1998 Reserves Papers Search for the "Best Estimate"

by Joanne S. Spalla

At the Casualty Loss Reserve Seminar on September 28-29 in Philadelphia, six sessions were devoted to the 1998 Call for Papers sponsored by the Committee on Reserves. The topic of this year's call was "Best Estimates for Reserves." The eleven papers submitted in response to the call explored the issues that actuaries face in recommending the "best" estimate to book.

In the first session, William Lakins described a model to determine "Efficient Estimators Through Data Segmentation." Richard Vaughan then discussed "Some Extensions of J.N. Stanard's Simulation Model for Loss Reserving," which explored a wide variety of questions arising in loss reserving.

In a concurrent session, Glen Barnett presented the paper, "Best Estimates for Reserves," which he co-authored with Ben Zehnwirth. The paper established a methodology to determine whether assumptions underlying standard actuarial techniques are supported by the data. Barnett went on to develop and test a probabilistic model of the loss process, using regression techniques to estimate the distributions.

In the third session, Richard Stein and Michael Stein explored the cognitive, operational, and methodological issues that affect the quality of the actuary's reserve estimate in their paper, "Sources of Bias and Inaccuracy in the Development of a Best Estimate." Paul Struzzieri and Paul Hussian, authors of "Using Best Practices to Determine a Best Reserve Estimate," employed a case study to illustrate a set of best practices that the actuary could follow at each step of the reserving process to minimize the impact of biases.

On Tuesday, Kathleen Blum and David Otto discussed terminology and communication issues surrounding the selection of a best estimate reserve. Their paper, "Best Estimate Loss Reserving: An Actuarial Perspective," surveyed the current accounting, tax, and actuarial regulations concerning the selection of a best reserve estimate. In the same panel, Michael McCarter and Jan Lommele, authors of "Is the 'Best Estimate' Best?," contrasted the "best estimate" wording of the NAIC Codification Issue Paper Number 55 with the "reasonable provision" wording of the annual statement instructions for the statement of actuarial opinion. The panelists also discussed practical considerations regarding the actuary's communication of a point estimate to management.

In the next session on "Estimating the Variability of Loss Reserves," Richard Sherman examined a new family of loss distributions that fit simulated claim data. Also during this session, Alfred Raws III presented the 1998 Reserves Prize paper, "Statistical Modeling Techniques for Reserve Ranges: A Simulation Approach," which he wrote with Chandu Patel. Raws described approaches from this prize-winning paper that utilized both simulation and judgment to generate a range of reserves and provide a basis for selecting the best estimate.

In the final session, both panelists discussed the application of utility theory to the selection of the best estimate. In "Using Utility Theory For Describing Best Estimate Reserves" Mark Littmann discussed the results of a survey of professionals, including actuaries, regulators, company management, and shareholders. He also presented a case study that illustrated different utility functions. Robert Buchanan, author of "The Philosophy of Reserving," discussed the issues of discounting, inflation, variability, and risk margins. His paper utilizes a time series of successive estimates to approach the best estimate.

Readers interested in learning more about these papers are directed to the fall edition of the CAS Forum, which was distributed in September. The papers are also available in the Download Library.