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.
Practical Application of Development Trend Extrapolation for Triangle GLMs
Triangle-based stochastic reserving models with payment period trends necessitate extrapolation of future trends. Using illustrative examples, we explore practical aspects of a mathematical extrapolation scheme recently proposed in (Hartl, 2015). We show how the method extends triangle-based analyses that ignore payment period effects, discuss to what extent the model captures uncertainty in future payment period trends, and explain how to incorporate exogenous assumptions about future payment period trends. Finally we demonstrate that the same method can be used to extend triangles beyond the last development period, and suggest a strategy to avoid leveraging extreme uncertainty in the tail.
Source:
2016 Casualty Loss Reserve Seminar (CLRS)
Type:
Concurrent Session
Moderators:
Bob Wolf
Panelists:
David Moore, Thomas Harti
Individual Claims Reserving with Stan
In 2006, Guszcsa and Lommele proposed a reserving technique which fit a Poisson to individual claims. When aggregate data are split, their individual model will devolve to an aggregate chain ladder approach. This may be feasible if the volume of segmented data which results is sufficiently large. However, as the data get more refined, the variability around estimated model parameters increases. This session will explore the use of Bayesian techniques and the Stan modeling language to use all of the data to enhance the estimation procedure for each segment.
Source:
2016 Casualty Loss Reserve Seminar (CLRS)
Type:
Concurrent Session
Moderators:
Bob Wolf
Panelists:
Brian Fannin
Improving Actuarial Reserve Analysis Through Claim-Level Predictive Analytics
Aggregate reserving techniques such as triangle analysis dominate current actuarial practice. However, there is much to be gained by analyzing the detail that is otherwise lost when triangles are created. Advances in predictive modeling are allowing companies to detect patterns in data which were not able to be detected in the past. Chris Gross will illustrate some of the approaches and the benefits of using predictive modeling techniques in actuarial reserving. His work ranges from adding additional insight to traditional reserving techniques by analyzing changes in case reserve adequacy or assisting with reserve segmentation, to a full reserve development model based on individual claim development and emergence behavior over the life of a claim. Brian Janitschke will discuss the experience at his company of implementing such a model at an insurance company in his role as Chief Actuary.
Source:
2016 Casualty Loss Reserve Seminar (CLRS)
Type:
Concurrent Session
Moderators:
Bob Wolf
Panelists:
Scott Anderson, Christopher Gross, Brian Janitschke
Creating an Informative Prior
Bayesian models not only allow a user to incorporate prior knowledge, they actually require the user to specify this prior knowledge. This session will investigate methods for taking benchmark patterns and prior beliefs about development patterns into a model by translating that knowledge into explicit distributions.
Source:
2016 Casualty Loss Reserve Seminar (CLRS)
Type:
Concurrent Session
Moderators:
Bob Wolf
Panelists:
Glenn Meyers, David Clark
Beyond the Chain-Ladder Framework: Generalized Linear Models for Reserving
As the level of sophistication of reserving tools increases in recent years, actuaries are gradually supplementing the traditional chain ladder framework with more advanced modeling techniques. Generalized linear modeling (GLM) employing aggregate data expands beyond the chain-ladder framework, which focuses only in the development year dimension, by modeling also trends in both the accident year and calendar year dimensions of a triangle. The session will briefly describe the GLM modeling framework and walk the audience through a simple example of how this is employed specifically within the reserving context. Furthermore an optimization routine that allows the modeler to fit the GLM in all three triangle dimensions simultaneously will be discussed.
The second part of the session will focus on GLM employing claim level detail data. The introduction of additional variables beyond the traditional two-dimensional model using "year" and "lag" enable us to identify patterns and trends that otherwise would be masked in aggregate data. Examples of various predictors that are employed in claim level detail modeling will be provided. In addition the session will describe the modeling structure of various individual claim level models, including the Incremental Paid method, the Taylor McGuire method, the Open claim method and the Frequency/Severity method. Advantages and associated challenges for each of these models will be also discussed.
Source:
2016 Casualty Loss Reserve Seminar (CLRS)
Type:
Concurrent Session
Moderators:
Bob Wolf
Panelists:
David Moore, Emmanuel Bardis
A Machine Learning Framework For Loss Reserving
For 50 to 100 years actuaries have estimated insurance liabilities with triangles, patterns and other assumptions and methods that are staples of modern actuarial analyses. Technology has led to improvements including stochastic methods, finer segmentation and frequent analysis, but we're now on the verge of extraordinary advances that will fundamentally transform actuarial analyses in the years ahead.
We will begin with a discussion of advanced modeling methods to estimate claim level liabilities. We'll review the challenges of claim level analysis and the benefits of claim triage to identify key characteristics early in the analysis of individual claims. The practical implications of Survival models, GLMs and machine learning methods will be reviewed. We will demonstrate how "early warning" insights from advanced analytics can inform and expedite a traditional quarterly reserving process.
The session will then review a first generation application of machine learning to actuarial analysis. Examples based on the CAS Loss Reserve Database will be presented using machine learning applications in R. We will consider the technical aspects of building and optimizing R models using hyperparameter tuning and model ensembling techniques. Out of time performance of machine learning methods will be compared to traditional projections. Finally, we will review considerations to improve transparency and overcome the black box nature of the models.
Source:
2016 Casualty Loss Reserve Seminar (CLRS)
Type:
Concurrent Session
Moderators:
Alessa Quane
Panelists:
Kevin Kuo, Thomas McIntyre, Andrew Golfin, Christopher Moore, Michaela Porter
Variance Papers: Parameter Risk
Paper #1: Estimating the Parameter Risk of a Loss Ratio Distribution—Revisited
by Avraham Adler
When building statistical models to help estimate future results, actuaries need to be aware that not only is there uncertainty inherent in random events (process risk), there is also uncertainty inherent in using a finite sample to parameterize the models (parameter risk). This paper revisits Van Kampen (2003) in replicating its bootstrap method and compares it with measures of parameter uncertainty developed using maximum likelihood estimation and Bayesian MCMC analysis.
Paper #2: A Note on Parameter Risk
by Gary Venter and Rajesh Sahasrabuddhe
Consideration of parameter risk is particularly important for actuarial models of uncertainty. That is because—unlike process risk—parameter risk does not diversify when modeling a large volume of independent exposures. Without consideration of parameter risk, decision makers may be tempted to underwrite higher volumes as a result of the apparent high degree of predictability in the mean outcome. However, the financial impact of parameter error is magnified by volume and doing so could have significant consequences for the firm. In this paper, we present an inventory of uncertainty models associated with various approaches that actuaries use in estimating model parameters.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
Alessa Quane
Panelists:
Gary Venter, Francis Gribbon, Avraham Adler
Research on Copulas
2016 Hachemeister Prize: Anas Abdallah, Jean-Phillipe Boucher, and Hélène Cossette, "Modeling Dependence between Loss Triangles with Hierarchical Archimedean Copulas"
In this paper, there is discussion around different ways to estimate reserves for incurred but unpaid losses. This paper discusses using a parametric approach with multivariate Archimedean copulas and hierarchical Archimedean copulas to model the dependence between lines of business and calendar years when calculating incurred but unpaid losses. The committee believes that this is a practical paper and topic worthy of bringing to the attention of US actuaries who might not consider using Copulas to model a dependence relationship between triangles of different lines of business. The concepts of this paper are applied to a dataset of a major US Property & Casualty insurer, as well as a large Canadian Property & Casualty insurer to illustrate the impact of these techniques. This will be useful reading for many actuaries who are looking to think differently about how to estimate reserves.
Variance: Paul Ferrara, Rahul Parsa, and Bryce Weaver, “A Linear Approximation to Copula Regression”
Recently, Parsa and Klugman (2011) proposed a generalization of ordinary least squares regression, which they called copula regression. Though theoretically appealing, implementation, especially calibration, of copula regression is generally more involved than for generalized linear models. In this paper a linear approximation to copula regression, for which implementation is similar to that for least squares regression, will be introduced. We proceed by investigating the connection between the proposed approximation to copula regression, and copula regression itself. In particular, we develop a set of criteria which ensure a predictable bias in the estimates from the linear approximation to copula regression.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
Bill Fischer
Panelists:
Jean-Philippe Boucher, Helene Cossette, Paul Ferrara, Anas Abdallah
Embedded Predictive Analysis of Misrepresentation Risk in GLM Ratemaking Models
The authors propose to embed the predictive analysis of misrepresentation risk in ratemaking models under the generalized linear modeling (GLM) framework. In particular, we treat binary misrepresentation indicators as latent variables under GLM ratemaking models, for rating factors that are subject to misrepresentation. By specifying a binomial regression model on the misrepresentation probability, a predictive analysis of misrepresentation risk can be embedded in ratemaking models that use regular claim experience data. The method allows for multiple factors that are subject to misrepresentation, while accounting for other correctly measured rating factors. We demonstrate the identifiability of the model through theoretical studies based on observable moments, as well as simulation studies with finite samples. The identifiability ensures valid inference on the parameters of interest, such as the rating relativities and misrepresentation odds ratios. The approach is illustrated by a case study using Medical Expenditure Panel Survey data.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
Ben Zehnwirth
Panelists:
Gary Vadnais, Michelle Xia
The Credibility of the Overall Rate Indication: Making the Theory Work
The square root credibility formula is often used by actuaries, but is a suboptimal predictor. This session presents the tools needed to convert to best estimate credibility for an overall rate indication, both in terms of the formula and advice for computing the key values.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
John Campbell
Panelists:
Joseph Boor, Michael Chen
The Capital Asset Pricing Model: An Insurance Variant
In the present study the authors declare that the CAPM is not actually based on specific distributions of risks but, instead, relies on the assumption of linearity between the asset and market rates-of-return. This observation encompasses not only numerous classical results on the CAPM but also opens up the possibility of using a vast family of joint distributions for which certain variants of the CAPM become suitable for pricing risk components within portfolios of financial and insurance risks. Consequently, the authors introduce a Weighted Insurance Pricing Model (WIPM), and this study is devoted to discussing its various applications, including pricing insurance risks having light- and heavy-tailed distributions, and following a variety of dependence structures of practical relevance.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
David Payne
Panelists:
Arthur Randolph, Edward Furman, Ricardas Zitikis
Data and Cybersecurity: Legal and Regulatory Developments
Recent cybersecurity breaches of insurance and other entities have prompted insurance regulators to devote additional resources to the issue. Regulators are stepping up their evaluations of insurers' cybersecurity measures, and are issuing additional guidance and creating new requirements that insurance entities must comply with. This session will consider the issues associated with some of these recent developments, as well as new cyber-related issues looming on the horizon.
The proposed cybersecurity model law of the National Association of Insurance Commissioners (NAIC) will be discussed, including a deep-dive into how its provisions would impact the insurance industry, and some of the criticisms of the NAIC's approach. Other activity at the state level will also be discussed, including some of the most recent guidance developed by insurance departments. Tips for working with regulators to avoid compliance pitfalls will also be shared. Lastly, the possibility of federal regulation will be considered, with an analysis of the pros and cons of such a scenario.
Source:
2016 Annual Meeting
Type:
General Session
Moderators:
David Payne
Panelists:
Dustin Loeffler, Fred Karlinsky, Lori Nugent
Automated Vehicles: For Whom the Bell Tolls?
It has been estimated that by 2030, 95% of vehicles produced will be fully automated; and that by 2050, 95% of the vehicles on the road will be fully automated and communicating with each other.
• What are the implications for the insurance industry?
• Is it a threat? Opportunity?
• What are the risks we face?
• What is our role?
• What should we be doing?
• When should we act?
Join us for a discussion on automated vehicles and its impact on society and the insurance industry?
Source:
2016 Annual Meeting
Type:
General Session
Moderators:
David Payne
Panelists:
Matthew Carrier, David Cummings, Michael Stienstra, Jonathan Charak
Florida Government Response to Hurricane Risk
Hurricane Andrew hit South Florida like a bomb in August 1992 and changed the way the State and the insurance industry respond to hurricane risks. The 2004-05 hurricanes showed us that this is also a frequency issue, not just severity. Not too many hurricanes since 2004-05 but history and science suggest that it’s just a matter of time before the next big one(s) make landfall.
In this session, three speakers will discuss different aspects of the government response to hurricane risk.
The first speaker will provide an overview and update of the Florida Hurricane Catastrophe Fund.
The second speaker will describe Florida Commission on Loss Projection Methodology process and standards for approving catastrophe models.
The third speaker will talk about Florida residual market trends.
Source:
2016 Annual Meeting
Type:
General Session
Moderators:
David Payne
Panelists:
John Rollins, Benoit Carrier, Anne Bert, Lorilee Medders
The Underwriting Cycle - Is It Economics or Behavior or Both?
What is old is new again. A historical perspective on the underwriting cycle. Is it really different today than it was 10, 20 or 30 years ago?
Many believe that interest rates, abundance of capital, M&A activity and inflation determine where we are in the insurance underwriting cycle. This may not be the case. In order to complete the picture, one must look at underwriting behavior and confidence. In our industry, many are more focused on the competitive pressures rather than having the self-confidence to enforce a disciplined, consistent and rationale underwriting focus despite where we are in the cycle. Do new tools like data analytics help with this confidence?
This panel with focus on the true impact and factors that influence insurance product pricing.
Source:
2016 Annual Meeting
Type:
General Session
Moderators:
David Payne
Panelists:
Kevin Madigan, Eduard Pulkstenis, Greg Hendrick, John Shettle
Workers' Compensation and Prescription Drugs
Prescription drug costs represent a significant portion of workers compensation (WC) medical costs and is one of the most active subjects of WC-related legislative activity. In this session, we will examine:
• The estimated Accident Year 2014 prescription drug share of WC medical costs
• The impact of price and utilization changes on prescription drug costs
• Potential prescription drug cost savings from a drug formulary
• Controlled substances
• Physician dispensing
• Brand name and generic drugs
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
David Payne
Panelists:
Philip Borba, David Colon, Wesley Griffiths
War and Peace without Pride and Prejudice: The Power of Persuasion
How to build relationships through professionalism and consultatively sell with the value add of Emotional Intelligence ...
Persuasiveness is probably the most important skill you’ll ever need in this life time. Imagine where the entire animal kingdom would be today, if it had failed to master persuasiveness. For one to influence another’s actions is an infinite power indeed.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Panelists:
Farnaz Namin
Using Predictive Analytics for Claims Modeling
Many companies that have successfully implemented predictive models in their claims operations are starting to see the positive impact of these analytical tools on both claim cost as well as duration. These companies are typically scoring claims based on the expected cost and/or duration of each claim and are using this information to triage claims. This session will provide some anonymized examples of the successful use of these predictive models and discuss how the change in claim duration and average cost per claim is impacting the payment and reserving practices of these companies and development patterns. The session will also address how to analyze the predictive efficiencies of claims models and how to modify the models over time to increase the predictive power.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
David Payne
Panelists:
Brian Brown, Todd Lehmann, Stan Smith
University Actuarial Programs and the Property and Casualty Industry – CAS University Award Winners Share Their Insights and Best Practices
In June 2016, four universities were selected as recipients of the 2016 CAS University Award, an honor created to recognize schools doing exemplary work in preparing students for a career in the property and casualty insurance industry. The schools honored in the program’s inaugural year include Illinois State University, University of California, Santa Barbara, University of Connecticut, and University of Illinois at Urbana-Champaign. A representative from each university will discuss how they are incorporating property and casualty into their curriculum, research, industry engagement initiatives, and innovations. Attendees will hear best practices on how companies can partner with universities and work together to prepare the next generation of property and casualty actuaries.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
David Payne
Panelists:
Roger Hayne, Krzysztof Ostaszewski, James Trimble
Sustainable ERM Working Party
As global trends —environmental, social, political, technological —continue to shift the foundations of our current business models, sustainability is becoming embedded in corporate behavior, metrics and strategy. This session will show evidence for a global Sustainability meta-trend and summarize business cases and best practices for insurance companies. It will directly serve the risk professionals with more relevant, current, practical and technical information on the new way of doing business and integrating Sustainability into ERM.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
Daniel Perry
Panelists:
Voon Lai, Thomas Wakefield
Surety - US & Latin American Perspectives
This session will introduce Surety reserving and pricing practices within the US and Latin America. Surety is a unique insurance product not commonly understood. With different characteristics, data patterns, and risks; traditional actuarial methodologies must be adapted. Come and explore US and Latin American perspectives on this important third party product.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
John Wade
Panelists:
Alejandro Ortega, Landon Mortensen, Wilson Mayorga
Superforecasting Working Party
The popular book "Superforecasting" describes people who produce unusually accurate forecasts of future event probabilities, the methods they use, and the science of forecasting evaluation. This session will provide an overview of the book, describe recent activities of the CAS Superforecasting Working Party, and provide an opportunity for questions and discussion.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
John Wade
Panelists:
David Ruhm, John Wade, Liza Wong
Solvency II - One Year On! A Comedy of Errors or Much Ado About Nothing?
Yes! It's finally happened. After a decade of build up, Solvency II incepted on 1/1/2016. Our talk will discuss observations from the UK market and Lloyd's - why some insurers had a smooth implementation while others struggled. Why did some insurers experience the trauma of Macbeth while others sailed through smoothly like The Two Gentlemen of Verona?
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
John Wade
Panelists:
Robert Wolf, Kendra Felisky, Jeffery Courchene
The Science behind Time Management and Productivity Hacks
An Insight driven strategy to build on your successes ...
We live in a world of industry and technology. Everyday millions and millions of stimuli enter our periphery and hijack our brains. It would seem that with so many advances in technology, we would get more done. But the truth is most of us don’t. Our brains are not designed to stay on task. They, similar to other organs of the body are meant to be trained to do what we want them to do.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Panelists:
Farnaz Namin
Reserving in Uncertain Economic & Legislative Conditions - Compare and Contrast California WC and Argentina Auto
Reserving during changing economic conditions, such as high inflation, is particularly complicated when there is a lag to pay claims. This session will investigate two markets that have both features - and compare and contrast the environment and methods applied in both markets to estimate unpaid losses. They will show that the standard development factor method does not work under changing economic conditions, and explore methods used in 2 jurisdictions that initially appear to be very different, but in fact have a lot in common.
Source:
2016 Annual Meeting
Type:
Concurrent Session
Moderators:
John Wade
Panelists:
Christopher Gross, Mark Priven, Alejandro Ortega
