Reinsurance Choices for Rating Resilience

Reinsurance as a Rating-Resilience Decision

Reinsurance can reduce the amount of loss an insurer retains, but the protection is not free and it does not remove every risk. A rating-relevant assessment therefore asks more than whether a treaty lowers modeled net losses. Management must also consider its cost, the resulting earnings pattern, how much capital relief is recognized, and whether the reinsurer will pay when a recovery is due.

The calculations below use simplified hypothetical models. They illustrate decision analysis, not the capital formula of a particular rating agency. Actual assessments may depend on the agency’s methodology, the insurer’s financial position, the treaty wording, and evidence about the reinsurer’s credit quality and collectability.

Definition: A reinsurance arrangement transfers some specified insurance losses from the insurer to a reinsurer in exchange for a premium or other consideration. The insurer’s gross loss is the covered loss before reinsurance; its retained loss is the gross loss less any recovery actually recognized. Counterparty exposure is the risk that the reinsurer does not pay some or all of an amount due.

For rating resilience, consider three distinct effects. First, reinsurance can change expected earnings: the cost of coverage may outweigh the expected recoveries. Second, it can reduce the variability of earnings by limiting the insurer’s retained losses. Third, it can affect the assessed capital position, but modeled relief may be reduced by counterparty risk, contract limitations, or the agency’s treatment of recoverables. A conclusion about one effect does not settle the others.

Separating Average Cost from Earnings Volatility

Suppose annual gross losses are represented by a random variable $L$. For a quota-share treaty, the insurer retains a fixed proportion $1-a$ of each covered loss, where $a$ is the ceded proportion. If the reinsurance premium is a deterministic amount $c$, the insurer’s loss cost after reinsurance is

L_{\text{net}}=(1-a)L+c.

This model assumes the treaty applies to all losses in $L$, the ceded percentage is constant, and the premium is fixed. It does not include variable commissions, reinstatement premiums, exclusions, defaults, or other treaty features.

Theorem: Under those assumptions, $E[L_{\text{net}}]=(1-a)E[L]+c$ and $\operatorname{Var}(L_{\text{net}})=(1-a)^2\operatorname{Var}(L)$. A fixed premium changes the average loss cost but not its variance.

The result follows directly from the rules for expectation and variance. For a constant $b$ and random variable $L$, $E[bL+c]=bE[L]+c$. Subtracting the mean from the net loss gives

L_{\text{net}}-E[L_{\text{net}}] =(1-a)L+c-\big((1-a)E[L]+c\big) =(1-a)\big(L-E[L]\big).

Squaring and taking expectations yields

\operatorname{Var}(L_{\text{net}}) =E\left[\big(L_{\text{net}}-E[L_{\text{net}}]\big)^2\right] =(1-a)^2\operatorname{Var}(L).

The premium cancels because it is constant. This variance result does not apply unchanged when the premium varies with losses, when recoveries are capped or conditional, or when counterparty default makes recoveries uncertain.

Worked Example: Quota Share and the Cost of Smoother Earnings

A hypothetical insurer models annual covered losses, in millions of dollars, as follows: loss of $0 million with probability 0.50, $20 million with probability 0.30, and $80 million with probability 0.20. It considers a 25% quota share with a fixed annual reinsurance premium of $7 million. Assume recoveries are fully collectible and ignore taxes and expenses other than the stated premium. Calculate expected net loss and compare gross and net standard deviations.

Method. Calculate the gross loss moments first, then apply the fixed-share and fixed-premium formulas. The premium is included in net expected loss but does not affect net variance.

The gross expected loss is

$$ E[L]=0.50(0)+0.30(20)+0.20(80)=6+16=22. $$

The gross second moment is

E[L^2]=0.50(0^2)+0.30(20^2)+0.20(80^2) =120+1{,}280=1{,}400.

Therefore, the gross variance and standard deviation are

\operatorname{Var}(L)=1{,}400-22^2=1{,}400-484=916, \qquad \operatorname{SD}(L)=\sqrt{916}=30.2655.

The insurer retains 75% of each loss. Its net loss outcomes, after adding the $7 million premium, are $7 million, $22 million, and $67 million. Their expected value is

E[L_{\text{net}}]=0.50(7)+0.30(22)+0.20(67) =3.5+6.6+13.4=23.5.

Equivalently, $0.75(22)+7=23.5$. The net variance and standard deviation are

\operatorname{Var}(L_{\text{net}})=0.75^2(916)=515.25, \qquad \operatorname{SD}(L_{\text{net}})=\sqrt{515.25}=22.6991.

To four decimal places, gross standard deviation is $30.2655 million and net standard deviation is $22.6991 million. The treaty lowers the modeled standard deviation by $7.5664 million, while increasing expected loss cost from $22 million to $23.5 million. The check is that the ceded share reduces the standard deviation by the same factor, 0.75, and the fixed premium changes the mean only.

Common mistake avoided: Subtracting the $7 million premium from the variance, or treating it as if it increases the standard deviation. A deterministic expense shifts every outcome by the same amount; it does not change the distance between outcomes and their mean.

Layered Protection: Attachment Points and Limits

Excess-of-loss reinsurance responds only to loss above an attachment point, and its payment is capped by the contract limit. For a single covered loss $X$, attachment $r$, and limit $u$ on the reinsurer’s payment, the recovery is

R=\min\big((X-r)_+,u\big), \qquad (X-r)_+=\max(X-r,0).

This form assumes the limit is expressed as the maximum payment by the reinsurer, not as the maximum loss covered including the insurer’s retention. Treaty wording may use a different convention, so the attachment and limit must be read carefully before calculating. Ignoring default for the moment, retained loss is $X-R$.

Worked Example: Expected Cost of an Excess-of-Loss Treaty

A hypothetical insurer faces one covered event per year. The event loss, in millions of dollars, is $5 million with probability 0.60, $25 million with probability 0.30, and $60 million with probability 0.10. It buys per-event excess-of-loss protection attaching at $10 million, with a $30 million limit on the reinsurer’s payment. The fixed annual premium is $4 million. Assume the reinsurer pays all covered recoveries. Calculate expected gross loss, expected recovery, and expected retained loss including the premium.

Method. Apply the attachment and limit to each possible event loss before taking expectations. The payment is zero below attachment, the loss above attachment within the layer, and no more than $30 million.

At a $5 million loss, recovery is $\min(0,30)=0$, so retained loss is $5 million. At a $25 million loss, recovery is $\min(25-10,30)=15$, so retained loss is $10 million. At a $60 million loss, recovery is $\min(60-10,30)=30$, so retained loss is $30 million.

Expected gross loss is

$$ E[X]=0.60(5)+0.30(25)+0.10(60) =3+7.5+6=16.5. $$

Expected recovery is

$$ E[R]=0.60(0)+0.30(15)+0.10(30) =4.5+3=7.5. $$

Expected retained loss before premium is $16.5-7.5=9$ million. Including the fixed premium, expected net loss cost is $9+4=13$ million. As a check, the retained-loss outcomes are $5 million, $10 million, and $30 million, whose weighted mean is $0.60(5)+0.30(10)+0.10(30)=9$ million before premium.

Common mistake avoided: Treating the limit as a $30 million maximum gross loss, or paying the full loss whenever it exceeds $10 million. For a $25 million loss, this treaty pays $15 million, not $25 million; for a $60 million loss, it pays $30 million, not $50 million.

Counterparty Exposure and Recognized Protection

A contractual recovery is not necessarily the same as cash ultimately collected. A rating-relevant analysis should identify when the reinsurer must pay, whether the claim is within the treaty, and how the reinsurer’s ability and willingness to pay are assessed. Collateral, dispute risk, concentration in a single reinsurer, timing delays, and mismatches between the underlying claim and the reinsurance terms can all affect the value of protection.

A simple model can show why the distinction matters. If a loss produces a recovery $R$, but the reinsurer pays in full only when solvent, the actual recovery depends on both the covered loss and the counterparty’s payment event. The two events need not be independent. In particular, losses across the market may place pressure on a reinsurer at the same time that an insurer most needs its recovery.

Worked Example: Default Risk Reduces Catastrophe Protection

A hypothetical insurer has a $50 million catastrophe loss with probability 0.10 and no loss otherwise. A treaty would pay $30 million for the catastrophe. Conditional on the catastrophe occurring, assume the reinsurer defaults with probability 0.08 and pays nothing in that case; otherwise it pays the full recovery. Assume no default exposure when there is no catastrophe. Calculate the expected retained loss and standard deviation with this counterparty risk, then compare them with fully collectible reinsurance and no reinsurance. Ignore premium for the volatility comparison.

Method. Build the joint outcomes using the conditional default probability. Do not multiply the catastrophe probability by an unconditional default probability: the problem specifies default probability conditional on catastrophe.

With counterparty risk, retained loss is $0 million with probability 0.90; it is $20 million when catastrophe occurs and the reinsurer pays, with probability $0.10(0.92)=0.092$; and it is $50 million when catastrophe occurs and the reinsurer defaults, with probability $0.10(0.08)=0.008$. These probabilities sum to $0.90+0.092+0.008=1$.

The expected retained loss is

E[L_{\text{net}}]=0.90(0)+0.092(20)+0.008(50) =1.84+0.40=2.24.

The second moment and variance are

E[L_{\text{net}}^2]=0.092(20^2)+0.008(50^2) =36.8+20=56.8, $$ $$ \operatorname{Var}(L_{\text{net}})=56.8-2.24^2 =56.8-5.0176=51.7824.

Thus, standard deviation is $\sqrt{51.7824}=7.1960$ million, to four decimal places. If the treaty is fully collectible, retained loss is $0 million with probability 0.90 and $20 million with probability 0.10. Its mean is $2 million, its second moment is $0.10(20^2)=40$, its variance is $40-2^2=36$, and its standard deviation is $6 million. With no reinsurance, the retained loss is $0 million or $50 million with probabilities 0.90 and 0.10; its mean is $5 million, variance is $0.10(50^2)-5^2=225$, and standard deviation is $15 million.

The check is that counterparty risk makes the reinsured result worse than the fully collectible result, but still less volatile than retaining the entire catastrophe loss in this simplified model. This conclusion is specific to the stated assumptions; it is not a guarantee that every treaty reduces volatility after counterparty risk.

Common mistake avoided: Counting the $30 million contractual recovery in every catastrophe outcome. When default occurs, the actual recovery is zero, leaving the full $50 million loss with the insurer.

Connecting the Calculation to Capital Adequacy

A reduction in modeled loss volatility can support a stronger capital position, but it does not automatically translate dollar for dollar into rating capital relief. A capital assessment may recognize a benefit for reduced retained exposure while also considering the premium’s effect on earnings, the quality of the reinsurer, and the possibility that recoveries are delayed, disputed, or unavailable. Different assessment methods may treat these factors differently.

For a simplified management comparison, define capital headroom as available capital less a specified capital requirement. If a treaty reduces the assessed requirement, but its premium reduces available capital and a counterparty adjustment increases the requirement, all three changes belong in the comparison. This arithmetic is a decision aid, not a substitute for an agency’s model.

Worked Example: Testing Whether Capital Headroom Improves

A hypothetical insurer has $120 million of available capital and a rating-relevant capital requirement of $100 million before a proposed treaty. Management estimates that the treaty would reduce that requirement by $18 million, but an additional $6 million counterparty adjustment would apply. The $4 million treaty premium is paid from available capital. Calculate headroom before and after the treaty, and assess the change under these assumptions.

Method. Calculate capital headroom consistently as available capital minus the adjusted requirement.

Before the treaty, headroom is

120-100=20\text{ million dollars}.

After the treaty, available capital is $120-4=116$ million dollars. The adjusted requirement is $100-18+6=88$ million dollars. Therefore, post-treaty headroom is

116-88=28\text{ million dollars}.

Headroom improves by $28-20=8$ million dollars. The check is to reconcile the change: an $18 million reduction in the requirement, offset by a $6 million counterparty adjustment and $4 million premium cost, gives a net improvement of $18-6-4=8$ million dollars.

Common mistake avoided: Reporting the full $18 million modeled capital reduction as the increase in headroom. The premium reduces available capital and the counterparty adjustment increases the requirement, so both must be included.

Choosing Protection: The Trade-Offs to Test

Reinsurance is most useful when the particular risks it transfers are important to the insurer’s objectives and the protection is dependable at an acceptable cost. A lower net-loss standard deviation may be valuable, but management should not select a treaty using that measure alone. Relevant questions include:

What is the economic cost? Compare premium, commissions, reinstatement terms, and any collateral cost with the expected recovery and the value of reduced volatility.
Which outcomes are actually covered? Check attachment points, limits, exclusions, event definitions, aggregate provisions, and any mismatch between the underlying policy and the reinsurance contract.
How reliable is the recovery? Consider credit quality, collateral, concentration, dispute and payment risk, and whether the reinsurer could be stressed during a widespread loss event.
What happens to capital and earnings together? A treaty may reduce adverse-loss volatility while its cost lowers expected earnings or available capital. The assessment should use a consistent horizon and account for the relevant adjustments.
Does the protection remain useful under stress? Test outcomes beyond the limit, multiple events, changing exposure, and plausible counterparty impairment rather than relying only on an average recovery.

A common decision error is to assume that lower retained losses necessarily mean a stronger rating position. The correct analysis distinguishes contractual coverage from collectible recovery, and modeled capital relief from the net change in capital headroom. It also distinguishes a lower expected loss from a less volatile loss: a fixed premium can increase the expected cost even while the treaty reduces volatility.

Key takeaway: Evaluate reinsurance as a combined earnings, volatility, capital, and counterparty decision. Calculate the covered recovery from the contract terms, include the cost, and assess how much protection is likely to be recognized and collected under the adverse scenarios that matter.

Check Your Understanding

Work through each hypothetical case. Monetary amounts are in millions of dollars unless stated otherwise.

A hypothetical insurer has annual gross-loss standard deviation of $12 million and buys a 40% quota share. The premium is fixed and recoveries are fully collectible. What is the standard deviation of net loss? (Answer: $7.20 million)
An excess-of-loss treaty attaches at $10 million and has a $20 million payment limit. Annual event loss is $8 million, $18 million, or $40 million with probabilities 0.50, 0.30, and 0.20, respectively. The fixed premium is $3 million. What is expected retained loss including premium? (Answer: $14.00 million)
A $12 million recovery is due if a covered event occurs. Conditional on that event, the reinsurer has probability 0.10 of paying nothing and otherwise pays in full. What are expected recovery and the standard deviation of recovery, in millions? (Answer: Expected recovery $10.80 million; standard deviation $3.60 million)
Before reinsurance, available capital is $120 million and the capital requirement is $100 million. A treaty reduces the requirement by $18 million, adds a $6 million counterparty adjustment, and costs $4 million from available capital. What is post-treaty headroom, and how much does it change? (Answer: $28 million post-treaty headroom; an $8 million increase)
A catastrophe loss is $40 million with probability 0.10 and zero otherwise. A treaty would recover $25 million, but conditional on catastrophe the reinsurer defaults with probability 0.20 and pays nothing; otherwise it pays in full. What is the standard deviation of retained loss? (Answer: Approximately $6.7823 million)

What You'll Learn