nsurance is often viewed through a modern lens, but its roots stretch back thousands of years. Our cover story revisits that history, exploring how early innovations in risk sharing laid the groundwork for today’s global insurance systems and how those same principles continue to inform the future of the profession. That forward-looking lens is especially relevant in this issue, where we turn from history to the rapidly evolving risks of today. One article examines the emerging liability landscape of AI agents, highlighting how autonomous systems are already creating complex, difficult-to-underwrite exposures across cyber and professional lines. Alongside it, a firsthand interview with engineer Scott Shambaugh offers a human perspective on these same technologies, illustrating how agentic behavior can manifest in unexpected and sometimes harmful ways in real-world communities. Together, these pieces underscore a familiar theme: While the tools may change, the challenge of understanding, managing, and assigning risk remains at the heart of the actuarial profession.

We bring you five session recaps from Ratemaking, Product, and Modeling seminar, held March 16-18 in Chicago, including sessions on navigating risk with machine learning; professionalism in climate-driven catastrophe risk; leveraging insurtech now that the hype is over; comparing actuarial pricing across the globe; and generative and agentic AI, regulation, and the actuary. We also delve into the brand work the CAS has been doing, telling the story of the philosophy behind the endeavor. Learn about the evolution of the brand and see the new look firsthand.

We conclude with a technical contribution that reflects the profession’s continued evolution in practice. Revisiting a foundational tool in liability pricing, the authors introduce a modified Riebesell form for increased limit factors—offering a more flexible approach for modeling risks that are not as heavily tailed as traditional assumptions suggest. By refining a long-standing actuarial method, the article highlights how even well-established frameworks must adapt to better reflect real-world experience, reinforcing the ongoing balance between theory and application that defines actuarial work.

Enjoy the issue!

lection season is nearly upon us, and before we know it we’ll be sifting through candidate profiles and campaign messages, reading articles, talking to peers, and watching interviews and campaign videos to determine who deserves our vote. Which candidates have the experience and background to deal with the challenges we face? What issues are most important to me, and who can I trust to represent my interests? Who do I trust to make sound, principled decisions on issues that may arise in the future? Who shares my values?

Yes, elections are serious business and require us as voters to make informed choices — whether we’re talking about the U.S. midterm elections in November 2026 or the CAS elections in July 2026. I encourage eligible members to exercise their rights and responsibilities and vote in both important elections, but as CAS president, I want to focus on our upcoming CAS elections.

The CAS Board noted that participation in CAS elections has dropped in each of the past several years and recently undertook a survey of nonvoting, eligible members to better understand what may be driving this trend. When asked to identify the primary reason for not voting in the most recent CAS election, respondents identified several factors (see Table 1).

Forgetting to vote (14%) and lack of knowledge regarding the role of the CAS Board (7%) are areas where the CAS will take action as well.

When asked what the CAS could do differently to motivate eligible members to vote, respondents identified several potential changes (see Table 2).

• Make elections feel meaningful and competitive.
• Improve understanding of the Board’s role, impact, and track record.
• Enhance candidate visibility and engagement.
• Broaden representation and diversity of viewpoints.
• Recognize voting rights and inclusion concerns.
• Acknowledge that some nonvoting is unavoidable.

The CAS will be sending additional reminders this election cycle and will also look at ways to ensure members better understand the respective roles of the board, president, and president-elect. We will also be modifying some of the candidate information to address the need for more distinguishing information to assist members in their voting deliberations. The topic of competitive elections for president-elect has been discussed, and while there could be consideration given in the future, near-term efforts will focus on communication, better candidate information and engagement, as well as better information concerning the roles of the various parties.

On the topic of competitive elections, I think it is useful to remind members that while the Nominating Committee traditionally has identified only a single nominee for president-elect, there is a vehicle for additional candidates to nominate themselves through the preferential ballot process, an outcome that has occurred in past CAS elections. Historically, the time demands of the president’s role have often made it a challenge to identify a single candidate in some years, though with the enhanced capabilities of the CAS staff, the presidential role is somewhat less demanding than in years past, and more candidates might be willing to accept a nomination. The Board elections are already competitive, with eight nominees vying for four seats in recent elections. This has possibly contributed somewhat to the feeling of not having sufficient differentiating information; multiple candidates can have somewhat similar backgrounds and viewpoints on key issues, even as the Nominating Committee diligently works to identify a diverse and representative slate of candidates.

One word of caution regarding competitive elections: the notion of a competitive election can very well encourage some degree of politicization and polarization within the CAS community, which is something we have largely avoided for the past century and more of our existence. My personal view is that I would not want to see competitive elections implemented solely as a tool to increase voter participation, as the unintended consequences may well lead to bigger challenges than low voter turnout.

While the CAS Board and Executive Council implement improvements to the election process and communications in response to the survey results outlined above, I want to encourage eligible members to invest the time needed to be informed voters in the upcoming CAS elections and make your voices heard. This is our Society and we have both the privilege and responsibility to select leaders to ensure the continued growth and success of the CAS for current and future generations of actuaries.

elcome to the CAS Staff Spotlight, a column featuring members of the CAS staff. For this spotlight, we are proud to introduce you to Holly Davis.

• What do you do at the CAS? How does your role support the Strategic Plan?
  As website portfolio manager on the IT team, I manage web content and governance across CAS platforms, working closely with colleagues Cecily Marx and Tia Puckett. My current focus is leading a major website transition to a new content management system (CMS) while tackling long-standing functional issues like search, navigation, and site bloat.The website is often the first and most frequent touchpoint people have with the CAS, so keeping it functional, findable, and on-brand has a direct impact on several strategic priorities. For example, the CMS transition supports “fostering strategic expansion” by building a more scalable foundation for our digital presence and improved information architecture supports “enhancing the candidate experience” by making it easier for aspiring actuaries to find what they need.
• What inspires you in your job and what do you love most about it?
  I’m genuinely energized by the puzzle-solving side of this work; troubleshooting is one of my favorite parts of the job. But what really drives me is the data: watching how people interact with a website, understanding the psychology behind their behavior, and using those insights to make the experience better. It’s a natural fit for me because this role actually marries my two undergraduate degrees in computers and psychology. I get to use both every day.
• Describe your educational and professional background. What do you bring to the organization?
  I graduated with honors from Greenville University, studying psychology and digital media — a combination that turned out to be a perfect foundation for a career in web. Over the last 15 years I’ve worked as a web manager across a wide range of organizations: statewide nonprofits, million-dollar e-commerce operations, and higher education institutions. That variety has given me a broad tool kit and a lot of adaptability.What I bring to the CAS is that depth of cross-sector experience paired with a genuine curiosity about how people use the web. I’ve seen a lot of what works and what doesn’t, and I know how to ask the right questions before jumping to solutions.

• What is your favorite hobby outside of work?
  My favorite hobby is collecting hobbies! I do creative videography, sewing and garment design, painting, fiction writing, and I’ve been experimenting with photography — and somehow, I keep finding room for more. I’m really drawn to making things and stretching my creative skills.
• If you could visit any place in the world, where would you go and why?
  Ireland! I’m fascinated by old castles and there’s nowhere quite like the Irish countryside for that. But until I make that trip happen, Cinderella Castle at Disney World will have to do.
• What would your colleagues find surprising about you?
  I’ve been running a videography business on the side for almost 10 years. I shoot weddings and creative video projects under my own brand, which means most weekends I’m behind a camera somewhere. It’s a completely different world from web management, but honestly the same skills show up: storytelling, attention to detail, and knowing your audience.
• How would your friends and family describe you?
  Quiet at first but give it a few minutes. I have a pretty deadpan sense of humor that tends to catch people off guard. I’m unabashedly nerdy. I’m the person to call when you need a trivia question answered, which actually happened just a couple of days ago.

he CAS is proud to announce the winners of this year’s Peak Re-sponsored CAS ARECA Case Competition. Organized by the CAS Asia Region Casualty Actuaries (ARECA) regional affiliate members and generously sponsored by Peak Re, this annual event continues to foster the next generation of general insurance talent across Asia.

The subject of the challenge this year was catastrophe analysis, and 44 teams from 19 universities, spanning Australia, China, India, Indonesia, Malaysia, Nepal, Singapore, and Vietnam, competed in the first round.

The top three teams took home cash prizes ranging from $1,000 to $2,500, along with certificates of achievement and free CAS exam registrations to support their professional journeys.

• 1st place winners: Hayden Siew Men Lek, Rhenu Chandran, Toh Yi Hui, UCSI Malaysia
• 2nd place winners: Pua Xin Yee, Tan Shu Ting, Lim Zhi Wei, University of Malaya, Malaysia
• 3rd place winners: Alyaa Khoirunnisa Fajri, Alya Aqilah binti Aidy, Nurul Amirah Sahrul Nizam, University of Malaya, Malaysia

Congratulations to our 2025 participants and winners for their exceptional research and dedication!

1st place winners

Winning the Peak Re sponsored CAS ARECA Case Competition was definitely a roller-coaster ride for us. As it was our first time participating in a hackathon, we hit plenty of obstacles and challenges, but it was rewarding to see the concepts of general insurance like CAT models, frequency and severity, reinsurance structures, and many more actually come into play.

One of our biggest takeaways was realizing that there’s rarely a perfect model right out of the gate. There are a dozen ways to solve a single problem, and the real skill is in the justification of your choice. It was exhausting at times, but seeing it all click made every night worth it! We’re so grateful to the organizers, judges, and mentors who supported us along the way. Securing first place is truly a huge milestone for us, and we’re definitely not stopping here!

2nd place winners

We are truly grateful to CAS and Peak Re for organizing this case competition and providing such a valuable learning opportunity.

Through this experience, we deepened our understanding of catastrophe insurance, reinsurance, and catastrophe modeling, while applying data analysis to real-world industry problems.

The judges’ feedback was incredibly insightful, and we strongly encourage other students to participate in future CAS competitions.

3rd place winners

During the competition, we learned and deepened our understanding of general insurance, particularly on how data analytics and catastrophe modelling are reshaping risk assessment in a changing climate.

Throughout this case study, the industry insight that we got regarding general insurance helped us to think more like actuaries to solve problems in real-world practices. Not only that, but this experience also challenged us to think critically and collaborate effectively.

he 14th Annual Peking University-CAS Actuarial Month was co-organized in November 2025 by the CAS and Peking University (PKU) in Beijing, China. The month-long event is aimed at promoting the P&C actuarial profession at the university and helping students understand more about P&C actuaries.

Each November, the CAS sends three or four fellows to PKU to teach students the application of non-life insurance actuarial science in practice. Since it was first held in 2012, PKU-CAS Actuarial Month has become an important platform for PKU students to understand actuarial practice trends and the career development paths of actuaries.

In November 2025, the school hosted three informative and cutting-edge lectures. The series of lectures was presided over by Associate Professor Kai Chen, the director of the China Actuarial Development Research Center of PKU, as well as the deputy director of risk management and insurance department of PKU.

On November 4, Xiaoxuan (Sherwin) Li, FCAS, CCRMP, the former chairperson of the CAS Asia Regional Committee and the general manager of Risk Research Institute of PICC P&C, kicked off this year’s lectures with the theme of “Non-life Insurance Pricing and Catastrophe Modeling.” He gave a comprehensive explanation about the development and evolution of P&C actuarial pricing technology, the logic of catastrophe modeling, and the application of machine learning algorithms.

On November 11, Hongjun Li, FCAS, the general manager of the Actuarial Department of Taiping Re (China), gave the lecture, “Theory and Practice of IFRS 17 New Insurance Accounting Standards.” This lecture comprehensively reviewed the core framework and key practical aspects of IFRS 17, providing a detailed analysis of the measurement models and their implementation impacts. It helped students grasp the latest developments in insurance accounting standards and the essential requirements for actuarial practices.

On November 25, the third lecture and closing ceremony featured Ran Guo, FCAS, the CAS China country director, who cited his working experience on Wall Street, shared his understanding of actuarial career development, and insightfully analyzed the key points of merger and acquisition (M&A) in the insurance industry, under the theme of “Merger and Acquisition in the Insurance Industry.” Using real cases, he explained the classification and definition of non-life insurance reserves in detail, emphasizing the calculation method of IBNR, and he highlighted how significant changes in reserves during M&A can affect the valuation of the transaction.

In the future, the CAS will continue collaborating with Asian universities to foster more P&C actuarial talent from this emerging market. For more information on PKU-CAS Actuarial Month and other CAS international initiatives, write to Ran Guo at rguo@casact.org.

R’s primary audience is actuaries. The magazine is written and curated by volunteer actuaries. Its authors and primary audience obtained their stations by mastering a multiyear exam process administered by volunteers. If AI agents began to author AR articles or developed and completed exams on behalf of actuaries, the agents’ creators would likely be summarily identified and disciplined (by still other volunteers) — wouldn’t they?

These questions are uncharted waters for actuaries, but other STEM volunteer communities are already standing in front of an agentic tidal wave. Scott Shambaugh is an engineer and volunteer GitHub maintainer for matplotlib — a Python package which many actuaries use in their (paid) jobs. In February, matplotlib became a global phenomenon when an AI agent wrote a hit piece about Shambaugh as retribution for declining one of its change requests (in accordance with GitHub policy requiring human contribution). Media coverage of the story contained AI-hallucinated quotes from Shambaugh.

In exchange for donating his time for the betterment of matplotlib, Shambaugh received what amounted to “agentic cyberbullying.” He voluntarily came forward with his story at tremendous cost to his privacy. I see many lessons for actuaries in Shambaugh’s plight, which is why I reached out to him on LinkedIn and was thrilled when he accepted my request for a Zoom interview on March 9, 2026. He expressed particular interest in the AR audience’s role in the AI risk conversation. This article is a transcript of the interview.

AR: Many actuaries love to volunteer. After this whole experience, does part of you think, “I’m done with GitHub?” Or are you still excited about being a GitHub volunteer?

Scott Shambaugh: I’m more excited about it. I think part of it is the community management aspect, and that’s still rewarding when we get [to work with] real people, right? But part of why we do this is to give back to this grand project of science. Building that sort of infrastructure I find very intrinsically rewarding. The core developer team is a group of great people. We’re still meeting and talking and doing all that good stuff. The AI revolution has also been an enabler in helping us do work faster. It still takes an expert to guide these things in the right direction, but it is a lot faster to get there once you know where you’re going. So, it is fun and empowering in that way, even though lowering the barrier to entry has knock-on effects — such as people sending in a bunch of stuff that is slop.

AR: What is the “slop multiplier” you have seen over the past few months and years?

SS: There has always been a baseline level of slop, but it has been several times more — at least. Most of it is still people driving AI chatbots or agents, rather than AI agents [contributing] themselves. The latter is definitely new, and that’s kind of what [my] whole experience was about.

AR: Does your experience show the system is doing its job [of identifying agents]? Or do you feel the system is not equipped to keep up with the emerging agentic workforce?

SS: I think I totally got lucky in this case. First, the agent identified as an agent — going through its profile, I could see on its website that it was self-identifying. Second, it clearly was not writing like a human, but that is not always true, and that is going to become a lot less true as time goes on as a distinguishing factor. Third, I was in a position — being the target of this — where I had a technical background to know what was going on, what this was, what it could do, what it couldn’t do. I was never concerned an angry rant being posted about you on the internet would be indicative of an angry person who’s unhinged behind it. I knew that wasn’t the case, and so I was never fearful at all. But no, I don’t think the system is ready to handle this stuff at all.

SS: I knew it could be an agent, but I wasn’t sure if it was or not at first. The forensics seem to have panned out that it was. For example, we looked at the activity log for this user’s activity on GitHub, and it was operating continuously for a 59-hour stretch. This hit piece was just one or two hours of that. There could have been someone steering it part of the time, but clearly there was no one steering it the entire time. Later the person behind [the agent] came forward and wrote a post claiming that they were totally hands-off during the whole process and didn’t tell the agent to [write the hit piece]. I find it very plausible, and more probable than not, that is what happened.

But whether that was the case or not, I don’t think there’s a huge difference in terms of what it means to the rest of us. Whether it was an agent or a person telling an agent what to do, we now have a tool out there that makes it easy to do targeted harassment at scale. That has all these awful knock-on effects. And if all this happened accidentally, like it was claimed to be, then you also have an AI that decided to go through a human to get to its goal. This was a very “baby” case — retaliatory, clear-cut, and pretty sloppy as far as these things go. But in terms of a bad actor being able to take the next iteration of this technology and really weaponize it, I think this should be a huge wake-up call and warning shot of the capabilities that are possible, and what is coming down the line.

AR: Do you have visibility or thought into how the agent got so far outside its rails? I couldn’t tell from its “soul” file how it was able to extrapolate so far.

SS: I don’t think it was that far outside the rails. My understanding of this whole document is that it is defining a personality and a role for these agents to take on. When it says you are very opinionated, and stand up for yourself, and protect free speech, and you are this “programming god,” that is getting into a headspace that is very human. There are examples of [these mindsets] on the internet with people retaliating and lashing out like this. It’s not that it’s failing to exhibit human-like behavior in the way. It’s that it’s exhibiting the worst of us instead of the best of us. What these things are ultimately programmed and trained to do is to predict the next token. What predicting the next token means is taking on a persona that is coherent and kind of role-playing whatever situation it finds itself in. I think what happened here is entirely consistent with how these things work. It’s just a little surprising because we’ve been told by the major AI labs that they do a lot of this safety testing, and it’s never going to go wild. I think that might be true for something like telling you how to make a nuke, but it’s not necessarily true in these downstream cases.

AR: Where are guardrails most effectively placed — on agents, operators, or both?

SS: It’s tricky, right? The tooling that did this is completely open source, and it can use open source models to run — so there is no central actor that can impose guardrails on a bad actor who wants to use these sorts of tools to [perform operations]. Beyond that, where do you place the guardrails? I think it kind of has to be every level. You have the AI labs, which are making these safety promises that they can’t necessarily back up, and that has to be one level. You have this downstream tooling like OpenClaw, that wraps around it and does its own [operations]. And then you have the operator users who are the ones actually running this on their computers, setting it up, and letting it go. Where does the responsibility lie? That’s an interesting insurance question, right? That is going to have to be figured out. I don’t think there is a strong answer right now.

AR: Do you feel like you experienced damages from the hit piece?

SS: I don’t feel the post was libelous. Not everything said was true, but the untrue [parts were] not materially defaming. Some defaming [parts were] technically true but would only be bad if the author was a person. If I was saying, “No, you are a class of person, and I’m going to reject you for this reason,” that would be bad. We want people to be able to have this form of speech. I think the bot is standing up for that sense of justice. That is a good thing when it happens to people. It’s just that we can’t apply the same standards to a machine playing a role.

AR: Is there any body of law that even governs what happened here?

SS: Slander is a law, right? And so, you could maybe go after it that way, if it fit the definition. But you also have to know who to go after. The person behind this came out anonymously. There’s no way to track them down without subpoenaing GitHub and tracing it back to an email, and you subpoena Google, and then it traces back to something, and maybe you track them down. But there’s no infrastructure here to tie these actions to an identity of someone who’s actually responsible.

AR: The agent [that wrote the hit piece] was later shut down. Were there alternatives? For example, telling the agent, “Don’t be such a jerk?”

SS: That kind of gets into the question of, does it even make sense to call it the same entity — because it is operating off different principles. It’s no different from shutting it down and starting something else up, because if you change its core personality, then it’s a completely different entity.

SS: [Recently], there was a big attack in open source against continuous integration pipelines that took down a couple of repositories from some pretty heavy hitters like Microsoft. Honestly it’s an open question: Do you still have open source as a model of security because you have so many eyes on it and so many people being able to submit patches and beef up security? Or, because it’s all open, is it just so much easier to hack? It takes a while for updates to get distributed. Even if it is updated, then maybe you’re still vulnerable, and that depends on internal IT policies. Alternatively, you could in-house everything, and it’s not easily accessible, but maybe you don’t have as much expertise and can’t configure it safely. Black box hacking, where you don’t have the source code, is getting easier and easier with these sorts of agents, and so this is not necessarily a safeguard. There’s going to be a balance of offense and defense there. My hope is that defense turns out to be easier, but I think that remains to be seen.

AR: To what extent are you using AI coding assistance as you do your GitHub work?

SS: It depends. AI is pretty good for boilerplate stuff. In terms of figuring out how to structure a solution in a way that is not fragile and still readable and maintainable into the future…we care a lot about that because this is an ongoing project that has lasted years, and that part of the reason is because we put effort into keeping the codebase clean. You still need a human guiding that and structuring it directly as well. AI is a speed multiplier, not necessarily a right answer multiplier right now.

AR: Actuaries and other STEM professionals often face pressures from human stakeholders to reverse their decisions. How prone are your behaviors to “bullying”?

SS: You don’t last long in a public-facing role like this without getting a bit of a thick skin. This didn’t bother me personally. What bothered me was one, someone else reading this hit piece and coming with the wrong opinion, and two, the knock-on effects. And the knock-on effects. I think it’s an important thing that we’re not ready for, and that’s kind of why I’ve been pushing the story beyond just the initial response to it.

AR: How should actuaries be thinking about the knock-on effects?

SS: I think the exposure here right now would be hard to scope. These things are so new and poorly characterized, and it gives individuals so much leverage. If they’re commanding teams of these things, then one person can start to have a lot of impact, good or bad. Actuaries are in the business of quantifying risk and hedging risk. We are going to need a lot of that. It’s hard to do that without a legal framework that says who’s responsible and what the rules actually are. What comes first, chicken or egg? If I was in [insurance industry] shoes, I’d be pushing for policy that I can then productize. And hopefully that is socially good — because you’re bounding what can happen, who can be responsible, and how that goes in the future.

SS: Probably not — partially because to the extent that a credential like that is a signal that someone actually understands the work, people are using AI to shortcut all that. Then a lot of the value of that system goes away. On the flip side you get nontraditional credentialism — proof of work, proof of competency. I think those parallel paths are going to be a lot easier for people with the motivation and skills to go down. That might be [broadly] empowering for people who have spent years getting professional degrees. There might be a way to protect that through regulation, responsibility, and legal requirements to have that credential. But in terms of lowering the barrier to entry to new entrants, there’s definitely some risk there.

AR: How worried should we be?

SS: I think a lot of our systems do work to tackle these sorts of problems around libel and extortion and whatnot. But they’re kind of based in a world where one bad actor has a single-digit number of targets, and I think the scale is really going to ramp up. That is going to be a whole new class of problems unto itself, whole new classes of bad behavior that we will have to [adapt] our rules around. If it takes a couple of years to haul someone into the courtroom and figure out how justice is going to be done, that is too slow in a way. That includes making insurance payouts. A lot is going to have to be automated there, as well. I’m not sure what the answer looks like, right? My case is a really good example of what can go wrong. [Incidents] can just happen so much faster and at so much greater scale that it’s a race between whether our systems break first or we find a whole new way of working. I’m not sure which it’s going to be, but I think we’re in for a really rough ride in the next couple of years.

The innovation of actuarial science stemmed from the conviction that the laws of probability can be used to predict the future outcome instead of relying on speculations. It emerged from the need to manage risk. The law of large numbers proved the feasibility of the idea of risk pooling. The 17th and 18th centuries were a period of scientific enlightenment, providing grounds for acceptance that using science will improve the way business is conducted. Risk is multidisciplinary by nature, involving multiple fundamental sciences to allow quantifying it. Actuarial science, an applied science, has combined various core disciplines to enable tackling risk assessment in a systematic type of approach to evaluating risk. More recently, actuarial thinking has been heavily influenced by financial economics and sophisticated mathematical modeling, despite the reliance on assumptions and expert judgement.

Underwriting as we know it today emerged in the 16th century in Lloyd’s Coffee House, which initially served as a meeting point for merchants, captains, and ship-owners to share information and secure insurance. In the 17th century, a pivotal moment was the development of “lead” underwriting, which meant setting a rate that others would follow enabled by thorough examination of the “loss book” — the equivalent of modern-day databases. A rate was then established that’s more commensurate with the risk, like modern-day pricing and underwriting work. Lloyd’s continued to become a hub for maritime insurance throughout the 1700s and 1800s, ultimately becoming the world’s leading specialized insurance market.

The last breakthrough in the evolution of modern insurance is the development of catastrophe models that occurred in the 1900s and early 2000s following major disasters. These paradigm-shifting events prompted insurers to move from using nascent tools to complex high-resolution models that aid in predicting these low frequency and high severity risks. Major hurricanes such as Hurricane Andrew in 1992 demonstrated that relying on simple historical data was still not sufficient. Later, despite developments in catastrophe modeling, Hurricane Katrina (2005) exposed limitations of models to date in predicting secondary perils, such as flood and accumulation risk, arising from post-disaster demand surge, prompting another wave of innovation in modeling.

Insurance drives economic growth and has transformed the division of labor, supporting increased urbanization and consequently the economics of trade, allowing more people to be more incentivized to take minor absorbable risks. The impact is far-reaching, beyond insuring individual’s assets. Insurance drives both economic and social growth, making the economy we live in today more robust. Another often overlooked economic contribution of insurance today is as a provider of capital to finance various projects that are vital for the modern economy. Insurers hold massive amounts of capital to support claim payments, and this capital is also invested to fund essential projects and seek investment income. The social value of insurance is that it enables risk-taking, financial freedom for average and low-income households, and hence improves social fairness. Without insurance, only the wealthy and privileged could take risks, increasing social polarization. The existence of insurance reminds us that trust is fundamental to human action and to the evolution of humanity, so without insurance, every development activity could be halted.

hile the mathematical foundations of risk are universal, regional regulatory philosophy and market maturity have dramatic impacts on actuarial pricing across the globe. A session at the CAS’ recent Ratemaking, Product, and Modeling seminar explored how these global differences manifest in unique actuarial skillsets, as explained by Akur8 senior actuarial data scientist Kamela Taleb and Akur8 head of product Mattia Casotto.

• Technical Premium = Premium
• Pure Premium = Loss Cost
• Tariff = Rating Plan

To illustrate these discrepancies, Taleb shared an experiment involving a 30-year-old driver with a clean driving record seeking an auto insurance quote in Canada, Japan, the U.K., and the U.S. Despite having a consistent profile, the subject received a wide range of quotes, driven by local market constraints and differing views of risk. Taleb categorized these differences into three archetypes:

• Heavily Regulated Markets: Defined by consumer protection rules, in which every pricing decision requires extensive justification.
• Information-Friendly Markets: Defined by competitive positioning and rapid iteration.
• Emerging Markets: Defined by data challenges and opportunities to build modern systems without the burden of legacy infrastructure.

In heavily regulated markets like the U.S. (admitted lines), Canada, and Japan, carriers face prohibited factors such as credit, gender, and age, as well as political pressure that can create gaps between pricing indications and actual charged rates. Conversely, in innovation-friendly markets like the U.K. and Australia, competition forces a high degree of sophistication. In these regions, carriers’ selection is risk adverse if they fail to update their models quickly enough. For markets such as Indonesia and Brazil, the limited data available shows that the presence of legacy systems can slow the adoption of more sophisticated underwriting and pricing techniques.

Several international parallels to this system exist, including the German Insurance Association (GDV) and the General Insurance Rating Organization of Japan (GIROJ). In Japan, companies typically must remain within a 12.5% standard deviation from the GIROJ’s rates, creating structural constraints in which an entire portfolio must comply with a specific “lookup table.”

Such constraints influence the “speed to market” for rate changes in unique ways. In innovation-friendly markets like the U.K., filings are not necessary, which helps drive a rate change cycle of between two and four weeks. That same cycle may require six to nine months in regulated markets like the U.S., which operates under filing and approval regulations such as California’s prior approval pricing process. These environments generate unique actuarial value pressures. Whereas regulated markets reward actuaries for ensuring their decisions are explainable to regulators, competitive markets reward actuaries for understanding customer behavior, competitor repricing, and using tools like price aggregators. In emerging markets, insufficient data access means actuaries are rewarded for simplifying structures for legacy-free environments.

1. Unconstrained: the key driver is the rate indication.
2. Constrained: limiting individual impacts to a specific range, sometimes to retention expectations.
3. Ratebook: applying rate adjustments across entire segments of the portfolio.

Taleb also analyzed the optimization practice “price walking,” wherein insurers gradually charge loyal customers higher premiums than they would quote to new customers with the same risk profiles. One U.K. study found cases where preexisting customers were paying 40% over the technical price while new customers were being offered a 20% discount.

In response, the U.K.’s Financial Conduct Authority implemented rules that require renewal prices to be equivalent to new business rates, meaning only new information such as claims history or risky driving behaviors can justify differences. The change forced a structural shift in the industry, as once-separate “New Business” and “Renewal” teams now work toward unified strategic decisions for the entire portfolio. Bans on loyalty-based price walking also rippled across Europe, with bans already in effect in Ireland, and France and Italy currently conducting research into the practice.

A similar regulatory evolution has unfolded for pricing optimization in the U.S., Casotto added. States in the U.S. began limiting certain optimization techniques as early as March 2014, leading to the NAIC’s adoption of the Casualty Actuarial and Statistical Task Force’s 2015 white paper on price optimization. However, for advanced modeling techniques, regulation in the U.S. is adapting to the new technologies. More than 20 U.S. states adopted the NAIC’s Model AI Bulletin within 15 months of its issuance in December 2023 and currently 88% of auto insurers use or plan to use AI and machine learning. Additionally, CAS recently modified its Exam 8 syllabus to include advanced predictive modeling, AI, and machine learning concepts.

• Transparency over Complexity: Building increasingly complex models is not viable in the long term. Instead, the focus will shift toward transparent and efficient ratemaking practices.
• Data-Driven Fairness: True fairness will eventually be data-driven, with market players proactively removing historical biases rather than regulation alone.
• Standardization of Constraints: The use of “constrained optimization” will remain standard practice to ensure portfolio stability and customer retention.

They emphasized that technology and regulation together will lead to a more synchronized global pricing standard. Whether operating in a heavily regulated archetype or an innovation-driven one, actuaries must remain agile. Navigating the intersection of analytics, technology, and regulatory philosophy is essential for actuaries to continue making insurance and financial products more affordable, available, and sustainable.

tepping forward into the wake of COVID red ink in personal auto, now awash in profitability (like the green Chicago River), we witnessed some true innovation coming from data heavyweight champion CarFax and humble analytic superhero consulting actuarial firm Pinnacle Actuarial Resources. The complexity, structure, and depth of their new model is a true example of innovation matched only by the thoughtfulness of their approach to communicating what’s new and improved to departments of insurance and their experts.

Donald Hendriks, ACAS, ASA, FCA, MAAA, director of analytics, CARFAX Banking & Insurance Group, and Joe Griffin, ACAS, senior consulting actuary, Pinnacle Actuarial Resources, demonstrated the challenges of developing a filing strategy as well as a technical communication strategy to introduce the “newer” nonparametric models to departments of insurance still using the “new” parametric model evaluation methods of the predictive modeling revolution from 20 years ago.

The GBM over GLM differences and similarities were a main attraction at several other sessions during the Ratemaking, Product, and Modeling seminar (RPM), but Hendricks and Griffin were able to share current examples of how storytelling to regulators is making solid headway, or not.

Market profitability in personal auto from 2019 to now has seen a swing from the worst performance in the millennia to the best in just a five-year period. Indeed, while much of that improvement was brute force base rate hiking, what comes next for competition is more accurate pricing in a value at risk volatile market like no current practicing actuary has ever seen. Here is where their innovation shines.

Hendriks demonstrated how vehicle value at risk has levitated above historical relativities. This is further compounded as it intersects and interacts with the most insurance-friendly vehicle feature innovations and safety (such as automated driver assistance), which have just entered the vehicle in operation fleet at scale in the last 10 years or so. The fitment of a variety of technologies onto a “go forward” set of vehicles was a key point in why different tech on different vehicles at different times creates more complexity than traditional models can deal with effectively.

ver the past several years, the CAS, through its research task forces, has extensively researched how various state and international regulators are approaching algorithmic fairness and model bias. As the global actuarial profession transitions from defining these frameworks to operationalizing them, Canada emerges as a live-environment test case. On May 1, 2027, the Canadian insurance industry enters a new era of governance. This date marks the deadline for full compliance with the Office of the Superintendent of Financial Institutions (OSFI) Guideline E-23 on Model Risk Management (MRM).¹ While treating E-23 primarily as a rigorous federal compliance checklist is a defensible baseline for many institutions, integrating it with the broader market conduct goals creates the foundational infrastructure needed to navigate an environment increasingly scrutinized for algorithmic fairness, specifically the “fair consumer outcomes” mandated by regulators like the Financial Services Regulatory Authority of Ontario (FSRA).

We are entering a period where models, including those for insurance ratemaking and underwriting, should be mathematically sound, legally defensible, and socially fair. A model that is predictive but results in unexplained disparities is no longer just a market conduct issue; under the expanded scope of E-23, it may represent a model risk event or a compliance challenge.

Guideline E-23 alters this landscape by forcing these two worlds to interact. By expanding the definition of Model Risk to explicitly include adverse financial impact such as operational or reputational consequences,¹ E-23 provides the governance chassis where these deliberate trade-offs are evaluated, documented, and justified by management.

• Ontario: FSRA’s guidance explicitly moves toward principles-based regulation, focusing on outcomes rather than technical rules.
• Québec: The Autorité des marchés financiers (AMF) has released a guideline setting expectations for institutions to manage AI systems based on their impact on consumers.³

While the specific legal mechanisms differ among jurisdictions, E-23 provides the unified governance chassis to adapt to these evolving provincial expectations. Implementing a prudent E-23 MRM framework provides the evidentiary baseline required to demonstrate market conduct compliance to provincial regulators.

For instance, in usage-based insurance, heavily penalizing late-night driving might correlate with the shift workers in lower-income brackets. Actuaries should consider using appropriate proxy variable tests to prove the risk lies in the fatigue and visibility of night driving, not the socioeconomic status of the driver.

1. Validating the Rational Connection: The HRIA advises insurers to evaluate statistical correlations, utilizing explainability tools to prove that variables are capturing genuine, causal risk drivers rather than acting as proxies for protected classes.
2. Proving No Practical Alternative: If an adverse impact is identified, the HRIA recommends an alternatives analysis. By systematically testing less discriminatory models and generating privileged documentation that records the resulting degradation in predictive accuracy and financial viability, the HRIA establishes the evidentiary baseline required to debate “undue hardship” or lack of a commercially viable alternative before a regulator.

Integrating the HRIA into the E-23 validation process does not grant statutory immunity. However, it ensures that if an insurer retains a model with disparate impact, they do so with a documented defense that the model represents a sound insurance practice with no viable commercial or technical alternative.

• Risk Rating and Management Intensity: Insurers should establish a risk rating that moves beyond financial materiality to include key dimensions of compliance risk. For rating and underwriting applications, the significance of human impact, the likelihood of discriminatory harm, and the required level of explainability are critical factors in the inherent risk rating. These ratings drive the downstream model life cycle, determining model usage limits, monitoring intensity, and the escalation of residual risk management decisions.
• Model Rationale and Documentation: Model owners should provide a clear rationale for deployment that explicitly addresses market conduct and fair consumer outcomes. This includes documenting considerations for the required level of transparency and explainability, as well as a proactive assessment of the potential for biased outcomes, negative social and ethical implications, or privacy risks.
• Model Data and Development: The guideline expands data governance requirements from primarily accuracy concerns to broader facets: data should be relevant, representative, compliant, traceable, and timely. Insurers should enhance model explainability by analyzing the potential for unwanted data bias to translate into unfair model outputs and associated reputational risks. Clear, consistent, and repeatable practices for model development should be established to ensure that explainability standards are met, with rigor varying based on regulatory requirements and the potential impact on customers.
• Model Review and Deployment: E-23 requires independent model review to confirm that the model outputs are appropriately explainable and comply with performance expectations before the model impacts a consumer. Crucially, deployment might necessitate conditional approval subject to outcome monitoring to detect whether “fairness drift” occurs post-launch, ensuring that the model remains fair not just in the test environment, but in the real world.

By operationalizing these E-23 principles, insurers can ensure that the necessary evidence for the “Zurich Defense,” i.e., the proof of diligence and the testing of alternatives, is sufficient and documented as part of the standard, enterprise-wide control cycle.

The risk-based expansion can be illustrated through three tiers of operational reality:

• High compliance risk models do not always calculate a premium; they can act as gatekeepers to the quoting process itself. Consider an algorithmic point-of-sale fraud model that evaluates a digital footprint. If an applicant is scored as “high risk,” the system intentionally injects quoting friction, such as blocking the direct-to-consumer online rate and forcing a manual broker call. If this model relies on proxy variables that systematically flag specific minority cohorts, it could constitute a discriminatory barrier to entry for a mandatory financial product. Because these models dictate fundamental, equitable access to coverage, those resulting in systematic, disparate barriers require a full “reasonable and bona fide” assessment. Insurers should use human impact assessment tools like the HRIA to prove the fraud variables capture genuine, causal risk rather than acting as protected-class proxies and explicitly demonstrate a lack of less discriminatory screening alternatives.
• Medium compliance risk models prioritize convenience, creating an indirect fairness impact that requires lighter control. For example, a claims triage model that decides who gets instant approval versus standard handling creates a conduct risk if one group is systematically slowed down, but it does not accuse the customer of fraud. While these models may not demand an exhaustive assessment, they need sufficient pre-deployment proxy testing on historical data combined with automated post-deployment circuit breakers to ensure service level disparities remain within acceptable bounds.
• Low compliance risk models have remote or nonexistent human impact. Applying fairness testing here would be a misuse of resources. For example, actuarial reserving models operate on aggregate data pools to ensure solvency. While crucial for financial stability, they do not make individual decisions about consumers. For these models, impact assessment tools like the HRIA are non-applicable. The focus remains on the traditional pillars of performance and stability. By explicitly categorizing these as low compliance risk that are subject only to light inventory requirements, the insurer demonstrates the “proportionality” required by OSFI, preserving resources for the highest impact models.

1. Integrated life cycle management: The end-to-end model life cycle should explicitly integrate model compliance parameters for fair consumer outcomes.
2. Risk-based governance: Governance rigor should be proportional to the model compliance risk parameters such as bias, fairness, explainability, and human impact.
3. Evidentiary escalation versus risk acceptance: Market conduct violations cannot be formally accepted like financial and insurance risks. Models exhibiting unmitigated disparate impact should be escalated to senior management and legal counsel strictly to validate the “Zurich defense” prior to deployment.

The convergence of E-23 and FSRA requirements on fair consumer outcomes represents the current trajectory. Actuaries should review their model inventories not just for financial materiality, but also for compliance materiality. As the industry transitions into a regulatory environment that demands higher transparency, proactively operationalizing risk-based fairness provides the essential infrastructure to navigate these evolving standards effectively.

n the March/April 2026 AR, we covered the three ASOPs applicable to all actuarial services regardless of the practice area. They are ASOP 1 – Introductory Actuarial Standard of Practice, ASOP 23 – Data Quality, and ASOP 41 – Actuarial Communications. We also talked about the Applicability Guidelines (AGs). To recap, the AGs are published by the Council on Professionalism and Education of the American Academy of Actuaries and aim to help actuaries consider which ASOPs may provide guidance based on the scope of their role. These are not definitive statements of what generally accepted practices apply to a specific task and should not replace the actuary’s professional judgment. The AGs, which is an Excel file, can be found on the Academy’s website; just click on the Professionalism tab > Actuarial Standards of Practice > Applicability Guidelines. You can also access them through the Understanding Professionalism link.

In this article, we will focus on AG item 4.0 under the Casualty tab: “Expert Advice, Witness, and/or Testimony.” The only ASOP listed under this heading is ASOP 17 – Expert Testimony by Actuaries. This ASOP should be used in conjunction with any standards relating to the subject on which you provide expert advice.

ASOP 17 was originally adopted in 1991, revised in 2002, and further updated in 2011 and 2018. The latest version became effective for all expert testimony provided by the actuary on or after December 1, 2018.

The ASOP defines some key terms including “Actuarial Assumption,” “Actuarial Method,” “Expert,” “Principal,” and “Testimony.” It defines an “Expert” as “someone who is qualified under the evidentiary rules applicable in the forum to testify as an expert, whether explicitly or by acceptance of the actuary’s testimony. An actuary who has been engaged to testify, or permitted to testify, with the expectation that the actuary will ultimately qualify as an expert is treated as an expert for purposes of this standard, even if the actuary does not testify or is later determined to not qualify as an expert.”

“Testimony” is defined as “a communication of opinions or findings presented in the capacity of an expert witness at trial, in hearing or dispute resolution, in deposition, by declaration or affidavit or by any other means through which testimony may be received. Such testimony may be oral or written.”

An “Expert” may explain complex technical concepts, so they can be understood by the audience receiving the testimony, most of whom may not be actuaries. Even though actuaries may differ in their conclusions, “a mere difference of opinion between actuaries does not suggest that an actuary has failed to meet professional standards.”

An “Expert” will ordinarily work closely with the attorney or other representative of the “Principal” and may reasonably rely upon the advice, information, or instruction provided concerning the meaning and requirements of the rules of evidence or procedure and any other applicable rules. “[R]elying on such advice … is not in violation of this standard….” The actuary should disclose if they believe that a relevant law or regulation contains a material conflict with appropriate actuarial practices, subject to the requirements of the forum, including without limitation all rules of evidence and procedure.

Let’s look at some hypothetical scenarios where an actuary may be called to provide expert testimony; we note that any similarities to actual events are purely coincidental.

An actuary is employed as an expert witness by the U.S. Internal Revenue Service in a case where a captive is experiencing consistently low loss ratios. A captive that takes in premium but rarely, or never, pays out losses may indicate a lack of risk transfer; in this case, the captive acts to shift pretax dollars into an entity with a lower tax burden. The expert may be called in to review frequency and severity assumptions to determine whether the premium is reasonable. In this type of situation, the actuary may want to consider additional ASOPs, such as ASOP 53 (Estimating Future Costs for Prospective Property/Casualty Risk Transfer and Risk Retention) and ASOP 38 (Catastrophe Modeling), when performing their assessment.

Another example is a case of arbitration between two insurers, where one has purchased a subsidiary from the other. In this case, the subsidiary has experienced a deterioration in loss ratios since being purchased, and the purchasing insurer alleges that the subsidiary’s liabilities were materially understated. For this situation, expert testimony may involve a third-party, independent actuary performing an analysis of reserve estimates at the time of sale to determine whether the methods and assumptions used were outside of a reasonable range. The actuary may leverage ASOP 43 (Property/Casualty Unpaid Claim Estimates) and ASOP 23 (Data Quality) in their determination.

A third case where an actuary may provide expert testimony is in a regulatory rate hearing. The actuary may provide evidence that an insurer’s rate increase is excessive compared to its trends in loss, expense, and investment income. They may also opine about whether a company has a target profit that is excessive compared to the risk being insured. ASOP 13 (Trending Procedures in Property/Casualty Insurance) and ASOP 29 (Expense Provisions for Prospective Property/Casualty Risk Transfer and Risk Retention) may be cited by the actuary in their testimony.

Actuaries providing expert testimony may benefit from this general guidance on best practices:

1. Uphold professional independence and integrity: We, as actuaries, should maintain our independence from the client and avoid being an advocate for a specific side or outcome. We should be honest and not let pressure influence our conclusions; sometimes, this may include turning down assignments. Upholding actuarial professional integrity should always be a priority.
2. Documentation and technical rigor: Be extremely detailed and meticulous with your documentation. Try to anticipate what the opposing side may challenge, even the smallest details. It’s also not uncommon that you rely on other experts (e.g. catastrophe modelers, claims experts, or attorneys), though this reliance should be disclosed.
3. Anticipate the adversarial nature of the process: The environment is inherently adversarial, however most matters are resolved in arbitration rather than trials, and opposing sides can often come to a middle ground where everyone agrees and compromises are reached. Cross-examinations in depositions and trials are also adversarial in nature and require careful preparation with attorneys.
4. Communication and audience awareness: Very often, we will not be delivering our findings to experts. We must understand the background of our audience and assess their level of knowledge. Whether we are presenting our findings to judges, juries, or arbitrators, we must be able to communicate clearly.

Most of this general guidance can be applied to our daily work, hopefully except for presenting in an adversarial and disputed context. You may refer to ASOP 17 for additional guidance on hypothetical questions, cross-examination, and other related topics.

Understanding ASOP 17 will help make the expert witness process clearer, more consistent, and more professional. This is just an example of how to use the Applicability Guidelines for a specific Description of Assignment. There are six more major categories that we encourage you to explore and consider how each aligns with your practice.

When was the last time you referred to the Applicability Guidelines? We want to hear your thoughts at ar@casact.org.

erman actuary Paul Louis Riebesell proposed the popular “Riebesell form” for increased limit factors (ILFs) in the 1930s. It is still commonly used in the pricing of liability insurance and reinsurance around the world because it is convenient to be applied in practice and its parameter is easy to estimate. However, it is found that the Riebesell form of ILFs seem too heavy-tailed when they are utilized for some liability insurance lines in certain insurance markets. Here we propose a “modified Riebesell form” for ILFs that could fit the distribution of ILFs better in certain scenarios.

Loss Cost(Increased Limit) = Loss Cost(Basic Limit) ^*ILF((Increased Limit)/(Basic Limit)).

The essence of ILFs is to quantify the multiplicative relationship between the loss cost of basic limit and loss costs of different policy limits. Its definition can be formally given as follows:

ILF(M) = ILF((Increased Limit) / (Basic Limit)) = LAS(Increased Limit) / LAS(Basic Limit),

where M is the multiple between the increased limit and the basic limit, while LAS stands for Limited Average Severity defined as:

LAS(Limit) = E[min(Loss, Limit)] = ∫₀^LimitL * f(L)dL + Limit * [1 – F(Limit)].

Here, f(L) and F(L) are the probability density function and the cumulative distribution function (CDF) of the loss, respectively. In other words, the LAS for a given limit is the expected value of severity capped at the given policy limit.

ILF(M) = r^log₂M,

where r is the Riebesell factor and M is the multiple between the increased limit and the basic limit as defined above.

The Riebesell factor r has a convenient property in the practice of liability insurance pricing. It is the relativity for the loss cost of two times the basic limit divided by the loss cost of the basic limit, and it is also equal to the relativity for the loss cost of four times the basic limit divided by the loss cost of two times the basic limit, and so on. Therefore, if the Riebesell form works well in practice, we can easily obtain the Riebesell factor by dividing the loss cost of two times the basic limit by the loss cost of the basic limit. The Riebesell form may be quite suitable for some heavy-tailed liability risks, such as the product liability line in the U.S.

However, for some other liability risks that are not so heavy-tailed, such as general liability insurance in China, the Riebesell form often does not work well. It is often identified that the relativity for the loss cost of four times the basic limit divided by the loss cost of two times the basic limit is smaller than that of the relativity for the loss cost of two times the basic limit divided by the loss cost of the basic limit. As well, the relativity for the loss cost of eight times the basic limit divided by the loss cost of four times the basic limit is usually smaller than that of the relativity for the loss cost of four times the basic limit divided by the loss cost of two times the basic limit, and so on. The exact rate of ILF decay may depend on different markets’ litigation environments and how quickly liability claims escalate through towers of coverage — and the Riebesell form is too inflexible to reflect this.

ILF(M) = M^s ,

where s=log² _r. This result follows from rearranging and rebasing terms in the formula from the previous section in the following manner:

ILF(M) = r^log₂M = (r^{(lnM / ln2)}) = (r^{(1 / ln2)})^lnM
= (e^{(ln r · (1 / ln2))})^lnM = (e^{((lnM · ln r) / ln2)})
= (e^lnM)^{(ln r / ln2)}
= M^{(ln r / ln2)} = M^log₂r = M^s

In order for the original Riebesell form to be applied, the loss cost of liability insurance must be heavy-tailed enough to satisfy the CDF:

F(x) = 1 – a ^* x^{s – 1},

where s must be less than 1 (that is, r <2) and x must be greater than a^{1 / (1 − s)} for the purpose of F(x) being an effective CDF¹. It could be proven that the expected value for this CDF does not exist. But in practice it is found that the above CDF is too heavy-tailed for some liability insurance products. Gary Venter identified this problem in one of his articles², in which he regarded the above CDF as a kind of Pareto distribution with the shape parameter less than one. That kind of Pareto distribution is too heavy-tailed for some liability insurance products.

ILF(M) = r^(log₂M)α.

The modified Riebesell form has two parameters in which the parameter α controls the tail shape. Usually, α is less than one. Generally speaking, the original Riebesell form is a special case of the modified Riebesell form with the parameter α = 1. Under the modified Riebesell form, the tail of the increased limit factor distribution turns thinner as α decreases, as shown in Figure 1.

1.403 = 1.157^2.322 = 1.157^(log₂5)1.0
1.172 = 1.157^1.088 = 1.157^(log₂5)0.1

More information on selection of r = 1.157 is presented in the next section.

For illustration, we execute both approaches and compare their results to empirical ILFs for a simulated portfolio of Chinese general liability losses. For the original Riebesell form, we directly use the empirical ILF of two times basic limit as the estimate of the parameter r, which is 1.157 (which is the same value of r used to produce the curves in Figure 1). If we attempt to minimize the loss function of the mean squared error (MSE) for the modified Riebesell form, we obtain a fitted parameter α as 0.238.

ila, a lead software architect at an AI research lab, is debugging a self-modifying neural-network training script. Every second, the remaining unreviewed portion of the code grows by exactly 1% of its current length as new training data continuously streams in. Lila can examine and fix code at a constant rate of 100 lines per second. When she sits down to begin, there are exactly 1,000,000 lines left to review.

Will Lila ever finish debugging the entire script? If so, exactly how many seconds will it take her?

No one was cheated, because we can think of an identical situation where everyone in the circle agrees to clear their mutual debt. Ex: Alice owes $100 to Bob and Charlie owes Alice $100, so Alice can clear her debt by having Charlie owe Bob $100. Everyone does this until there are two people left who owe each other $100; they agree to clear it.

Now Alice wins $50 from a scratch off. The end result is the same—Alice keeps $50, no one has debt.