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.
R-9: Practical Applications of Credibility Related to Large Account Pricing
Credibility is a simple concept but often difficult to apply in practice. This talk will discuss two applications of applying credibility to large account pricing for incorporating the excess losses and for smoothing changes over time periods.
The typical approach to considering an account’s experience above the basic limit is to analyze the burn costs in the excess layers directly. But burn costs in higher layers are extremely volatile and right skewed, and do not perform well with typical credibility methods. An alternative approach to incorporating this information will be discussed where the excess losses are used to modify the severity curve used to calculate the ILF. Such an approach considers all available information in the same way as analyzing burn costs, but works better for higher, volatile layers. It may also be easier to implement as it does not involve developing and selecting a burn cost for each layer.
The second part of the talk will describe a method for credibility weighting changes in accounts’ losses over time, with applications for profitability studies and pricing models as well. The typical approaches for all three are either ad hoc or insufficient. Often, loss development is confused and mixed with the estimation of changes. But a proper forecast should involve an accurate assessment of possible changes. A method will be shown that incorporates random walks and related state space modeling functionality (which will be explained) within a regression framework that is robust, relatively simple, and that is designed to be suitable for presentation.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Cheng Khang Saw, Peter Lowth
Panelists:
Uri Korn
R-8: Understanding the Changing Used Vehicle Market
The ever changing used vehicle market has given insurers an opportunity to create a new underwriting or pricing variable. Not all used vehicles are in good condition and vehicles in poor conditions have shown to have higher loss costs. By more accurately pricing used vehicles, insurers can save one to two points on their bottom line.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Houston Cheng, Brett Nunes
Panelists:
Christopher Whipple, Daniel Pickens
R-7: Rate Deployment in a Digitally Transforming Insurance Environment
Insurers have the need to deploy rates, rating structures, or models in a matter of hours, not weeks or months. Deploying these items quickly to real-time production environments enables better organizational agility and nimbleness, and in turn a better customer experience. In this session, Earnix along with DataRobot will explore how automating both pricing and non-pricing related models, decisions, workflows, and handoffs through analytically based software improves the ability for insurance companies to respond to market forces. We will also look at how updating and replacing core systems as well as selecting, automating, and deploying machine learning models can be accomplished quickly via modern software systems.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Houston Cheng, Dion Oryzak
Panelists:
Dror Pockard
R-6: Cyber Risk and Insurance: How are the Market and Rating Methods Evolving?
This session will feature a discussion around the relatively recent emergence of cyber risk and corresponding cyber insurance market, challenges presented to insurers by this new and dynamic landscape, and the intersection of cyber security efforts and traditional insurance practices in helping to address insurer uncertainty in writing cyber risk. The session will also examine current progress and lessons learned from the development of a comprehensive rating plan that can used by insurers and reinsurers to quantify cyber risk.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Peter Wu, Houston Cheng
Panelists:
Eduard Alpin, Joshua Pyle
R-5: Workers Compensation—State of the Market
An overview of the current state of the Workers Compensation line will be presented, including a review of financial results, recent trends, and a discussion of where the line might be headed.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Jeffrey Baer, Pamela Sealand Reale
Panelists:
Jay Rosen, Brent Otto
IET-5: The Changing Role of the Actuary in the Face of Disruption (Livestream)
Many jobs across the economy are quickly becoming disrupted as a result of advanced technologies like cognitive automation and artificial intelligence, as well as alternative workforce arrangements. Are actuaries next? This session will introduce the concepts underlying robotics and cognitive automation, and will also explore alternative talent models such as use of off-balance sheet resources. Using a process called Pixelation, this session will explore the many tasks that actuaries perform and what parts of our roles could be automated, augmented or replaced. At the same time, we will discuss how this disruption can have a positive impact on the actuarial role, how actuaries are thriving in this emerging environment by embracing new technologies and an innovation mindset, and how the profession can ultimately provide even more value to our constituents.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Christopher Gross, David Shleifer, Houston Cheng
Panelists:
Rob Galbraith, Stefan Peterson, Kelly Cusick
IET-4: The "Nitty Gritty" of Developing Valuable Microinsurance Products
This presentation will introduce the microinsurance product development process and how microinsurance pricing differs from the traditional market. Speaker Michael J. McCord will take the audience through the pricing process, highlighting the role of the actuary in microinsurance, and provide real world casualty microinsurance product development examples drawn from his 25+ years working in the field. Participants will walk away with practical examples and a more concrete understanding of how valuable casualty microinsurance is developed and priced.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Betty-Jo Walke, Peter Lowth
Panelists:
Michael McCord
IET-2: Internet of Things: How Autonomous Cars Drive Better Than Humans And How to Insure Them
Vehicles are being developed with autonomous driving features. Some of these autonomous features are being rolled out into both personal lines vehicles as well as commercial transportation network systems and other fleets of vehicles. This presentation will demonstrate how the risk exposure is changing, how the insurance product needs to evolve to cover the changing insurance needs, and how to develop an insurance product that uses InsurTech solutions to price and deliver the appropriate coverages. We will compare autonomous driving behavior to human driving behaviors and share the results of a telematics model that scored both humans and autonomous driving. We will then demonstrate how autonomous driving reduces crashes and in turn reduce insurance costs, but creates new types of exposure from autonomous features (cyber and product liability) that were not traditionally needed in an auto insurance product. Finally, we will present how to construct an insurance product that effectively covers autonomous vehicle exposure, uses InsurTech solutions like Telematics to quantify and communicate the autonomous feature impact on insurance exposure/rates, and provides relevant modern coverage for evolving insurance market needs. Get out of the dinosaur era and into the future with new modes of transportation insurance.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
David Shleifer, Pamela Sealand Reale, Brent Petzoldt
Panelists:
Edward Ford, Sheri Scott
RP-9: Big Data and the Role of the Actuary
The use of Big Data and new analytical techniques is disrupting established practices, procedures, and markets in the highly regulated insurance industry. The American Academy of Actuaries Big Data Task Force recently released the monograph "Big Data and the Role of the Actuary". This session will cover current and emerging practices, regulatory considerations, and professional issues related to Big Data and insurtech. Actuaries, who must comply with the Code of Professional Conduct and the actuarial standards of practice, will need to be aware of these issues as they help their companies and clients navigate through this changing environment.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Francis Gribbon, Stephen Watkins, Brent Petzoldt
Panelists:
Robert Miccolis, Dorothy Andrews
RP-8: The Regulatory Response to Insurtech
Regulation presents both a hurdle and useful input to the development and acceptance of insurance technology. The National Association of Insurance Companies (NAIC) and state regulators have put much effort into assessing a constructive regulatory response.
The discussion will be centered around topics such as price optimization, use of Big Data in fraud detection or claim settlement amounts, and regulator treatment of complex pricing algorithms in rate filings (where the regulator may not have expertise to evaluate).
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Fred Karlinsky, Cheng Khang Saw, Brent Petzoldt
Panelists:
Elizabeth Dwyer, Peter Kochenburger
RP-7: Lights! Camera! Action!
The Committee on Professionalism Education will present a new round of skits! The skits provide a venue to discuss how you can use professionalism in practice, and audience participation is encouraged! This session will earn continuing education in the area of Professionalism.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Andrea Cablayan, John Burkett, Betty-Jo Walke, Pamela Sealand Reale
Panelists:
George Levine, Peter Royek, Mary Hosford, Michael Chen
RP-6: Regulatory Perspectives in Ratemaking and Pricing: Something Old, Something New
In this session, you, the audience, have the opportunity to select our panel’s discussion topics! Hear from a panel of three eminent regulatory actuaries who are active in the ratemaking and pricing realm as to how they and their states view current burning issues. These topics (4-6) may include:
• Predictive modeling/GLMs as used in pricing and underwriting in general
• Pricing that incorporates policy renewal expectations/Use of Retention Models
• Other price optimization topics
• Usage based insurance (UBI)
• Controversial/proxy rating variables
• Machine Learning and Big Data
• Role of actuaries vs. data scientists
• Role of regulators vs. companies vs. consultants
• Confidentiality of proprietary models
• Catastrophe modeling as used in pricing
• Capping (of premium/rates/rate changes) and other renewal pricing topics
• Other topics suggested by the audience
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Jonathan Taccone, Andrea Cablayan, Andrea Cablayan, Su Wash
Panelists:
Lisa Chanzit, Cara Blank, Lawrence Steinert, Wanchin Chou
R-4: Cluster Analysis in Loss Development
Cluster analysis is a collection of statistical tools designed to group similar data objects. In this session, we will look at these tools applied to grouping of similar loss development patterns.
In ratemaking, loss experience is typically aggregated into development triangles according to the needs of state-by-state rate filings. In loss reserving, losses are typically aggregated based on financial reporting categories. Cluster analysis can help to investigate more optimal ways of grouping the data. This session will discuss some useful ways to visualize the clusters and share some practical successes and challenges in applying clustering on loss development triangles. Topics include “hard” clustering such as k-means and “soft” or “fuzzy” clustering with similarities to traditional credibility theory.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Andrea Cablayan, Chris Fievoli
Panelists:
David Clark, Victoria Jiang
R-3: Advances in Personal Lines Risk Classification
Insurance risk classification continues to become more sophisticated and refined, fueled by enhancements in data, analytical techniques and computing power. This session will open with a brief discussion of some recent risk classification developments have been covered in the press. We will focus on personal lines insurance and discuss real examples of novel risk classification used in pricing and what issues the classifications aim to address, what data and techniques are used to develop these classifications, and what implementation challenges they may present. We will then examine how advances in telematics data and vehicle safety features changed and will continue to change risk classification.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Betty-Jo Walke, Kristen Dardia Turner, Peter Lowth
Panelists:
Ron Lettofsky, Katey Walker
M-17: Roof Intelligence: A Complete Picture
Using data to understand risk is not a new concept. However, with advances in aerial imagery and machine learning, it is now possible to collect even more granular and accurate data about roofs and property characteristics. Adding this new digital data to the current tabular data that has historically been used leads to a much more dynamic and complete view of property risk. This session will dig into the processes that are used to quantify the aerial imagery data, and how it can be used to develop that more complete assessment of risk.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Edward Yao, Michael Solomon, Cheng Khang Saw
Panelists:
Tanya Havlicek, Howard Kunst, Taylor Brown
M-16: The “Data” part of “Big Data”: Data Pre-Processing Emerging Topics
This session will cover current concepts, techniques and tools relating to data preprocessing for predictive models. Hadley Wickham’s “tidy data” concept will be introduced. We will also introduce some of the “tidyverse” packages, including tidyr and dplyr.
The session will also cover feature engineering to create new variables with more predictive ability. Common dimension reduction techniques such as using clustering to create alternatives to class codes or medical diagnostic codes will be presented. Other cardinality reduction methods such as pre-processing categorical variables with trees will be introduced.
We will also show some methods to transform numeric data, such as variable transformations and methods of selecting cut points for binning.
Some of the illustrations will use a public dataset such as the loss reserving data pulled from NAIC Schedule P. The presentation will also illustrate some of the techniques on a workers compensation claims data set from a modeling competition.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
David Shleifer, Pamela Sealand Reale
Panelists:
Louise Francis, Scott Sobel
M-15: Advances in Cyber Risk Modeling
Cyber risk offers insurers an opportunity to grow their business and help society become more resilient in the face of a new threat. Eye-opening incidents such as Dyn and Wannacry illustrated the potential for catastrophic losses, but insurers remain challenged to understand how often cyber incidents can occur and how these can affect the performance of their business. As a result, cyber risk management decisions tend to be made with limited data and too much weight is put on intuition or broad assumptions. Risk models help insurers quantify the likelihood, severity, and economic and insured financial impact of cyber incidents. In this session, you’ll learn about the advances being made in the development of probabilistic cyber risk models and how these can be leveraged to inform risk management decisions.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Fred Karlinsky
Panelists:
Pamela Eck
R-2: Insurance-Linked Securities: How to Grow Past Cats When the Wind Keeps Blowing
Insurance-Linked Securities (ILS) markets were founded to provide property coverage against natural catastrophes. ILS markets have historically provided market leadership and stability around catastrophe events. Learn how the ILS sector has grown through two active catastrophe years. And, see how capital markets capacity can support new specialty risks – such as cyber – while still meeting the needs of its original constituency.
Join us for a lively discussion of how ILS capacity can complement current capacity to help clients achieve target ROE’s and investors achieve attractive financial returns.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Edward Yao, Chris Fievoli, Brent Petzoldt
Panelists:
Zachary Breslin, Tom Johansmeyer
M-7: Process Mining in P&C Insurance
Insurers are continuously expanding their data analytics toolkits. One new technique is process mining, which evolved to better measure industrial processes. Today, leading insurers are using this technique to improve measurement of and provide transparency into difficult to quantify human costs, such as Underwriting overhead, ULAE/ALAE, and service costs. This session provides an overview of the technique and applications.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Jonathan Taccone, Garrett Bradford
Panelists:
Anthony Beirne, Jonathan Wong
R-1: Don’t Put All Your AALs in One Basket: Event Level Data and Pricing Catastrophic Concentration of Risk
Insurance risk pooling seeks to spread variability across a large number of independent risks. However, for insurers exposed to natural catastrophes, risks are not independent. Significant dependencies may exist between individual policies, and surcharges may be warranted for individual risks more correlated with an insurer’s portfolio.
Although pricing for concentration of risk is a longstanding theme within historical actuarial literature, modern catastrophe simulation models have enabled considerable improvements to traditional techniques. Catastrophe models are most often used to obtain estimates of average annual losses (AAL); but other model outputs at the simulated event level are not as widely used. Actuaries can utilize these model components to measure correlations between individual risks, and calculate concentration of risk charges that are more granular and accurate than ever before. In this session, we will review traditional techniques for pricing the concentration charge, describe the stochastic data outputs of a catastrophe model, and provide quantitative examples of how the event level simulation output could be used by a primary insurer to accurately reflect its reinsurance cost at the territory level.
Using the quantitative methods described, this session will explore these crucial elements of risk concentration in pricing:
1. How to assign credibility to catastrophe model results in remote layers
2. How sensitive certain methods may be to tail assumptions
3. Techniques for concentration pricing with limited data
4. Regulatory concerns for concentration pricing
5. Why ideal territorial ratemaking approaches for reinsurance cost may differ from those for loss cost
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
David Shleifer, Kristen Dardia Turner, William Frierson, Peter Lowth
Panelists:
Cody Webb, Victoria Gomez, Tim (Chi-Fan) Wei
M-14: It's Time to Replace MLE with Bayesian Shrinkage for Model Estimation
Fitting models by Bayesian shrinkage instead of MLE improves model accuracy (i.e., it reduces estimation and prediction variances). This previously required ad hoc approaches, but is now a systematic methodology, with clear procedures and a goodness-of-fit measure. It can be applied to most actuarial models, like regression and GLM models, and even reserve development models.
This session, coming from a research project done for the Ratemaking Committee, discusses the advantages of both frequentist and Bayesian shrinkage and shows how to apply them. Several classification ratemaking models are fit as illustrations.
The session will be an interactive discussion. The steps to do the modeling will be set out, along with an introduction to R packages to implement it. The handouts will have more detail, including code that the attendees can use to experiment and/or modify for their own applications.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Kristen Dardia Turner
Panelists:
Gary Venter
M-13: Sparsity Blues: How to Improve Your Predictive Model with Categorical Data
Linear models don't always handle categorical data well. And yet, categorical predictors are common in classification rating plans. Credibility has long been a tool which can smooth predictions across segments. This presentation will illustrate decision trees and naive Bayes as alternatives to a strict credibility-focused modeling approach.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Fred Karlinsky
Panelists:
Brian Fannin
M-12: Python for Actuaries
This session will introduce the Python programming language. We will review some of the significant libraries that are relevant to actuaries. The session will walk through a machine learning exercise using the scikit-learn package
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Jonathan Taccone, Fred Karlinsky
Panelists:
Brian Fannin, John Bogaardt
M-11: Overview and Practical Applications of Machine Learning Methods in Pricing - Part 2
The term “machine learning” covers a range of methods that can be powerful, with very practical benefits, in pricing and other insurance applications. Such methods can aid in further improving GLM results or more broadly bring valuable insights to complex problems. There can also be a number of practical challenges in using these methods effectively. This is the first of two sessions reviewing a range of commonly used methods and illustrating different ways they are being applied in insurance, including some case study results. These sessions will focus on the high-level mechanics of each method and the benefits/challenges of their application – as opposed to the underlying technical details. Part 1 will focus on tree-based methods including random forests and GBMs. Part 2 will cover other popular methods including penalized regression methods and neural networks. Each part can be attended without the other.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Brett Nunes, Garrett Bradford
Panelists:
Benjamin Williams, Graham Wright
M-10: Overview and Practical Applications of Machine Learning Methods in Pricing - Part 1
The term “machine learning” covers a range of methods that can be powerful, with very practical benefits, in pricing and other insurance applications. Such methods can aid in further improving GLM results or more broadly bring valuable insights to complex problems. There can also be a number of practical challenges in using these methods effectively. This is the first of two sessions reviewing a range of commonly used methods and illustrating different ways they are being applied in insurance, including some case study results. These sessions will focus on the high-level mechanics of each method and the benefits/challenges of their application – as opposed to the underlying technical details. Part 1 will focus on tree-based methods including random forests and GBMs. Part 2 will cover other popular methods including penalized regression methods and neural networks. Each part can be attended without the other.
Source:
2019 Ratemaking, Product and Modeling (RPM) Seminar
Type:
Concurrent Session
Moderators:
Brett Nunes, Garrett Bradford
Panelists:
Benjamin Williams, Graham Wright
