of recoverables and four banks holding €2,000m of cash. Type 2 are many and small: €800m of premiums receivable. Each type 1 counterparty gets a loss given default and a probability of default from its credit quality step (Art 199): 0.002% for AAA, 0.01% AA, 0.05% A, 0.24% BBB, 1.2% BB, 4.2% below. A reinsurer's LGD includes half of its risk-mitigating effect, because a default also takes the cover away (L3.8.10: Reinsurance in depth).

Type 1 exposure (assumed)AmountRM_reLGDPD
4 reinsurers rated AA, each600300375.00.01%
2 reinsurers rated A, each300150187.50.05%
4 banks rated A, each5000500.00.05%
Total LGD3,875

The charge is not a sum of PD × LGD. Art 201 builds a variance from two parts: V_inter, which treats counterparties in different PD classes as correlated through the economy, and V_intra, which adds the lumpiness of a few large names within a class. Here V_inter is 1,692.0 and V_intra 354.7, so σ = €45.2m, 1.17% of total LGD. Below 7% the charge is 3σ (Art 200): €135.7m. Type 2 is €150m, and the two combine as √(T1² + 1.5·T1·T2 + T2²) = €267.3m rules.

The variance weighs both size and quality. Put all €3,000m with one A-rated reinsurer and type 1 becomes €198.0m, the module €325.9m; with one AA reinsurer it is €132.5m, because the better rating offsets the concentration. All the cash at one bank gives €162.8m; the six reinsurers one step lower each, €190.2m. Reinsurance always trades underwriting risk for counterparty risk (L2.6.07: Insurance and reinsurance); the standard formula prices the trade.

Non-life: premium and reserve

Premium risk is that next year's claims on next year's premiums are worse than expected; reserve risk that the claims already incurred cost more than reserved (L3.8.04: Premium risk and the underwriting cycle). Art 115 charges both together as 3 × σ_nl × V_nl, where each segment's volume is premium plus reserve, and its standard deviation blends Annex II's σ_prem and σ_res with a correlation of 0.5. σ_prem is multiplied by 80% for non-proportional reinsurance in segments 1, 4 and 5: motor vehicle liability, fire and other damage to property, and general liability (Art 117(3)).

SegmentV_premV_resσ_prem × NPσ_resσ_s
1. Motor liability2,338.52,712.18.0%9%7.41%
2. Motor own damage1,620.0238.78.0%8%7.54%
4. Property (home and commercial)2,712.7990.26.4%10%6.46%
5. General liability1,528.82,606.011.2%11%9.69%
σ_nl, V_nl14,746.95.74%

Combined across segments with Annex IV's correlations, σ_nl × V_nl = 847.1 against 1,154.4 added up; the charge is 3 × 5.74% × 14,746.9 = €2,541.3m. Level 3 simplifies in two directions: it leaves out the expected premiums of existing and future contracts in V_prem (Art 116(3)), which would raise the volume, and the geographical diversification of Annex III, which would lower it. A firm may replace some σ with undertaking-specific parameters from its own data, with supervisory approval (Directive Art 104(7)).

Non-life: lapse and catastrophe

Non-life lapse (Art 118) is the loss if 40% of policies ended where that raises technical provisions; it bites only when the premium provision counts future profits, which RiskAtoms' does not, so it is nil. Catastrophe (Art 119) combines four parts: natural catastrophe with non-proportional property reinsurance, man-made (motor, marine, aviation, fire, liability, credit), and other, as √((nat + NP)² + man-made² + other²). The nat cat sub-module, in outline on RiskAtoms' zones illustrative:

Peril (Arts 121–124)RegionsCharge, netRegions independent
WindstormDE, BE, NL, LU, FR488.5300.5
FloodDE, BE817.3751.2
HailDE, IT, FR72.772.7
EarthquakeIT513.5513.5
Nat cat (perils independent)1,084.3lower 1,035.3

Fire, on the largest concentration of buildings within a 200-metre radius (here one building, €160m gross; Art 132), costs €2m net of the surplus treaty and per-risk cover. So cat is €1,084.3m against the event model's €703.2m, and the non-life module, premium and reserve and cat combined at 0.25, becomes €3,002.0m against €2,801.2m. The standard formula's scenarios are calibrated to a market-average portfolio, not to RiskAtoms' zones; a firm whose own model says less can ask to use a partial internal model (L4.6.09: Internal models). Catastrophe modelling: L3.8.05: Catastrophe risk.

Life: seven shocks

Each life sub-module is an instant, permanent change to one assumption, and the charge is the fall in own funds it causes, applied only to the policies where it hurts:

Sub-moduleArticleShockRiskAtoms
MortalityArt 137+15% mortality rates206.5
LongevityArt 138−20% mortality rates1,887.2
DisabilityArt 139(no life disability cover)0.0
LapseArt 142max(up, down, mass): mass431.2
ExpenseArt 140+10% expenses, +1pp inflation111.4
RevisionArt 141(no non-life annuities)0.0
CatArt 143+0.15pp mortality for 12 months86.5
Life (7×7 matrix)Art 136sum 2,722.82,054.0

Longevity dominates: the annuity book loses €1,887.2m if mortality falls 20% (L3.8.07: Longevity risk), while the term book loses €206.5m if it rises 15% (L3.8.06: Mortality risk). The two are correlated −0.25 in the matrix: a firm with both is partly hedged. Lapse is the largest of up 50%, down 50% (at most 20 points) and mass lapse: here mass, at €431.2m (L3.8.08: Lapse and policyholder behaviour). The seven add to €2,722.8m; the matrix gives €2,054.0m.

Delegated Regulation Art 136(3), in Level 3's order
MortalityLongevityDisabilityExpenseRevisionLapseCat
Mortality1.00−0.250.250.250.000.000.25
Longevity−0.251.000.000.250.250.250.00
Disability0.250.001.000.500.000.000.25
Expense0.250.250.501.000.500.500.25
Revision0.000.250.000.501.000.000.00
Lapse0.000.250.000.500.001.000.25
Cat0.250.000.250.250.000.251.00

Health

Health splits by technique: similar-to-life (SLT) health mirrors the life sub-modules, not-similar-to-life (NSLT) health mirrors non-life premium, reserve and lapse, and health catastrophe adds mass accident, accident concentration and pandemic (Arts 144–163). SLT and NSLT correlate 0.5; catastrophe 0.25 with both. RiskAtoms' income protection is SLT: disability-morbidity €191.3m (35% more incidence in year one, 25% after, and 20% less recovery where recovery rates are below 50%; Art 156) and expense €13.0m, combined at 0.5 to €198.1m (L3.8.09: Expense, health and morbidity risk). Level 3 leaves out SLT lapse and revision and health cat.

The formal version

Delegated Regulation (EU) 2015/35 Arts 114–117, 119, 136–143, 189–202:

LGDre=max⁡(50% (Rec+50% RMre)−F Coll, 0),LGDbank=depositVinter=∑j,kpj(1−pj) pk(1−pk)1.25(pj+pk)−pjpk TLGDj TLGDk,Vintra=∑j1.5 pj(1−pj)2.5−pj∑i∈jLGDi2SCR1={3σσ≤7% ΣLGD5σσ≤20% ΣLGDΣLGDotherwise,σ=Vinter+VintraSCR2=90% LGDreceivables>3m+∑15% LGDi,SCRdef=SCR12+1.5 SCR1SCR2+SCR22σs=σp,s2Vp,s2+σp,sσr,sVp,sVr,s+σr,s2Vr,s2Vp,s+Vr,s,NLpr=3 σnl VnlNLCAT=(NLnat+NLnp)2+NLmm2+NLother2,SCRlife=∑i,jCorrij LiLj\begin{aligned} \text{LGD}_{re} &= \max\big(50\%\,(\text{Rec} + 50\%\,\text{RM}_{re}) - F\,\text{Coll},\ 0\big), \qquad \text{LGD}_{bank} = \text{deposit} \\ V_{\text{inter}} &= \sum_{j,k} \frac{p_j(1-p_j)\,p_k(1-p_k)}{1.25(p_j + p_k) - p_j p_k}\, TLGD_j\, TLGD_k, \qquad V_{\text{intra}} = \sum_j \frac{1.5\,p_j(1-p_j)}{2.5 - p_j} \sum_{i \in j} \text{LGD}_i^2 \\ \text{SCR}_{1} &= \begin{cases} 3\sigma & \sigma \le 7\%\,\Sigma\text{LGD} \\ 5\sigma & \sigma \le 20\%\,\Sigma\text{LGD} \\ \Sigma\text{LGD} & \text{otherwise} \end{cases}, \quad \sigma = \sqrt{V_{\text{inter}} + V_{\text{intra}}} \\ \text{SCR}_{2} &= 90\%\,\text{LGD}_{\text{receivables} > 3m} + \textstyle\sum 15\%\,\text{LGD}_i, \qquad \text{SCR}_{\text{def}} = \sqrt{\text{SCR}_1^2 + 1.5\,\text{SCR}_1\text{SCR}_2 + \text{SCR}_2^2} \\ \sigma_s &= \frac{\sqrt{\sigma_{p,s}^2 V_{p,s}^2 + \sigma_{p,s}\sigma_{r,s} V_{p,s} V_{r,s} + \sigma_{r,s}^2 V_{r,s}^2}}{V_{p,s} + V_{r,s}}, \qquad \text{NL}_{pr} = 3\,\sigma_{nl}\,V_{nl} \\ \text{NL}_{\text{CAT}} &= \sqrt{(\text{NL}_{\text{nat}} + \text{NL}_{\text{np}})^2 + \text{NL}_{\text{mm}}^2 + \text{NL}_{\text{other}}^2}, \qquad \text{SCR}_{\text{life}} = \sqrt{\textstyle\sum_{i,j} \text{Corr}_{ij}\, L_i L_j} \end{aligned}
  • TLGDjTLGD_j is the total LGD of counterparties with PD pjp_j. The 1.25 and 1.5 are calibration constants; the variance is a sum over pairs, like s⊤Css^{\top}Cs (M-C06: The normal distribution).
  • Every shock is a percentage of a rate, a volume or a sum insured (M-A01: Percentages); applied per policy, only where it raises technical provisions.

Try it

Concentrate the panel, downgrade it, take away the risk-mitigating effect, and let receivables go overdue. Watch σ move against ΣLGD: if σ passes 7% of ΣLGD the charge jumps from 3σ to 5σ. A four-step downgrade gets there whether one reinsurer holds all the reinsurance or the panel of six shares it, the panel only just (7.05% against 9.84%).

Try itConcentrate, downgrade, and watch the charge

Reinsurance panel

Bank deposits

Counterparty module€267.3m
σ as share of ΣLGD1.17%
V_inter1,692.0
V_intra354.7
σ = √V€45.2m
ΣLGD€3,875m
Reinsurers ratedAA (PD 0.01%)
Band (Art 200)3σ

Illustrative, € millions. Reinsurer LGD = 50% × (recoverables + 50% × RM_re), RM_re shared in proportion to recoverables, no collateral; bank LGD = the deposit. SCR = 3σ while σ ≤ 7% of ΣLGD, 5σ up to 20%, else ΣLGD. Type 2 = 90% of receivables from intermediaries over three months due + 15% of the rest. Module = √(T1² + 1.5·T1·T2 + T2²).

Worked example

Nat cat in outline

RiskAtoms' sums insured by country from Level 3's zones, every risk weight 1, regions run through the cat XL (€500m xs €500m and €1,000m xs €1,000m, one reinstatement each) illustrative.

From sums insured to the cat charge

Step 1 of 5
  1. 1

    Windstorm, Germany

    LDE=Q×SI=0.07%×180,000=126A: 80% L+40% L=100.8+50.4=151.2 (both below the 500 retention)B: 100% L+20% L=151.2\begin{aligned} L_{DE} &= Q \times SI = 0.07\% \times 180{,}000 = 126 \\ \text{A: } & 80\%\,L + 40\%\,L = 100.8 + 50.4 = 151.2 \ (\text{both below the } 500 \text{ retention}) \\ \text{B: } & 100\%\,L + 20\%\,L = 151.2 \end{aligned}

With counterparty at €267.3m as well, the SCR on the rules in outline is €5,155.9m and the ratio 180.4% (L4.6.05: The standard formula: structure).

Regulatory lens

Delegated Regulation (EU) 2015/35 Arts 114–163 (underwriting) and 189–202 (counterparty), with Annexes II and V–VIII; Directive 2009/138/EC Arts 104–105 as at 4 Nov 2026.

In the EU

Delegated Regulation (EU) 2026/269 amends the counterparty articles from 30 January 2027, among them Arts 189, 192, 199, 201 and 202. It also amends underwriting articles: Art 117 (reinsurance adjustments), nat cat and man-made cat (Arts 120–132, with new regional factors), Art 142(6), and health Arts 148 and 157. The life shocks of Arts 137–141 and 143 are not among the amendments we read. Read the shocks above as the rules in force at 31 December 2025.

In the UK

The PRA Rulebook keeps the same modules in its Standard Formula Part; firms use undertaking-specific parameters and internal models with PRA approval (L4.6.13: Solvency UK and global standards).

Results are disclosed by module in the SFCR, with the detail of each sub-module in the quantitative reporting templates sent to the supervisor (L4.6.11: Governance, ORSA and reporting).

Spark Atoms

Spark AtomYour model or ours
The standard formula's nat cat factors are set for an average European portfolio. For RiskAtoms they give €1,084.3m, against €703.2m from its own events: about €106m more SCR after diversification and the tax credit. That gap is the usual reason non-life insurers seek approval for a partial internal model for cat.
Spark AtomWhy 1.5, not 2
Type 1 and type 2 are combined with 1.5·T1·T2, a correlation of 0.75: defaults of reinsurers and of brokers are expected to cluster in the same bad year, but not perfectly.

Key takeaways

  • Counterparty default: type 1 (reinsurers, banks, derivatives) from a variance over PD classes, 3σ, 5σ or ΣLGD; type 2 at 15% and 90%; combined with 1.5·T1·T2. RiskAtoms: €135.7m + €150m → €267.3m.
  • Reinsurer LGD = 50% × (recoverables + 50% × RM_re): a default loses money owed and protection. Concentration and downgrades raise the charge.
  • Non-life premium and reserve = 3σ_nl V_nl from Annex II's σ by segment (€2,541.3m); lapse nil here; cat = √((nat + NP)² + man-made² + other²).
  • Nat cat in outline: Q × sums insured by region, two-event scenarios net of reinsurance, regions and perils aggregated: €1,084.3m, against €703.2m from Level 3's own model.
  • Life: mortality +15%, longevity −20%, disability, lapse ±50% or mass 40%, expense +10% and +1 point, revision 3%, cat +0.15 points; 7×7 matrix. Health: SLT, NSLT and cat, with 0.5 and 0.25.

Reaction

Check what stuck

Question 1 · objective 1

A reinsurer owes €600m of recoverables, and its share of the risk-mitigating effect is €300m. There is no collateral. What is the LGD, in €m?

€m

Question 2 · objective 1

Receivables are €800m, of which €40m are due from intermediaries for more than three months. What is the type 2 charge, in €m?

€m

Question 3 · objective 2

Windstorm in Germany: the regional factor Q is 0.07% and the weighted sum insured €180,000m. What is the specified loss, in €m?

€m

Question 4 · objective 1

Type 1 is €135.7m and type 2 €150m. What is the counterparty module, in €m?

€m

Question 5 · objective 3

Which lapse scenario sets RiskAtoms' lapse charge, and why?

Glossary

Type 1 exposures
Type 1 exposures: few, usually rated counterparties whose default is diversified only a little: reinsurance, derivatives, cash at bank, letters of credit (Art 189(2)). Charged through a variance across PD classes (Arts 200–201).
Type 2 exposures
Type 2 exposures: many small, usually unrated ones: receivables from intermediaries and policyholders, mortgage loans and other credit exposures outside the spread sub-module (Art 189(3)). Charged with flat factors, 15% and 90% (Art 202).
Loss given default (LGD)
Loss given default (LGD): for reinsurance, 50% of the recoverables plus half the risk-mitigating effect, less collateral (Art 192(2)); for cash at bank, the deposit.
Risk-mitigating effect (RM_re)
Risk-mitigating effect (RM_re): how much lower the underwriting (or market) risk charge is because of the reinsurance, the hypothetical charge without it less the actual (Art 196). A reinsurer's default loses that protection as well as the money owed.
Volume measure
Volume measures: V_prem is the larger of next year's and last year's net earned premium, plus expected premiums on existing and future contracts (Art 116(3)); V_res is the net best estimate of claims outstanding.
Standard deviation by segment
Standard deviation σ: Annex II sets σ_prem and σ_res for each of 12 segments. They are combined within a segment, then across segments with Annex IV's correlations, into σ_nl.
Natural catastrophe sub-module
Natural catastrophe sub-module: windstorm, earthquake, flood, hail and subsidence, as specified losses Q × weighted sums insured in each region, run through two-event scenarios net of reinsurance (Arts 120–125).
Mass lapse
Mass lapse: the instant discontinuance of 40% of policies (70% for some group pension business) where that would raise technical provisions (Art 142(6)).

Deep Dives & Origins

Deep Dives

Origins

  • Commission Delegated Regulation (EU) 2015/35, Arts 119, 136–143, 144, 189, 192, 196, 199–202: EIOPA Solvency II Single Rulebook; legislation.gov.uk text.
  • Commission Delegated Regulation (EU) 2026/269 of 29 October 2025, Art 1 (LawPlayer copy of the Official Journal text).

Watch this lesson

Video version

Prefer to watch? The same lesson as a 12-minute video will sit here, made from this page's script.