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.
Enterprise Risk Management and the Actuarial Profession
The 2004 ERM Symposium was the first ERM-related seminar sponsored by the CAS that did NOT have a session called "ERM - Fad or Fact?" or something along those lines. While it is fairly clear that ERM has staying power, the implications for the actuarial profession are less certain. This panel will explore the opportunities and challenges ERM presents our profession, and provide some insight into the activities inside and outside the actuarial profession that will profoundly influence the role of the actuary in the ERM process.
Source:
2004 Annual Meeting
Type:
concurrent
Moderators:
Rita Zona
Panelists:
Donald Mango, Mark Verheyen
Keywords:
enterprise risk management, risk profession, derivatives risk management
Emerging Issues in Medical Malpractice
Much has been said of the medical malpractice insurance crisis-causes, insurance failures, tort reform, and the ever-shrinking market that has resulted. This session will put a more futuristic spin on the topic and address a few areas where actuaries and company executives are moving forward with solutions to some of the problems being faced. Nontraditional risk financing alternatives are becoming so common in the medical malpractice arena, we can no longer refer to them as nontraditional. As the number of risk retention groups has increased in recent years, the demand for feasibility studies, pricing, reserving, and captive management has created a growing practice area for actuaries. The actuary's role in this emerging field will be presented.
The panel will also discuss the emerging use of predictive modeling for medical malpractice underwriting and risk segmentation optimization. While predictive modeling is heavily used for personal lines and standard commercial lines products, there are still unpaved roads to be explored among commercial specialty lines. The challenges of using these advanced pricing tools for physician owned companies will be presented.
Source:
2004 Annual Meeting
Type:
concurrent
Moderators:
Rita Zona
Panelists:
Kevin Bingham, Richard Lord
Keywords:
medical malpractice, Emerging Issues
Copulas and Common-Shock Models - an Update
This session will focus on the different approaches that two popular models take to describe correlations between two random variables in insurance applications. Panelists will discuss the underpinning of each model type and describe under what circumstances it would be appropriate to employ each model.
Source:
2004 Annual Meeting
Type:
concurrent
Panelists:
Glenn Meyers, Gary Venter
Keywords:
Copulas and Common-Shock Models
Catastrophe Life and Personal Accident
Reinsurance capacity for Catastrophe Life and Personal Accident (PA) treaties was plentiful and available at very good prices prior to the September 11 attacks. Since then the market went through a complete and irreversible transformation in terms of price and availability. The changed market has witnessed an increasing reliance on property/casualty actuarial skills and concepts to underwrite, price and manage accumulation risk of life and PA insurers.
During this presentation, we will discuss the underlying reasons for this emergence and will present applications of P&C models and tools relating to terrorism and earthquakes, two of the main accumulation risks faced by Life and PA insurers.
Source:
2004 Annual Meeting
Type:
concurrent
Moderators:
Rita Zona
Panelists:
David Lalonde, Pierre-Yves Corre
Keywords:
Reinsurance, Catastrophe Life and Personal Accident, terrorism and earthquakes
Captives
Captives are an alternate type of self-insurance vehicle that affords businesses some flexibility with their risk management strategies. One of the principle reasons why captives are formed is to address concerns over the affordability and availability of insurance.
Over the course of the past several years, the use of captives has increased. For many mid- to large-sized companies insurance prices have continued to increase. As a result, the benefits of forming a captive have far out-weighed the benefits of choosing an alternative insurance program, such as a large deductible program. Insurance availability also continues to play a role in the increased use of captives.
Further, places of domicile for a captive are no longer primarily offshore in Bermuda or in Vermont. Other states are now offering opportunities for forming captives.
This session will address the reasons why a company would form a captive, the recent growth in the formation of captives and key issues surrounding forming and maintaining a captive.
Source:
2004 Annual Meeting
Type:
concurrent
Moderators:
Avivya Stohl
Panelists:
Charles Woodman, Andrew Sargeant
Keywords:
Captives
California Workers Compensation
Do the math: AB 749 + (AB 227 / SB 228) + SB 899 = X. Panelists will discuss the latest estimates of the financial impact to be expected from each of these newly enacted bills. While a summary of the "bottom line" will be provided in a general session, this session will break down the components into the expected reduction to medical, indemnity, vocational rehab, and other expenses.
Need help explaining this mess to your company executives? The panelists will provide suggestions on how to translate the double-digit reductions boasted by Governor Schwarzenegger into the single-digit rate reductions recommended by the WCIRB, then to the approved rate change, and eventually getting to the renewal rate change your company will need to sell.
After having recognized the legislative changes in your pricing, you will likely be making parallel adjustments in your reserve analyses. The panel will discuss what methodologies are being used to incorporate these benefit level changes into ultimate loss projections, and the extent to which auditors are "buying into" these anticipated savings. The panelists will also discuss considerations for tailoring the estimates to your own company's book. In a state marred by deteriorating workers compensation reserve positions, we will get a regulator's perspective on an explicit provision in the reserve analysis for "anticipated expense reductions."
Source:
2004 Annual Meeting
Type:
concurrent
Moderators:
Patrick Crowe
Panelists:
Ronald Dahlquist, Joanne Ottone
Keywords:
California workers compensation,Reform,Legislation
Adjusting For Prior Experience Using Credibility Theory
This session will explore how actuaries in different operations adjust for prior experience when setting valuation assumptions, new rate assumptions or renewal rates for specific policies. Both theory and actual application will be explored. The speakers will introduce some elementary theory, including Buhlman credibility, ridge regression, cross validation along with examples from classification and ratemaking for personal home/auto insurance. They will discuss methods in use with Workers Compensation Boards across North America. They will also discuss applications related to Group Life and Long Term Disability Insurance.
Source:
2004 Annual Meeting
Type:
concurrent
Moderators:
Patrick Crowe
Panelists:
Christopher Monsour, Gary Walters, Robert Hinrichs
Keywords:
Buhlman credibility, ridge regression, cross validation along with examples from classification and ratemaking for personal home/auto insurance, Workers Compensation Boards across North America
Actuaries in Nontraditional Roles
Actuaries are continuing to take on positions in nontraditional areas of practice, including those related to risk management, financial management, underwriting, claims, and public policy.
The panelists will discuss their nontraditional positions, including their day-to-day activities and the obstacles they face being actuaries in nonactuarial roles.
Source:
2004 Annual Meeting
Type:
concurrent
Moderators:
Bernard Horovitz
Panelists:
Stephan Christiansen, John Ferrara, Duncan McCallum
Keywords:
risk management, financial management, underwriting, claims, and public policy
AAA Updated Research on the NAIC Risk Based Capital (RBC) Formula
The NAIC Risk-Based Capital (RBC) formula implemented in the early 1990s has been deemed an improvement over prior minimum capital standards. The RBC formula provides a ranking process for property casualty insurers with regards to regulatory intervention. But, in recent years, the formula has been criticized because the process does not detect troubled companies "early" enough, despite the fact that the formula was not intended to be used as an "end-all" solution to capital adequacy standards. The American Academy of Actuaries P&C Risk based Capital Committee have been working to test the feasibility of enhancements to the formula, including a "trend test" solution, similar to that used for life insurers. In essence, this effort is an attempt to evaluate and possibly improve upon the early warning responsiveness of the current RBC formula with regard to company impairment. The panel will constitute representatives from the AAA risk based capital committee subgroup charged with developing corresponding recommendations to the NAIC. The panel discussion will summarize the research to date, including feedback from the NAIC in its deliberations with regards to the committee's recommendations.
Source:
2004 Annual Meeting
Type:
concurrent
Moderators:
Bernard Horovitz
Panelists:
Chris Nyce, Anthony Phillips
Keywords:
NAIC Risk-Based Capital, P&C Risk based Capital Committee
Actuaries and Actuarial Issues Around the World
The actuarial profession is facing numerous challenges around the world. As the tools of risk assessment and risk analysis have become widely available to more people and the process of risk management has moved beyond the traditional field of insurance, our various publics' perceptions of what services the actuary provides and the value of those services are changing.
What changes must we make to adapt to the new public perception of a risk management professional?
What must we do to continue to demonstrate our integrity and competence to our current publics and gain the confidence of potential new publics?
What are the hard choices that we must make? Do we need to change our education, standards of practice development, and discipline in order to ensure continuing public confidence?
How do these questions also address the survival of the actuarial profession?
A distinguished panel of leaders of international actuarial organizations will present their views on these questions. They will speak from the experiences in their own countries and from a perspective of the profession globally.
Source:
2004 Annual Meeting
Type:
general
Moderators:
Bernard Horovitz
Panelists:
Robert Wilcox, Harvie Brown, Graham Rogers
Reserve Adequacy and the Underwriting Cycle
A question that is commonly asked is whether there is a correlation between financial reserve adequacy and the underwriting cycle. The panelists will provide a brief summary of their interpretations of this issue, and comment on trends and recent and pending legislation that may affect this issue in the future.
Some of the potential topics of discussion are the impact of Sarbanes-Oxley, revised actuarial reserve opinion rules, and insurance regulatory issues. Transparencies between reserving and pricing/underwriting functions within companies will also be examined. Panelists will also discuss how this issue affects outside rating agencies view of both individual companies and the insurance market in general.
Source:
2004 Annual Meeting
Type:
general
Moderators:
Bernard Horovitz
Panelists:
Michael Angelina, Meyer Shields
Reinsurance Worldwide
When many of us started in the actuarial profession, the traditional view of reinsurers was presented in exam texts as "following the fortunes" of the primary customers. Through liability crises, natural catastrophes, terrorist attacks, asbestos and environmental claims, and the like, that early characterization of reinsurance no longer seems adequate. In many respects, it now appears that reinsurance coverage availability and pricing considerations often are significant elements in marketplace developments for primary insurers, for self-insureds, and for the insuring public. Reinsurance is a worldwide business.
This panel will address the current drivers of the reinsurance market, ever more sophisticated catastrophe modeling techniques, financial and underwriting risk analytics, and emerging liability theories. Panelists will represent the perspectives from various reinsurance "clusters" in North America, Bermuda, and Europe.
Source:
2004 Annual Meeting
Type:
general
Moderators:
Guy Avagliano
Panelists:
Michael Wacek, Donald Alexander, James Standard
Other Modeling Techniques
Exploring a variety of useful techniques not covered in other sessions, this session will pay particular attention to tree-based modeling, as exemplified by the "Classification and Regression Tree" (CART) algorithm. CART is both intuitive and easy to use in any number of settings. While it is often discussed as a predictive modeling tool, it is also useful as an exploratory data analysis, data visualization, and model analysis tool. Time permitting, brief heuristic discussions of such modeling techniques as naive Bayes, kernel smoothing, Neural Networks, and MARS (Multiple Additive Regression Splines) will also be offered. Finally, the "Expectation Maximization" (EM) technique for imputing missing data will be discussed.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
Guy Avagliano
Panelists:
Louise Francis, James Guszcza
Keywords:
CART, MARS, neural net, decision tree, GLM, gains chart, lift curve, model comparison, spam data
Discrete Predictive Modeling
Most actuaries are comfortable with various statistical techniques for making predictions of the value of a continuous variable. However, there are many problems in insurance in which one needs to make predictions about a discrete variable, for example, whether a customer will stay or leave or whether a claim is fraudulent. Such problems are called "classification" problems. This session will introduce the specialized language used to discuss such problems and the basic statistical techniques for analyzing classification problems. The fundamental techniques (naive Bayes, nearest neighbors, discriminant analysis, logistic regression, and classification trees) will be presented, along with appropriate measures of goodness-of-fit for classification problems. In particular, session panelists will discuss how goodness-of-fit criteria can be chosen to link directly to the business purpose of the model. Finally, panelists will introduce some basic models of similarity, where the definitions of the classes are not known in the data being analyzed, as would often be the case in a fraud study.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
Guy Avagliano
Panelists:
Christopher Monsour
Generic Model Validation and Diagnostic Techniques
Modern statistics has greatly enriched the actuary's ability to select the optimal predictive model and accurately estimate a model's predictive power. This session will discuss in some detail the various forms of "cross-validation." Among cross validation's many helpful properties is the fact that it can be applied regardless of the modeling technique used and therefore can be used to compare models produced by entirely different techniques. Furthermore, cross-validation enables one to measure a model's strength in terms of "lift curves," which have more operational and business meaning than such standard measures as adjusted R2 and deviance. In addition to cross validation, the generic estimation technique known as "bootstrapping" will be discussed. Bootstrapping is an intuitive brute-force technique for estimating confidence intervals that has natural applications both in insurance predictive modeling and loss reserving.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
Richard Bill
Panelists:
James Guszcza
Keywords:
model validation, lift curve, cross-validation, gains chart, bias variance tradeoff, model selection, bootstrapping, variance
Generalized Linear Models I
This basic introduction to generalized linear models will begin with a basic review of common statistical methods that actuaries and other analysts have used for many years. No familiarity with modeling methods will be assumed. Topics will include introduction to regression and ANOVA. Methods of performing goodness-of-fit and diagnostic tests to examine the underlying assumptions of the regression model will be presented. The session will then discuss the often-encountered problem of what to do when the model assumptions are violated, which will lead to an introduction to GLMs. Where possible, the session will illustrate the statistical principles with applications that can be performed in Excel.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
Richard Bill
Panelists:
Louise Francis
Predictive Modeling for a Commercial Insurance Company with Little or No Data
Predictive modeling for a start-up company or company with little or no data of its own might seem like a non-starter. However, data and off-the-shelf models are available from various third-party vendors. This session will explore the data and techniques available to help a company get off the ground with predictive modeling.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
Richard Bill
Panelists:
John Wilson, Scott Bronstein, Tom Gershbach, A. Phinney
Fraud-Fighting Actuaries: Mathematical Models for Insurance Fraud Detection
This session will give an overview of the insurance claim fraud problem, emphasizing claims processing and fraudulent and abusive claims detection. Most estimates of the extent of the problem involve large aggregate dollars in losses and loss adjustment expense. Criminal fraud is distinguished from systematic abuse by exaggerated losses and calls for different detection and handling strategies. Recent research results using Massachusetts auto injury data and models for claim severity, claim classification, and the liability settlement process will be covered. An expanded role for the company actuary will be discussed.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
Richard Hofmann
Panelists:
Richard Derrig
Predictive Modeling for Homeowners
Many ratemaking applications of GLM have focused on auto examples. This session will discuss some rating elements that are of particular concern for homeowners insurance, like deductibles and amount of insurance. Panelists will also present research results utilizing detailed property characteristics as predictors of claim frequency and severity. This use of predictive modeling has potential application in risk assessment and pricing.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
Richard Hofmann
Panelists:
A. Cummings, Deanna Nix, David Bragg
Keywords:
deductible, GLM, truncation, dispersion
Predictive Modeling for Personal Automobile Pricing
The use of generalized linear models (GLMs) has been increasing rapidly in the U.S. marketplace. Because of the large number of rating factors, data availability, and the competitiveness of the market, personal lines automobile insurance seems to be leading the way with respect to the use of these techniques.
The panel of practitioners will discuss the use of GLMs for personal lines auto pricing with specific emphasis on class plan and other relativity analysis, insurance/underwriting scoring, vehicle classification, territory boundary development, and profitability optimization.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
G. Nyce
Panelists:
Roosevelt Mosley, Geoffrey Werner
Alternative Data Sources for Modeling
Predictive models developed are only as good as the data used to develop them. Insurers have begun implementing predictive models using internal pricing data, but are also looking to develop better models with more data items. This session will discuss alternative sources including demographic data, vehicle information, fire station, credit information, and vehicle crash test information. We will discuss what information is available, how the information is collected, and how to append the data to an insurer's internal information.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
G. Nyce
Panelists:
Roosevelt Mosley, John Wilson
Dimension Reduction
As a profession we would like to understand all of the underlying drivers of losses, however, it is not always possible or even practical to include all possible variables in the final rating plan. A thorough analysis can use different techniques to consider all of the different variables and should provide the user with statistics and outputs as to which structure is best to use. This session will discuss different approaches and reasons to reduce or group the number of variables. It will also talk about pros and cons of automated approaches, like stepwise regression, as well as the impact of forced dimension reduction (regulatory constraints). Ways to compare different models and the effectiveness of the new structure will also be discussed.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
G. Nyce
Panelists:
Thomas Hettinger, Serhat Guven
Keywords:
Dimension Reduction; Multivariate Analysis
Credit Scoring-An Insurance Department Perspective
The use of personal credit information in underwriting homeowners and personal automobile insurance is a significant topic of concern for regulators. Numerous insurance industry studies have demonstrated a strong correlation between an individual's "financial responsibility" as measured by a credit score and the expected loss. These studies claim to have shown that the use of credit scores in underwriting unfairly harms low income and minority consumers, and are therefore unfairly discriminatory. Other studies have been presented to demonstrate that credit scores are valid predictors for all classes of insureds and conclude that the use of credit scores for underwriting does not discriminate unfairly.
Other unresolved issues continue to present a major concern for insurance regulators. Some of these are directly related to the use of multivariate underwriting models and the industry's reluctance to open the "black box." Others relate to improper application of credit information and insurance company disclosure when credit information is being used. Insurance departments are unable to satisfy consumer complaints when premiums increase from deterioration in a policyholder's credit score, especially when the policyholder has a clean driving record.
Insurance departments have addressed these concerns in a number of ways. Some state regulators would prefer to ban the use of credit information, while others have accepted this information as a valuable underwriting tool and have developed rules to mitigate the consumer concerns. This panel will discuss the use of credit scoring models from a regulator's perspective.
Source:
2004 Fall SIS- Predictive Modeling
Type:
general
Moderators:
Frank Zizzamia
Panelists:
Michael Miller, Dudley Ewen, Brent Kabler
Modeling Lessons From Across the Pond- Insights from Decades of Deregulated Modeling
Generalized linear models have been applied in personal lines ratemaking and other insurance applications for over a decade in the U.K. and much of Europe. These less-regulated environments have provided fertile ground for actuaries to hone this craft. This session is designed to share modeling techniques and developments that would be of interest to North American actuaries. The session will begin with a discussion on the use of GLMs in rating plan design and common rating factors. The session will then focus on some recent regulatory developments in the European Union, particularly the increased interest in protecting consumer rights and the need for insurers to justify the degree to which they can reasonably segment their prices.
Finally, the session will cover advances in modeling techniques, including the use of external data in European personal lines ratemaking, spatial smoothing techniques in personal lines territorial analyses, and price elasticity modeling and price optimization techniques in deregulated markets.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
Donald Rainey
Panelists:
Michael Brockman, Duncan Anderson
Keywords:
Duncan Anderson, GLM, Generalised Linear Model, Generalized Linear Model, Price optimisation, Price optimization, Spatial analysis, Spatial smoothing, External data, Geodemographic data, Bonus Malus, Retention modeling
Generalized Linear Models II: Applying GLMs in Practice
This session will discuss how the statistical concepts introduced in Generalized Linear Models I are used in practice. After a brief summary of model formularization, this session will review model diagnostics that aid in both variable selection and the choice of model form. Typical model forms (e.g., multiplicative Poisson, Gamma and Tweedie, and logistic) and their output will be presented as well as the practical consequences of choosing between different model forms. Modeling pitfalls (e.g., near-aliasing) and modeling refinements (e.g., interaction variables, artificial restrictions, smoothing and combining models across claim types) will be discussed. Finally, the session will discuss GLM applications, including ratemaking, retention analysis, and profitability analysis.
Source:
2004 Fall SIS- Predictive Modeling
Type:
concurrent
Moderators:
Donald Rainey
Panelists:
Duncan Anderson
Keywords:
Duncan Anderson, GLM, Generalised Linear Model, Generalized Linear Model, Generalised Linear Models, Generalized Linear Models, Aliasing, Interactions, Tweedie, Statistical premium rating, Statistical ratemaking
