Search Presentations
STAY TUNED! If you are anticipating additional search filters by attribute and level to align with the CAS Capability Model, it is coming later this Summer. As the CAS begins to code recorded sessions by specific attributes and levels (starting with the 2023 Annual Meeting), these will be tagged in the CAS database of presentations going forward and should be searchable.
But you may use the Capability Model now to help you identify topics. For example, if you want to move up one level under the content area “Functional Expertise,” you may search topics in the particular functional area to expand your knowledge.
Recorded content is searchable by Capability Model attribute and level in the CAS Online Library.
Refine Your Search
The presentation materials are offered in connection with CAS professional education offerings. © 2022 Casualty Actuarial Society. All Rights Reserved. The presentation materials may contain copyrighted content the use of which has not been specifically authorized by the copyright owner. You are permitted to view and print the materials for personal/professional noncommercial research purposes. Except for the foregoing, you agree not to reproduce, distribute, modify, create derivative works, or commercially exploit the presentation materials without prior written permission from CAS. Please direct any copyright permission inquiries regarding use of the presentation materials to acs@casact.org.
STAY TUNED! If you are anticipating additional search filters by attribute and level to align with the CAS Capability Model, it is coming later this Summer. As the CAS begins to code recorded sessions by specific attributes and levels (starting with the 2023 Annual Meeting), these will be tagged in the CAS database of presentations going forward and should be searchable.
But you may use the Capability Model now to help you identify topics. For example, if you want to move up one level under the content area “Functional Expertise,” you may search topics in the particular functional area to expand your knowledge.
Recorded content is searchable by Capability Model attribute and level in the CAS Online Library.
Introduction to Increased Limit Factors
This session will present an overview of increased limits ratemaking. Participants will cover general concepts, such as calculating limited average severities, and practical problems with developing increased limit factors (ILFs) from a distribution of loss data. The session will also provide an overview of excess and deductible pricing and will discuss common approaches for calculating ILFs.
Source:
2012 Ratemaking and Product Management Seminar
Type:
workshop
Moderators:
Jed Isaman
Panelists:
Jared Smollik
Ratemaking Relativities
This session will examine the various methods that actuaries use to allocate overall average rates to various subdivisions of a line of business, including territories, classifications, and tiers. Some of the methods discussed will consist of univariate, multivariate, and generalized linear modeling techniques.
Source:
2012 Ratemaking and Product Management Seminar
Type:
workshop
Moderators:
Jed Isaman
Panelists:
Christopher Cooksey
Overall Rate Level Indication Considerations
Covering the basic foundations of the ratemaking process, topics for this session will include data organization for premium and losses, data adjustments such as current rate level, loss development and trend, and the determination of the expense and profit provisions.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Jed Isaman
Panelists:
Mark Komiskey
Just How Credible Is That Employer?
Come learn about the latest developments in Individual Risk Rating. NCCI has recently completed an experience rating plan review and is currently doing a review of its workers compensation excess loss factor methodology. This session will provide some upcoming changes for improving the performance of the Experience Rating Plan. Estimated impacts will be provided including some hypothetical examples.
For excess loss factors, multilevel modeling and GLMs provide viable solutions to the challenge of estimating loss severity across states, hazard groups, and claim groups. As statistical noise can cause indications to swing violently when estimating severities at such a granular level, the major concern is one of stability. A GLM introduces a statistical framework and the use of multilevel modeling produces endogenously credibility-weighted estimates. The approach proves advantageous over the one historically employed.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Jed Isaman
Panelists:
Tom Daley, Chris Laws
A Predictive Diagnosis of Workers Compensation Claims Costs
Two consultants will discuss their experiences with workers compensation claims predictive models that were designed to reduce costs-with one of the presentations focusing in detail on medical cost drivers. Both presentations will outline the goals and objectives of the projects, the modeling processes and outputs, and the current and potential benefits derived from leveraging workers compensation claims predictive modeling.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Richard Lauria
Panelists:
Gary Ciardiello, Stephen DiCenso
Current Issues in Workers Compensation
Loss Cost Components and Industrial Structure: The industrial structure of the U.S. economy has experienced significant changes during and in the wake of the 2007-2009 recession. While employment in health care has kept rising at a steady rate, employment in manufacturing and construction has shrunk markedly. These changes in industrial structure affect the loss cost components in workers compensation aggregate ratemaking. The influence of the change in industrial structure on loss cost components is quantified using a statistical model that processes annual state-level aggregate ratemaking data.
Obesity and Time to Return to Work: The effect of obesity on the time to return to work is quantified using a large data set of workers compensation claims. Both a Kaplan-Meier estimator and a proportional hazard model are employed; both approaches control for interval-censoring (due to the waiting period) and right-censoring. The time to return to work for obese claimants is a multiple of the reading for comparable non-obese claimants. These findings agree with a 2007 study by Duke University.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
Frank Schmid, Chris Laws
Workers Compensation-Selected State Issues
Issues arising and concepts applied in preparing and defending recent pricing filings in the states of New York, Pennsylvania, and Delaware will be presented, explained and/or compared. Subjects addressed will include loss development, claim frequency experience and trending, severity trends, and recognizing the effects of law amendments, including the 2007 New York reforms, in preparing pricing filings. Detail from recent state filings will be used to support the speakers’ presentations.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
Timothy Wisecarver, Martin Heagen
Workers Compensation Ratemaking-An Overview
The panel will review the essential components of a typical rate filing from the perspective of NCCI, other bureaus, and from the view of companies in loss cost jurisdictions. The discussion will highlight coverages, exposure bases, and data sources used for workers compensation ratemaking.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
Jay Rosen, Daniel Perry
Risk Load/Cost of Capital: How Reinsurers Consider these in Reinsurance Rates for Property Cat Covers
The panel will discuss how risk load/ and/or the cost of capital is considered in reinsurance rates for property cat covers. This will be done from the perspective of a reinsurer and also from the perspective of a reinsurance broker. This session should be of interest to attendees that have direct insurance on property exposed to the potential of significant losses from catastrophe events.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
Gary Venter, Ira Robbin, Ronald Wilkins
Risk and Return Considerations in Ratemaking: Calculating the Profit Provision
After you have the projected loss costs and expenses, the final step in deriving the indicated premium is to load in the underwriting profit provision. But what is the right number? This session will supply not one, but several, answers to that question. It will survey different approaches, from those mandated by regulators to those used by corporate pricing actuaries for internal profitability analysis. The assumptions and parameter selections for each method will be discussed and the sensitivity of results to key parameters will also be explored. The session will have a practical focus with an emphasis on clarifying basic concepts and highlighting key distinctions between different methods.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
What Makes a Good Rate Filing?
What makes a good rate filing? It depends on whom you ask. Now is your opportunity to hear from individuals who have extensive experience with the filing process at companies, bureaus, and insurance departments.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
Ken Creighton, Deborah Rosenberg, Carl Sornson
State vs. Federal Regulation
President Barak Obama signed the Dodd-Frank Act on July 22, 2010, and pursuant to that law, the Federal Insurance Office (FIO) has begun its work. The debate about the role of federal regulation continues and intensifies. There seem to be strong arguments by several insurance companies-some in favor of federal regulation, others very much opposed. This session will examine the current state of federal regulation and present two differing viewpoints about the future. In the spirit of a formal debate each side will be allowed to present their case, and then later will be given an opportunity to respond to the other side. Finally, the session will conclude with questions from the audience.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
Edward Collins, Jimi Grande
Ask a Regulator
Each panelist will give a brief overview of current issues in the regulatory arena, followed by a roundtable discussion of the following topics, as well as any others raised by audience members.
Best practices for rate and product submissions
Incorporating cat models, predictive modeling, and other cutting-edge analysis into rate filings
Regulatory standards for evaluating new products
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
Anne Kelly, Ken Creighton, Howard Eagelfeld, Carl Sornson
Avoiding Litigation and Disciplinary Risk
As a practicing actuary you may face the risk of professional litigation or disciplinary risk in pursuing your profession. Learn how to reduce these risks for yourself and your company, from a panel of experts including an attorney who specializes in defending actuaries in professional litigation and a former member of the ABCD. The session is expected to be highly interactive! This is the opportunity to improve your understanding of both the litigation and disciplinary processes, to learn steps that you can take to reduce the probability of your becoming involved in either process, to learn how best to manage the process should you become involved, and to get your questions answered.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
Michael Toothman, Ronald Lepinskas
Customer Lifetime Value - Opportunities and Challenges
Customer lifetime value (CLV) is a useful tool in marketing and customer relationship management (CRM). CLV has been gaining ground in the insurance industry over the last several years. Despite the theoretical simplicity of CLV, it is fraught with difficulty when applied in practice. In this talk, we will discuss critical issues to consider when modeling and implementing CLV applications within the insurance industry
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
Mohamad Hindawi, Gregory Firestone
Product Managers & Actuaries: A Comparison of Professions
Over the past few years, more and more Commercial Lines insurance companies have been hiring product managers. What roles do these product managers serve, and how do they interact with Commercial Lines actuaries? What makes a successful commercial actuary? A successful commercial product manager? We'll discuss these questions and more with a panel of individuals who have been both Commercial Lines product managers and actuaries.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Alessandrea Quane
Panelists:
George Busche, Donna Glenn
Large Scale Analysis of Renewal Discounts for P&C Insurance
In this session, we discuss the issue of whether a price discount for renewal business is warranted for property and casualty insurance. The discount is motivated by the fact that new business with insurance coverage lapse, or new business in general, may perform worse than renewal business. We will support the discussion with a large amount of real industry data: 25 books of insurance business with a total amount of almost $29 billion of premium. The data cover all of the primary property and casualty lines of business, including personal auto and homeowners as well as commercial business owner's policies, auto, WC, GL, and property. Our discussion will show that new business universally has a higher loss ratio and a lower retention rate than renewal business across all the 25 books of business. We will attempt to offer reasons as to why such differences exist between new and renewal business for insurance.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Kathryn Hyland
Panelists:
Cheng-Sheng Wu, Hua Lin
Machine Learning: A Complementary Approach for Product Management and Book of Business Segmentation
Learn how machine learning can help you validate and refine your best and worst profiles in your book of business and how you can use this knowledge to manage your book more profitably.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Kathryn Hyland
Panelists:
Stan Smith
Case Studies in Adding Variable Interactions to GLMs
Model development using Generalized Linear Models (GLMs) has traditionally focused on the search for and inclusion of main effects variables. More recently interactions between variables have become the aim of many practitioners. Two main approaches have dominated. Some have looked for two variable interactions by quasi exhaustive search of pairs of variables. Others have used tree growing algorithms, such as CART, to build compounds of two variables or more. Both methods produce compound variables that are understandable or insightful. This talk explores how modern methods can produce highly compound variables, of large numbers of variables, that are not represented as trees but continuous scores. Does the real world involve large interactions between a great number of variables? Their combined power to represent real insurance signals as well as their impact on GLM based model development is explored using a number of real P&C data sets.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Panelists:
Paul Beinat, Chun Li
The How, Why, and When of PRIDIT: Examples from Hospital Quality and Fraud Detection
PRIDIT, or Principal Component Analysis of RIDIT Scores, is a relatively new and versatile technique for producing a rank-ordered score for the intensity of a latent variable, by identifying a monotone relationship between this latent variable and a set of ordinal predictor variables. PRIDIT has previously been used to explore fraud, hospital quality, and risk management. Many actuaries involved or who wish to be involved in predictive modeling may wonder how PRIDIT works, why they should use it, and when it is an appropriate model to fit available data. This talk will introduce a general description of PRIDIT that was developed with input from practicing actuaries. It will also present two applications to hospital quality and fraud detection.
This actuarial introduction to the method will then be used to show how it is applied to the problem of determining hospital quality. The motivation for the determination of hospital quality comes from health actuaries--the Society of Actuaries Health Section provided grant funding for the work that the speaker will describe. The example will be generic, so that anyone who has personal experience with a hospital should be able to relate to the problem of determining overall hospital quality from many measures of hospital performance. The goal is that attendees will get an understanding of the method with hospital quality that will allow them to apply PRIDIT to their line of business.
A more detailed application will be presented to fraud detection. The attendees will be able to see how the PRIDIT method is used to analyze this realistic and important problem in the P&C insurance industry, with two data sets from the practice. Also will be discussed in this context is a fraud rate estimation method designed based on PRIDIT that can be used to monitor fraud activities within the company and assess the effectiveness of any detection efforts.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Panelists:
Robert Lieberthal, Jing Ai
GLM II
GLM I provided the case for using GLMs and some basic GLM theory. GLM II will be a practical session outlining basic modeling strategy. The discussion will cover topics such as overall modeling strategy, selecting an appropriate error structure and link function, simplifying the GLM (i.e., excluding variables, grouping levels, and fitting curves), complicating the GLM (i.e., adding interactions), and validating the final model. The session will discuss diagnostics that help test the selections made.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Jin Li
Panelists:
Ernesto Schirmacher, Lenard Llaguno
GLM I
Do terms such as "link function," "exponential family," and "deviance" leave you puzzled? If so, this session will clarify those terms and demystify generalized linear models (GLMs). The session will provide a basic introduction to linear models and GLMs. Targeted at those who have modest experience with statistics or modeling, the session will start with a brief review of traditional linear models, particularly regression, which has been taught and widely applied for decades. Session leaders will explain how GLMs naturally arise as some of the restrictive assumptions of linear regression are relaxed. GLMs can model a wide range of phenomena, including frequencies and severities as well as the probability that a claim is fraudulent or abusive, to name just a few. The session will emphasize intuition and insight in addition to mathematical calculations. Illustrations will be presented using actual insurance data.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Jin Li
Panelists:
Tanya Havlicek, Ernesto Schirmacher
Model Validation - Know Your Audience
The explosion of predictive modeling techniques during the past decade has led to more sophisticated pricing throughout the insurance industry. As a predictive modeler builds his or her model, the fundamental question arises, "How do I know this works?" Various statistical techniques are employed to resolve any doubts, and the results are then presented to senior management. Senior management then asks the identical question, "How do I know this works?" The response to this question is not the same because the modeler must provide a concise, simplified response at a higher level of detail that also incorporates business considerations. Finally, the results frequently find their way into a rate filing where the regulator asks the question, "How do I know this works?" Again, the modeler must support a different answer that takes into account proprietary considerations, answers concerns about excessive, inadequate, or unfairly discriminatory rates, and also allows a regulator to compare insurance filings across companies.
This session will examine model validation from three perspectives - modeling, management, and regulatory. Techniques to be discussed will include data partitioning, tests of statistical significance (confidence intervals, p-values, Type III tests, etc.), cross-validation, bootstrapping, and lift charts. The goal of this session will be to provide attendees with the tools necessary to answer the question, "How do I know this works?" for three distinct audiences.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Jin Li
Panelists:
Larry Haefner, Richard Piazza, Amber Popovitch
Variable Selection Using Elastic Net
The availability of data and the low cost of storage have created an explosion of the data available for predictive modeling. This created an opportunity and a challenge for the modelers who are faced with thousands of variables to choose from. Variable selection methods are powerful tools in analysis of such high dimensional massive data.
Penalization regression is a set of regression techniques that impose constraint/penalty on the regression coefficients. One of these techniques, which is well known, is Least Absolute Shrinkage and Selection Operator (LASSO), introduced by Tibshirani in 1996. Elastic Net is a relatively new technique that generalizes LASSO. It was introduced by Zou and Hastie in 2005. Elastic Net is very powerful and has many advantages over other variable selection techniques. It encourages a grouping effect, where strongly correlated predictors tend to be in or out of the model together.
In this presentation we will give an overview of Elastic Net and its use as a variable selection tool. We will discuss some of its advantages over other variable selection techniques.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Jed Isama
Panelists:
Mohamad Hindawi
Territorial Ratemaking
Territorial ratemaking (and highly dimensional predictors in general) has been an area of active actuarial research lately. Methods currently employed range from the traditional ones, such as one-way relativity analysis and GLMs, to the more modern ones, such as spatial smoothing and clustering. In essence, the newer approaches try to incorporate some domain knowledge in solving the problem, such as spatial adjacency or other similarity measures. In this session, panelists will compare and contrast these methods with some machine learning approaches based on a version of rule induction.
Source:
2012 Ratemaking and Product Management Seminar
Type:
concurrent
Moderators:
Mark Scanlon
Panelists:
Gary Wang, Eliade Micu
