Topic: Applying and Evaluating Generalized Linear Models
Generalized Linear Models (GLM) is a statistical tool generating increasing interest among casualty actuaries. GLM is frequently used for classification ratemaking in personal automobile lines, but the underlying theory offers the potential for broader application. There are a number of textbooks and papers that describe and advance the theory of Generalized Linear Models, but there is a need for practical guidance in the application of this theory to specific actuarial problems.
The 2004 Discussion Paper Program will focus on the practical application of GLM. Potential authors are encouraged to present case studies in detail sufficient for the reader to reproduce results. Ideally, the Program will be a forum for exchanging ideas of practical value to current practitioners as well as to potential new users of GLM.
Some examples of topics for discussion might include:
- How to conduct and interpret "model checking" tests, e.g., residual plots.
- A survey of the different models most relevant to insurance data. This topic would include model assumptions and tests to validate those assumptions.
- Applying GLM to areas outside of pricing, e.g., reserving or marketing strategy.
- Identifying factors relevant to specific lines of business. Economic or insurance reasons why a model should include a particular factor or interaction of factors.
- Modeling interactions between factors.
- The connection between credibility theory and GLM. How to incorporate credibility concepts within GLM.
Accepted papers will be published in the Spring 2004 Discussion Paper Program book and will be posted on the CAS Web Site. Authors will have the opportunity to present their papers at sessions scheduled during the CAS Spring Meeting in Colorado Springs, Colorado on May 16-19, 2004.
The 2004 Discussion Paper Program is managed by the Continuing Education Committee. The Committee encourages your participation in the Program. The results will contribute to the written body of knowledge of the CAS and to the success of the 2004 Discussion Paper Program.
Schedule for the Program
To participate in the Discussion Paper Program, prepare and submit a proposal that includes a title and short description of your paper. The description should outline the key issues of your paper and the conclusions you intend to reach. It would be helpful if you could describe the methods you might use if the paper is to be technical in nature. This would give the reviewing committee the best possible opportunity to guide you in reaching the widest possible audience.
When a proposal is submitted, the Continuing Education Committee members do not know the name of the author as they decide to accept or reject the proposals. This is done to make sure the process is as fair as possible. You can help by omitting any references to companies or people in your proposal.
The general schedule for the 2004 Discussion Paper Program is as follows:
Deadline for Proposals
The proposals must be received by September 30, 2003. They should be submitted to:
Mr. Roger W. Bovard
Discussion Paper Chairman
c/o CAS Office at firstname.lastname@example.org
If you wish to discuss an idea for a paper with Roger, he would welcome your call at (818) 879-4406.
Acceptance or Rejection of Proposals
By October end, the Committee on Continuing Education will have reviewed each proposal and made a decision to accept or reject it. The Chairman will contact each author regarding the Committee's decision.
Monitoring the Progress
Timely completion of the papers will be necessary so that all the papers can be made available to CAS members well before the 2004 Spring Meeting. The Chairman will be in regular contact with each author and will provide general guidance in completing the paper. Please note that the papers should be prepared in accordance with the "Guides for Submission of Papers" in the 2003 CAS Yearbook (pp. 352-359) and on the CAS Web Site at http://www.casact.org/about/index.cfm?fa=guides. It is hoped that the authors will also submit their papers for publication in the Proceedings; however, papers must be submitted separately to the Proceedings for independent consideration.
The Discussion Paper Chairman must receive the completed paper and a 200-word abstract by December 19, 2003.
Approving the Completed Paper
Each paper will be reviewed and screened by the Committee on Continuing Education. Occasionally, an initially accepted proposal may result in a paper that does not meet the minimum standards established by the Committee either due to relevance of material or quality of exposition. When that happens, some re-writing may be required, or the paper may not be included in the program.
Presentation at the CAS Meeting
Discussion of Papers
All authors will be invited and expected to discuss their papers at the sessions scheduled during the 2004 Spring Meeting in Colorado Springs, Colorado. Authors will be contacted in advance of this meeting regarding the nature of the presentation and coordination with the program.
Preparation of Bound Papers
All accepted papers meeting the deadlines will be printed, bound, and distributed in advance of the meeting.
Michelbacher Prize Committee
The Michelbacher Prize Committee will review submissions anonymously and decide upon a winner in accordance with the criteria sent to the authors once their proposals have been accepted. The Michelbacher Prize, in the amount of $1,500, will be awarded to the author of the best paper and presented at the 2004 Spring Meeting.