8. Reinsurance Pricing, Reserving and Profitability

Introduction and context

Reinsurance converts uncertain insurance losses into a financial arrangement that can be priced, reserved, monitored and evaluated. A reinsurer does not simply charge a percentage of the cedant’s premium. It estimates the expected cost of covered losses, considers volatility and catastrophe potential, allows for acquisition and administration expenses, provides a return for capital and risk, and accounts for the timing of cash flows. Pricing and reserving are therefore closely connected: inadequate pricing weakens future profitability, while inadequate reserving can make apparently profitable treaties economically unsound.

This chapter examines the principal methods used to price reinsurance, including experience rating, exposure rating, burning-cost analysis, catastrophe-model outputs and rate-on-line analysis. It then considers premium adequacy, reserve estimation, incurred-but-not-reported claims, cash-flow effects and treaty profitability. The emphasis is on practical interpretation rather than mechanical calculation. In professional practice, actuarial models, underwriting judgement, contract wording, data quality, capital requirements and governance all influence the final decision.

Learning objectives

After studying this chapter, the reader should be able to:

1. Distinguish experience rating, exposure rating and burning-cost methods.

2. Explain the construction of expected loss cost and the loading of expenses, risk and profit.

3. Interpret rate-on-line and other reinsurance pricing measures.

4. Explain the role of catastrophe models and their limitations.

5. Estimate unpaid claims, including IBNR, and relate reserves to cash flow.

6. Assess premium adequacy and treaty profitability using appropriate performance measures.

7. Identify data, model, governance and ethical risks in reinsurance pricing and reserving.

The economic logic of reinsurance pricing

The starting point of a reinsurance price is the expected loss cost. For a proportional treaty, the reinsurer may receive a share of the cedant’s premium and pay the same share of covered losses, subject to commissions and other terms. For a non-proportional treaty, such as excess-of-loss cover, the reinsurer pays losses within an attachment and exhaustion range. The price must reflect the probability and severity of losses entering that layer.

A simplified pricing equation is:

Required premium = expected loss cost + expenses + risk/capital loading + profit loading − investment-income allowance.

The expected loss cost is not necessarily the same as the historical loss ratio. Historical claims may be immature, affected by inflation, distorted by unusual catastrophes, or generated by a book of business that has changed materially. The pricing actuary therefore adjusts data for claims development, exposure growth, limits, deductibles, wording, inflation, social trends, geographic concentration and changes in underwriting.

Expenses commonly include brokerage, acquisition costs, internal administration, claims handling, data and modelling costs, taxes and the cost of collateral or retrocession. The risk loading compensates the reinsurer for uncertainty around the mean loss, including parameter risk, process risk and catastrophe accumulation. The profit loading provides a target return consistent with capital consumed and strategic objectives.

A useful distinction is between technical price and market price. The technical price is the price indicated by the risk analysis. The market price is the amount that can actually be negotiated. A market price below the technical price may be accepted temporarily for strategic reasons, but the difference should be explicit and governed rather than hidden in optimistic assumptions.

Pricing methods and when they are appropriate

Experience rating

Experience rating uses the cedant’s historical losses to estimate the future cost of a treaty. It is most useful where the portfolio is sufficiently large, stable and credible, and where past claims resemble future claims. A basic form is:

Indicated loss cost = adjusted historical layer losses ÷ adjusted exposure.

For an excess-of-loss treaty, historical losses are first developed to ultimate values and trended to the prospective period. They are then adjusted to the proposed attachment point, limit and contract terms. If a treaty has experienced losses of 8, 12 and 10 million over three years, and adjusted exposures are 100, 110 and 120 million, the simple historical loss ratio is 30/330, or 9.09 per cent. That figure is only a starting point; credibility, development, trend and changes in exposure must be considered.

Experience rating becomes unreliable when the portfolio is small or highly volatile. A loss-free history does not demonstrate that the risk is loss-free. Similarly, one severe event can dominate the result. Credibility procedures address this problem by blending the experience indication with an exposure-based indication:

Blended indication = credibility weight × experience indication + (1 − credibility weight) × exposure indication.

The credibility weight should reflect the volume, quality and stability of data rather than commercial preference.

Exposure rating

Exposure rating estimates losses from the characteristics of the insured portfolio rather than relying primarily on its own historical claims. Exposure measures may include insured values, payroll, revenue, vehicle counts, limits, geographic distribution or catastrophe-zone aggregates. Industry loss curves, damage functions or portfolio benchmarks are applied to those exposures to estimate the probability and severity of losses entering a reinsurance layer.

Exposure rating is particularly important when experience is sparse, when the treaty covers a new portfolio, or when catastrophe losses are infrequent but potentially severe. It is also useful for assessing changes in limits and attachment points. Its weakness is dependence on external assumptions and the quality of exposure data. A portfolio that appears similar to an industry benchmark may differ in construction quality, deductibles, claims practices or geographic concentration.

Burning-cost method

The burning-cost method is a practical form of experience rating often used for working-layer and excess-of-loss treaties. Historical losses falling into the layer are aggregated and divided by a measure of exposure, commonly subject premium:

Burning cost = layer losses ÷ subject premium.

Suppose adjusted losses within a treaty layer total 6 million and subject premium is 60 million. The burning cost is 10 per cent. If expenses, risk and profit loadings total 25 per cent of premium, the indicated premium is approximately 6 ÷ (1 − 0.25) = 8 million, or 13.33 per cent of subject premium. This calculation must be interpreted carefully. Losses should normally be developed, trended and adjusted for changes in limits, attachment, claims inflation and coverage. Catastrophe losses may be separately analysed or capped when the objective is to estimate attritional performance.

Catastrophe-model output

Catastrophe models combine hazard, exposure, vulnerability and financial modules. A windstorm model, for example, may simulate event frequency and intensity, estimate physical damage to insured locations and apply policy conditions before producing treaty-level loss distributions. Important outputs include:

• Average annual loss (AAL): the mean simulated annual loss.
• Occurrence exceedance probability (OEP): the probability that at least one event exceeds a specified loss in a year.
• Aggregate exceedance probability (AEP): the probability that total annual loss exceeds a specified amount.
• Return-period loss: a loss associated with a selected exceedance probability, subject to the model’s interpretation.

For a catastrophe excess-of-loss layer, the modelled expected loss to the layer may be combined with expenses and a risk margin. AAL alone is not a complete price because two portfolios with the same AAL may have different tail risk, diversification and capital requirements. Model output should therefore be supplemented by scenario testing, exposure validation and expert judgement.

Rate-on-line

Rate-on-line (ROL) is a common market measure for non-proportional reinsurance:

ROL = reinsurance premium ÷ reinsurance limit.

A premium of 12 million for a 100 million limit produces a 12 per cent ROL. The reciprocal, often called the payback period, is 8.33 years, although it should not be interpreted as a forecast that the layer will suffer a full loss only once every 8.33 years. ROL is useful for comparing layers and market prices, but it does not reveal attachment probability, expected loss, reinstatement provisions, expenses or capital consumption. A low-attaching layer may have a lower limit but substantially higher expected loss than a remote catastrophe layer with the same ROL.

Premium adequacy and treaty economics

Premium adequacy requires comparison between the price and the full economic cost of the treaty. A simple expected combined ratio is:

Expected combined ratio = expected losses and loss adjustment expenses ÷ premium + expenses ÷ premium.

A ratio below 100 per cent suggests underwriting profit before considering capital costs and some broader corporate expenses. However, the appropriate target depends on volatility, required return, duration of liabilities, investment income, tax, retrocession and capital allocation.

For proportional treaties, commission is a central pricing variable. A high cedant commission reduces the reinsurer’s premium available to fund losses and expenses. Sliding-scale commissions can align remuneration with loss performance, but they introduce complexity and require clear definitions of loss, premium, caps, floors and settlement dates. Profit commissions can also redistribute underwriting results and should be analysed together with claims development and contract accounts.

Profitability should be evaluated on both underwriting and economic bases. Useful measures include underwriting margin, return on allocated capital, risk-adjusted return on capital, economic value added and present-value profit. The analyst should distinguish gross results from net results after retrocession, brokerage, reinstatement premiums, collateral costs and overhead allocation.

Reserving: from reported claims to ultimate liability

A reinsurer’s reserve is its best estimate of obligations arising from covered events, together with appropriate risk margins where required by the applicable accounting or regulatory framework. The basic relationship is:

Ultimate loss = paid loss + case outstanding + IBNR.

IBNR, or incurred-but-not-reported reserve, includes claims that have occurred but have not yet been reported and development on reported claims that is not fully captured in case reserves. In practice, IBNR is often a residual actuarial estimate rather than a directly observable category.

Claims development methods

The chain-ladder method uses historical development factors to project current cumulative claims to ultimate values. It is effective when development patterns are stable and the portfolio is mature. The Bornhuetter–Ferguson method combines an a priori expected loss ratio with observed development and is often more stable for immature years or volatile portfolios. The expected-loss method applies an expected loss ratio directly when little credible claims experience is available. Frequency-severity methods separately model claim counts and amounts and can be useful where exposure information is strong.

A reserving analysis should examine paid and incurred triangles, changes in case reserve adequacy, large-loss activity, claim reporting delays, settlement practices, inflation, coverage disputes and exposure mix. Reinsurance reserving also requires attention to contract features: attachment, limits, reinstatements, hours clauses, event definitions, aggregate caps, exclusions, commutations and reinstatement premiums.

Cash flow and the time value of money

Premium is often received before claims are paid. This creates investable funds, but it also creates obligations and liquidity risk. Claims may be paid rapidly after a catastrophe, while premium may be received in instalments. A treaty can be profitable on an undiscounted basis yet produce adverse short-term cash flow if claims are accelerated or collateral is required.

Discounting is relevant when liabilities are long-tailed and payment timing is material. The present value of expected claims depends on the selected yield curve, currency, duration and liquidity assumptions. Investment income should not be treated as a substitute for adequate underwriting price: asset returns are uncertain and may not be available when catastrophe claims require payment.

An applied case: pricing a regional property catastrophe layer

Background

This illustrative case concerns Meridian Re, which is considering a one-year property catastrophe treaty for a regional insurer. The proposed cover is 50 million excess of 100 million, with one paid reinstatement. Subject premium is expected to be 200 million. The cedant supplies exposure data by location, construction type and territory, together with five years of loss experience.

Situation and analysis

The historical experience contains only one material catastrophe and is not considered fully credible. After development, inflation adjustment and contract-term adjustment, the experience-based expected loss to the layer is 7 million. An exposure-based catastrophe model produces an AAL of 8.5 million. The underwriting team assigns 35 per cent credibility to experience and 65 per cent to the model:

Blended expected loss = (0.35 × 7) + (0.65 × 8.5) = 7.975 million.

The team adds 0.8 million for expected reinstatement-related cost, 1.2 million for expenses and 2.0 million for risk and profit loadings. The indicated premium is therefore 11.975 million, approximately 12 million. The ROL is 12/50 = 24 per cent. The model also indicates significant tail risk because losses above the attachment are concentrated in a small number of coastal zones.

Before accepting the price, Meridian Re validates geocoding, checks whether insured values are replacement-cost or market-value measures, tests demand surge and inflation assumptions, and runs alternative hazard and vulnerability views. It also considers whether one reinstatement is sufficient and whether the cedant’s exposure growth assumption is credible.

Strategy and outcome

The cedant requests a premium of 10.5 million. Meridian Re does not simply reduce the price. It offers alternatives: a higher attachment, a lower limit, or a conditional price subject to verified exposure data and a more restrictive event definition. The final negotiated treaty has a premium of 11.5 million with improved data warranties and a revised reinstatement provision. The expected technical margin is narrower than the original indication, but the data and wording improvements reduce uncertainty.

Learning points

The case demonstrates that model output is an input rather than an automatic price. Credibility blending can avoid over-reliance on sparse history or an external model. ROL facilitates communication but does not replace expected-loss and tail-risk analysis. Contract wording and data quality can be as important as the numerical premium.

Common pitfalls, governance and professional responsibility

Pricing and reserving errors often arise from inconsistent exposure definitions, silent changes in attachment points, double counting of reinstatement premiums, failure to develop claims, inappropriate inflation assumptions or treating a model’s central estimate as fact. Analysts may also underestimate correlation between territories or lines of business, particularly when a single event can affect several treaties.

Governance should include documented assumptions, independent review, version control, model validation, sensitivity testing, clear authority limits and monitoring of actual-versus-expected results. Underwriters, actuaries, claims specialists, exposure managers and finance professionals should challenge one another constructively. The board and risk committee need information about aggregation, tail exposure, reserve uncertainty and capital adequacy rather than only headline premium volume.

Ethically, pricing should not exploit information asymmetry unfairly or conceal uncertainty from cedants and stakeholders. Claims reserving must be prudent, unbiased and consistent with contract terms. Conflicts of interest should be disclosed, especially where remuneration depends on growth or reported profit. Regulatory reporting, accounting standards, solvency requirements, sanctions controls and data-protection obligations must be observed in the relevant jurisdiction.

Digital tools and emerging developments

Modern reinsurance operations increasingly use automated exposure ingestion, geospatial analytics, cloud-based catastrophe platforms, claims dashboards and predictive reserving systems. Machine-learning techniques can identify reporting delays, segment claims and improve anomaly detection. However, predictive performance can deteriorate when the future differs from the training data, as occurs after legal, climatic or economic regime change. Explainability, data lineage, bias testing and human oversight are essential.

Climate-change uncertainty is encouraging scenario analysis beyond historical experience. Reinsurers are also examining secondary perils, cyber accumulation, systemic business interruption and supply-chain dependence. Parametric covers use objectively measured indices, such as wind speed or earthquake intensity, and can improve speed of payment, but basis risk must be analysed carefully. Alternative capital, collateralised reinsurance and insurance-linked securities make capital-market pricing increasingly relevant to treaty decisions.

Best-practice workflow

A disciplined pricing and reserving process normally follows these steps:

1. Define the contract, covered peril, attachment, limit, reinstatement terms and accounting basis.

2. Validate exposure, premium and claims data and document exclusions and changes in mix.

3. Develop and trend losses, separating attritional, large and catastrophe experience.

4. Select and, where appropriate, blend experience, exposure, burning-cost and catastrophe approaches.

5. Estimate expected loss, expenses, risk capital and profit requirements.

6. Test sensitivity to inflation, frequency, severity, exposure growth, model assumptions and wording.

7. Assess premium adequacy, cash flow, liquidity, retrocession and capital consumption.

8. Set initial and ultimate reserves using suitable actuarial methods and update them as evidence develops.

9. Monitor actual-versus-expected results, reserve deterioration, aggregation and emerging risks.

10. Record judgement, approvals and deviations from technical indications.

Chapter Summary

Reinsurance pricing begins with an estimate of the expected cost of covered losses and adds expenses, risk capital and profit requirements, while considering investment income and cash-flow timing. Experience rating is strongest where credible history exists; exposure rating is valuable for new or sparse portfolios; burning cost is a practical historical method; and catastrophe models help estimate infrequent, severe losses and tail distributions. Rate-on-line is a useful market ratio but cannot substitute for layer-level loss and capital analysis.

Reserving requires estimation of ultimate claims, including IBNR, using methods such as chain-ladder, Bornhuetter–Ferguson and frequency-severity analysis. Treaty profitability must be assessed after considering claims development, reinstatements, retrocession, expenses, capital and liquidity. Sound practice combines quantitative models with expert judgement, governance, data validation, sensitivity testing and ethical transparency.

Key Terms

Expected loss cost: The estimated average cost of covered losses before expenses and profit loadings.

Experience rating: Pricing based primarily on the cedant’s adjusted historical loss experience.

Exposure rating: Pricing based on portfolio exposures and external loss distributions or benchmarks.

Burning cost: Historical layer losses expressed as a proportion of subject premium or another exposure base.

Credibility: The degree to which an experience sample is considered reliable for predicting future results.

Catastrophe model: A model combining hazard, exposure, vulnerability and financial terms to estimate catastrophe losses.

Average annual loss: The mean simulated or expected loss in a year.

OEP: The probability distribution of the largest event loss in a year.

AEP: The probability distribution of total annual loss in a year.

Rate-on-line: Reinsurance premium divided by the limit of the layer.

Attachment point: The loss level above which an excess-of-loss layer begins to pay.

Exhaustion point: The loss level at which the available limit of a layer is fully used.

IBNR: Estimated liability for incurred claims not yet reported or not fully developed.

Chain-ladder method: A reserving method that projects ultimate claims using historical development factors.

Bornhuetter–Ferguson method: A reserving method blending an a priori expected loss with emerging claims experience.

Premium adequacy: The extent to which premium covers expected costs, risk capital and required return.

Reinstatement premium: Additional premium restoring some or all of a reinsurance limit after a loss.

Combined ratio: Losses and expenses divided by premium.

Basis risk: The risk that an index or contract payment does not match the actual economic loss.

Risk-adjusted return on capital: Return evaluated relative to the capital required for the risk assumed.

Review Questions

Short Answer Questions

1. What is the difference between technical price and market price?

2. Define burning cost and state its basic formula.

3. What do OEP and AEP measure?

4. Why is IBNR included in ultimate loss estimates?

5. State two limitations of rate-on-line as a pricing measure.

Long Answer Questions

1. Explain the experience-rating process for an excess-of-loss treaty, including adjustments required before relying on historical losses.

2. Compare experience rating, exposure rating and burning-cost methods, indicating when each is most appropriate.

3. Explain how catastrophe-model outputs are used in reinsurance pricing and why AAL alone is insufficient.

4. Discuss the principal actuarial approaches to reserving and their suitability for immature or volatile portfolios.

5. Explain how expenses, capital, profit, investment income and cash-flow timing influence treaty profitability.

Analytical / Application Questions

1. A layer has adjusted historical losses of 9 million and subject premium of 75 million. If expenses are 20 per cent of premium and the required risk and profit loading is 10 per cent of premium, calculate an indicated premium using a simple burning-cost approach and discuss two further adjustments that may be needed.

2. Experience rating indicates an expected loss of 6 million and exposure rating indicates 10 million. If credibility assigned to experience is 40 per cent, calculate the blended expected loss and explain why the result should not be accepted without sensitivity testing.

3. A treaty has a premium of 15 million and a limit of 60 million. Explain its ROL, identify why the figure may be misleading when comparing it with another layer, and list the additional information needed to assess profitability.

22. Multiple Choice Questions with Answers

Question 1

Which pricing method is generally most suitable when a cedant has sparse historical claims experience but detailed exposure information?

A. Exposure rating

B. Pure experience rating

C. Unadjusted burning cost

D. Claims settlement analysis only

Correct Answer: A. Exposure rating

Question 2

A treaty layer has adjusted losses of 6 million and subject premium of 60 million. What is its burning cost?

A. 1 percent

B. 6 percent

C. 10 percent

D. 16.7 percent

Correct Answer: C. 10 percent

Question 3

What does rate-on-line primarily measure?

A. Expected claims divided by total assets

B. Reinsurance premium divided by layer limit

C. Subject premium divided by claims incurred

D. Capital divided by underwriting profit

Correct Answer: B. Reinsurance premium divided by layer limit

Question 4

Why is average annual loss insufficient by itself to price a catastrophe layer?

A. It excludes all attritional claims

B. It does not describe tail risk and loss distribution adequately

C. It is always based only on reported claims

D. It measures premium rather than loss

Correct Answer: B. It does not describe tail risk and loss distribution adequately

Question 5

Which item is included in the relationship Ultimate loss = paid loss + case outstanding + IBNR?

A. Brokerage income

B. Investment profit

C. Incurred-but-not-reported claims

D. Shareholders’ dividend

Correct Answer: C. Incurred-but-not-reported claims

Question 6

An experience indication is 7 million and an exposure indication is 8.5 million. If experience receives 35% credibility, what is the blended indication?

A. 7.35 million

B. 7.98 million

C. 8.15 million

D. 8.50 million

Correct Answer: B. 7.98 million

Question 7

Which reserving method combines an a priori expected loss ratio with emerging claims experience?

A. Chain-ladder

B. Bornhuetter–Ferguson

C. Simple payback method

D. Rate-on-line method

Correct Answer: B. Bornhuetter–Ferguson

Question 8

A premium of 12 million is charged for a 50 million reinsurance limit. What is the ROL?

A. 4.17 percent

B. 12 percent

C. 24 percent

D. 62 percent

Correct Answer: C. 24 percent

Question 9

Which development most directly creates liquidity pressure for a reinsurer?

A. Claims are paid rapidly while premium arrives in instalments

B. Claims are permanently eliminated by modelling

C. The attachment point is documented clearly

D. Exposure data are independently validated

Correct Answer: A. Claims are paid rapidly while premium arrives in instalments

Question 10

What is a key governance response to reliance on catastrophe models?

A. Accept the central estimate without review

B. Remove underwriting judgement from pricing

C. Perform validation, sensitivity testing and exposure-data checks

D. Use only the highest modelled loss

Correct Answer: C. Perform validation, sensitivity testing and exposure-data checks