The last bout of serious inflation in the United States occurred in the late 1970s. In casualty insurance, high inflation coincided with deteriorating reserves and underwriting results. Many believe that the risk of future inflation is higher than it has been in many years. If rising inflation levels impact settlement of claims that are open or are currently unreported, then increased inflation risk leads to increased reserve risk.
Accurate capital assessment requires assessing reserve risk, and while there are many ideas of what “reserve risk” encompasses, there are no clear definitions. As a result, insurers are not sure how to proceed. The North American actuarial profession has yet to establish standards of practice for measures and disclosures of insurer reserve risk; without such standards in place, companies are understandably reluctant to venture into unknown territory. A sound measure of reserve risk is vital to the executive team and Board, who need to understand the margin of error and potential bias in the reserves.
Guy Carpenter & Company, LLC recently conducted a study using 25 years of casualty insurance loss development data at the industry level. Guy Carpenter’s analysis found that industry reserves tend to be set at inadequate levels during soft cycles. In fact, the underlying actuarial indications themselves tend to be cyclically inaccurate. Even estimates that are intended to be conservative may be based on inaccurate analysis.
Solid and reliable reserve estimates require a rigorous statistical forecasting model similar to those used in data mining and forecasting exercises. Unfortunately, most of the standard actuarial methods are “based on” statistical models but are not statistical models themselves. Companies typically use a collection of methods to produce a “range of estimates.” Examining and evaluating the estimates is a key part of the actuarial work papers from the reserve review. Actuarial standards require that management’s best estimate (carried) reserve be within the range of reasonable estimates. However, selecting a carried reserve in that range tells the chief financial officer nothing about probability. How likely is the carried amount to be adequate? By how much could it be deficient, and with what degree of probability? In order to understand these questions around reserve risk, we need access to new statistical reserving methods that make full use of statistical models.
Reserve forecasts are based on the assumption that the past is representative of the future. Several methods for statistical assessment of reserve risk that have been popularized also assume an unchanging world, and therefore miss the most important risk. The real risk for insurers lies in the possibility that the future will not be the same as the past. Inflation risk is an obvious reason, although not the only reason. Accurate risk evaluation requires a forecast of possible volatility in future inflation and through an analysis of past patterns of inflation, an analysis of possible future effects of inflation.
Utilization of rigorous statistical models provides the framework needed for accurate reserve establishment for casualty insurers. Guy Carpenter’s Dynamic Reserve Model (DRM™), a practical, hands-on tool using a groundbreaking methodology, assesses reserve levels, reserve risk and prospective risk levels.