|
1. We've developed a two-day course on the theory of linear statistical modeling and its actuarial applications. We give particular attention to modeling loss triangles. In 2005 we presented this course four times at three reinsurance companies, most recently in Munich for the actuaries of GE/Frankona (Press Release). The linear statistical model is a powerful and practical tool for the prediction of multivariate means and variances; in fact, one valid statistical model is worth more than any number of deterministic methods. This course is much less expensive than statistical software, and lets (re)insurance companies be self-reliant and creative. Contact us to find out.
2. For years we've championed utility-theoretic pricing, as opposed to risk-adjusted discounting and return on equity. Let the curious open the following two files for a presentation at a CAS Limited-Attendance Seminar in September, 2005:
Utility-Theoretic Underwriting.ppt Utility-Theoretic Underwriting.xls
It may be difficult to understand the files without the commentary; however, the ideas are developed in our 2003 paper The Valuation of Stochastic Cash Flows, which you can download at www.casact.org/pubs/forum/03spforum. Naturally, it'd be our pleasure to talk to you about this, the science of risk taking.
3. Many actuaries attempt to make statistical models out of the Chain-Ladder method. Recently we wrote an in-depth critique on the bias of this method, which appeared in the Fall 2007 issue of Variance. You can download here a copy of Chain-Ladder Bias: Its Reason and Meaning. From it you will learn the true reason why this method commonly overpredicts, and will gain a new understanding of credibility. | 
