Claims Reserving

Claims reserving is the process by which an insurer estimates the amount it will have to pay in future for losses that have already happened. Insurance is unusual in that the price is charged before the cost is known: premiums are received up front, while claims may be reported and settled months or years later. The reserve is the insurer's best assessment of those outstanding obligations at the balance sheet date.

The estimate is produced mainly by actuaries, using data from the claims department. It is usually the largest liability on a general insurer's balance sheet, so its accuracy determines reported profit and capital. The resulting figure is the claims reserve, also called the loss reserve or, under current accounting standards, the liability for incurred claims.

Outstanding claims fall into several groups:

• Case reserves. These are estimates set by claims handlers for each claim that has been reported but not yet paid in full.
• Incurred but not reported. This is an allowance for losses that have occurred but have not yet been notified to the insurer; see incurred but not reported (IBNR) claims.
• Incurred but not enough reported. This allows for future increases in the cost of claims already reported, where case reserves prove too low. It is often combined with the previous item.
• Reopened claims. Some closed files will need further payment.
• Claims handling costs. The future loss adjustment expense (LAE) of settling all these claims is reserved as well.

Reserves are calculated gross and net of reinsurance, since part of the cost will be recovered under the insurer's reinsurance cession.

Most methods start from a development triangle, a table that shows how claims from each period of origin have grown over time. Building and interpreting these tables is the subject of claims triangulation. The main techniques are:

• Chain ladder. Past patterns of growth from one stage to the next are summarised in a loss development factor for each stage and applied to recent periods to project their ultimate cost. It assumes that the past pattern will repeat.
• Expected loss ratio. The ultimate cost is set as premium multiplied by an assumed loss ratio, ignoring the claims experience so far. It is used where data is thin.
• Bornhuetter-Ferguson. This blends the two: claims already reported are taken as they are, and the amount still to emerge is based on the expected loss ratio. It is more stable for recent, immature periods.
• Frequency and severity. The number of claims and the average cost are projected separately.
• Stochastic methods. Statistical models and simulation produce a range of outcomes and not just a single figure.

Actuaries normally apply several methods to paid and incurred data, compare the results and select an estimate using judgement about changes in the business, claims handling, inflation and the legal environment. Classes of business in which claims emerge slowly, such as liability, are described as long-tail and are much harder to reserve than short-tail classes such as property.

Different frameworks set different bases. Under IFRS 17, which took effect in 2023, the liability for incurred claims is a probability-weighted estimate of future cash flows, discounted to present value, plus an explicit risk adjustment for uncertainty in non-financial risk. Under Solvency II in the European Union, technical provisions consist of a discounted best estimate and a risk margin, and they are the starting point for the solvency capital requirement. United States statutory accounting for property and casualty insurers generally requires undiscounted reserves and an annual opinion from an appointed actuary.

Insurers are also expected to understand the uncertainty around the figure. Reserve risk, the risk that reserves prove inadequate, is one of the main components of regulatory capital and is examined in the own risk and solvency assessment (ORSA).

A change in the estimate for earlier years flows straight into the current year's profit. If prior-year claims turn out cheaper than reserved, the surplus is released and improves the result, an effect known as a reserve release or favourable development. If they turn out dearer, reserves are strengthened and the result worsens. Analysts therefore look at the combined ratio both with and without prior-year development, and review the published tables of how past estimates have run off.

Because reserves involve judgement, they can be used to smooth results: prudent in good years, optimistic in bad. Persistent under-reserving has been a leading cause of insurer failure, since it overstates profit, encourages underpricing and may come to light only after years of growth. Reserving therefore depends on reliable case estimates from claims administration and on consistent premium data from premium accounting.

Reserve estimates can be upset by factors that past data does not capture. Claims inflation, driven by wages, repair and medical costs, raises settlements above the level assumed. Court decisions and legislation can widen liability or change compensation, sometimes with retrospective effect. Latent claims, of which asbestos-related disease is the best-known example, may emerge decades after the policies were written. Changes in the mix of business, policy terms or the speed of claims handling distort development patterns. Governance arrangements, including independent actuarial review, reserving committees and scrutiny by auditors and the board, exist to test the assumptions against these risks and to protect the insurer's policyholder surplus.

• IFRS Foundation
• European Insurance and Occupational Pensions Authority