Long-Term Growth Strategy
Looking ahead to our futureJUNE 2020
This last article in the Annual Report series looks at the evolution of the Society of Actuaries (SOA) long-term growth strategy and related initiatives. The insurance industry continues to experience unprecedented change due to advances in areas such as artificial intelligence, data analytics and InsurTech. With related shifts in the actuarial profession expected to occur, in 2019, the SOA Board of Directors began an in-depth study of these trends to determine how they will affect insurance markets and the role of actuaries. The Board will develop a new long-term growth strategy based on the findings from this study.
Q: We’re going to cover a few different areas within the topic of long-term growth strategy. To kick things off, let’s talk about how AI is transforming work for professions in general and specifically for the actuarial profession.
Christine Hofbeck: Let’s start by explaining what artificial intelligence (AI) is so we’re all on the same page. In simple terms, AI is the ability of a computer to accept and process information in a similar way to a human brain. It involves perception, problem solving, reasoning and learning. The computer perceives or takes in information, processes that information appropriately, and continues to learn as it processes. AI is the ability of computers to do tasks that used to require a human brain.
So how does AI transform work for actuaries and other professions? When you think about actuarial work, what we do for basically all or most of our jobs is use historical data to calculate a prediction. Then we make a judgment on what to do with that prediction. Going forward—for actuaries as well as other professions—we expect that machines are going to be doing more of the prediction piece, which will leave the creativity and judgment pieces for us humans. But this goes beyond divvying up work, because with the advent of artificial intelligence, both the nature of risk and our understanding of risk are changing, so we’re starting to see a more interconnected world. Climate change, pandemics, and other external events are introducing new risks. AI and other advances are giving us new insight on those risks, with new data and more of it. This makes some risks harder to insure in traditional ways, but it also creates opportunities for innovation.
Right now, we’re starting to see both a need and an opportunity to transform insurance. Our challenge as actuaries will be to think more broadly about what insurance is. What does it mean to measure and manage risk? It’s a big opportunity for us if we have the right skills to meet the challenge. We need to be comfortable with learning to innovate, pushing the envelope, and transforming, adapting, communicating, understanding and influencing those around us.
Q: Tim, given your work with and role on the Board over the past couple of years, would you like to add anything from that perspective?
Tim Rozar: Christine’s definition is spot-on. AI extends well beyond our traditional concept of prediction or predictive analytics. Whenever I consider what this means for actuaries, it is about finding the right balance: What is the role of the computer? What is the role of the expert? How do they work together? Are they in competition or collaboration with each other?
Think of how spreadsheets revolutionized actuarial practice however many years ago. They could have been viewed as a threat, but instead actuaries became experts and harnessed them as an additional tool in their toolbox. I think that’s the same way we should view artificial intelligence and other forms of advanced analytics. I think one of the risks facing actuaries is that someone else, some other profession, might do that in a better, more effective way than we do. These tools do require us to develop a new skill set or a new set of capabilities in order to use them well, but they could advance the landscape of actuarial practice in profound ways. We bring the domain expertise to synthesize and contextualize the information and predictions that these tools enable.
To me, though, one of the most important parts of this is that our responsibilities don’t end there. We must also provide a filter on appropriateness, on the ethical or socially responsible use of these tools. Predictions made by a machine are paradoxically “dumb” in that they may be very accurate from one point of view but may be completely inappropriate in a broader sense, whether that is marketing, risk segmentation, privacy, or whatever. That’s where the actuarial profession can play an important role, making sure we’re thinking it through, that we’re responsible stewards of the industry and that we’re not blindly following bad models.
Q: That’s a great segue into how the nature of risk and our understanding of risk are changing. Jennifer, how is the SOA addressing these types of challenges?
Jennifer Haid: As you say, our understanding of risk is changing, because the nature of risk is changing. As an example, I can no longer define the mortality of the woman who lives on a farm outside Toronto by her age, gender and smoking status. I need to consider her location, her socio-economic markers, the broader macroeconomic landscape. The risk profile of an individual is uniquely linked to her environment and circumstance.
All of the really cool technology that is being developed, has the ability to help us as actuaries understand these more complex ecosystems and to use that insight to construct products that protect the financial outcomes of people and corporations that participate in these ecosystems.
These advancements in technology foreshadow a world where actuaries will not be required to perform the calculations and basic work that machines can now do on our behalf. Instead, our value will be in working with the output of that process to answer key questions like – does the output make sense, what do the results mean, what are the implications for pricing, product design, risk management, etc. We will continue to be called on as interpreters — to present complex information to key decision makers within our organizations so we can develop the next generation of risk protection products.
The SOA is committed to prepare actuaries —through the examination process and through the lifelong learning curriculum— for our evolving roles. There are opportunities to upgrade technological skills, ‘just in time’ training for key regulatory changes, seminars on key market innovations, and opportunities to enhance communication skills through volunteer activities. I’m taking advantage of as much of it as I can as I prepare myself to respond to these changes
Q: As AI provides new insights, what opportunities is the SOA exploring?
Andy Rallis: The SOA has an opportunity to provide the appropriate training for our candidates and members. We are continuing to evaluate our education and syllabus to make sure that we’re on top of all the latest developments and that the training we provide is relevant to the work that actuaries are doing.
If you look at the technology we were using in the past to predict mortality, we had these precalculated calculations called commutation functions. The actuaries who enter the field today would probably be astounded to see those thick books of thousands and thousands of numbers that were precalculated just so that we could predict an age when somebody might die. Keep in mind those were the tools we had at the time. Today we have tools that are orders of magnitude more sophisticated, and it’s imperative that we make sure our actuaries are training in the skills needed to use these kinds of tools. As the technology evolves, the actuarial skill set has to evolve to accommodate it. That doesn’t mean that the end goal of making predictions has changed. We are still in the business of making predictions; it’s just that the range of information and the way we analyze it has evolved quite a bit. So education is one of those areas of opportunity.
Obviously we have roles to play in research and developing techniques. And in researching, make sure that we’ve accounted for biases in the data, things that are very hard to detect. But we have to make sure that we’re serving the public interest. So when we’re analyzing all of the data from the new sources we’re exposed to, we don’t embed in them things that are not good for society. I think that’s an avenue of research too.
Hofbeck: I just want to underscore what Andy said about how things change. One of the challenges we face as actuaries is that at times evolution can be scary and confusing. So we look at our history as a profession over the last 100 years and see that we’ve been the experts at the forefront of prediction. Even though learning new skills or trying to embrace artificial intelligence is scary, even hard work, it’s going to be necessary if we want to remain the experts at the forefront of prediction. The SOA is not overlooking the fact that sometimes change is hard, but it’s also necessary.
Greg Heidrich: When you’re in a world that’s changing as dramatically as ours is, and you’re seeing risks that you didn’t see before or didn’t recognize as being interconnected in the ways they’re proving to be today, it’s really critical to have a good understanding of how the business works and the context in which the business is operating. It is importance to emphasize all of the contextual knowledge that actuaries are bringing to the fields and the businesses they serve. Historically, actuaries have brought that knowledge to the industries in which they’ve worked, principally the insurance industry and employee benefits. So even as we respond to these new tools and capabilities that AI is bringing us, we have to keep that strong contextual knowledge too.
Haid: We’ve traditionally operated within our own specific communities, so a pension actuary knows the techniques and tools related to pensions, an insurance actuary knows the same related to the products they work with, etc. However, each of us is applying a similar toolkit to our specific area of practice and because of that, we are able to extrapolate the toolkits we already have to new problems. The more we leverage each other’s knowledge, the better success we’ll have adapting to the changing nature of risk.
Heidrich: It also emphasizes the importance of adaptability. It’s not that the world isn’t going to use specialists; it does need specialists and actuaries are specialists in particular areas. But the world is going to demand adaptability, and all of our work with employers—the research we’ve been looking at through this long-term growth strategy project has told us that high adaptability is a skill that’s going to be in great demand in the future, and actuaries need to cultivate that.
Q: We were talking earlier about how there’s so much more data out there than before, how everything tracks in real time, and the idea of computation becoming cheaper and able to do more calculations. Can we talk a little bit more about those aspects?
Rallis: With traditional actuarial work grounded in really traditional statistics—in the sense of having a small amount of data and needing to make extrapolations from that—the question is, how many deaths do you need for something to be credible?
In a way, we’re faced with exactly the opposite kind of problem today. We have access to vast amounts of data—sensors in our automobiles, our cell phones, maybe even our clothing—and what we’re faced with is the challenge of finding the information that’s hidden in that vast amount of data and deriving meaning from it. This requires new techniques, which are really the predictive analytic techniques.
It’s the same objective of finding the maximum amount of information, but in one case you’re starting with a little bit of data, and in the other case you’re starting with a lot of data. Either way, you still have to turn it into information.
Haid: With that wealth of information, we no longer have to rely solely on historical data: we can now get data in real-time. For example, we can collect real-time information about a person’s activities, like how many steps they’re taking, whether they’re taking their medication on a regular basis, whether they’re filling their prescriptions. All of that information can help us assess risk in real time.
This also allows us to take on a more proactive role in the lives of the people and companies we serve. For example, if we’re collecting real-time data on the risks people or companies are taking, we can work with our clients to achieve healthier outcomes – whether on a personal level, because we understand the drivers of healthy life expectancies, or at the corporate level, because we understand best practices for workplace safety and cyber security. This enables us, as actuaries, to help influence our world so it’s a safer, healthier place for people to be. I think that’s a really powerful set of motivators for folks who are looking for meaning and how to make a social impact in today’s world.
Hofbeck: When we look to new sources of data or get creative about using new data to make our predictions as well as or better than we did yesterday—we also might see an improved customer experience, decreased expenses and streamlined operations. Traditionally when an insurer is going to write life insurance, they may send a nurse out to a person’s house to take fluid samples. It’s costly for the insurer; it takes time; it’s annoying for the customer to have to sit and wait for the appointment and then have to give blood and urine samples. Now we can use pharmacy data or even selfies (a photograph of a person’s face) to identify risk factors. These methods are far less invasive; they make the customer much happier during the process; they make the decision faster; they make the decision cheaper; you can get the premium in the door quicker; and we can make just as accurate or even more accurate predictions based on faster, cheaper data. There are just so many benefits of continuing to think outside the box on how we can use new forms of data, or more data in better ways.
Heidrich: I’m interested in your reactions and what you’re seeing given the current situation that we’re all in with the COVID-19 pandemic. For instance, there are tons of new models of people interacting and exchanging data. One is the explosion in telehealth or telemedicine delivery. I’m curious if you think that the current environment we’re operating in is accelerating those trends? Will it have any permanent impact on the kinds of trends you’ve been talking about? Or will things go back to how they were developing in the world before all this happened?
Rallis: We’re seeing the complexity of the data leading to very complex interactions with the models. I find it fascinating that the models have to adapt in real time. There’s a feedback mechanism between the models and the actions taken in the real world based on those models. Social distancing, the development of treatments, the movement of medical equipment between sites—they’re all predicted by models. Those models are then influencing what’s happening in the real world in real time. It’s very different from the way we normally build models, where we’ve studied data over a long historical period and are trying to predict trends in the future.
Rozar: True, we’re seeing a couple of things. One is a strong demand for real-time data with which to feed the models, and that data just hasn’t been there early on in the pandemic. There are lots of reasons for this; for example, the availability of testing was slow to develop in some markets and the definition of how to count cases, or even deaths, is inconsistent from country to country, or even city to city. But the models are only as good as the data that go into them. And then there’s the feedback loop. We’re trying to come up with predictions that will influence behavior, but the behavior then influences the accuracy of the prediction, so there’s this constant chase to figure out what measures are actually working and what is just coincidental. It makes it very difficult to project with any certainty into the future.
Hofbeck: Speaking of real-time data — a request came across my Twitter feed recently asking anyone who has contracted coronavirus to send recordings of their cough to whichever organization is collecting this data. The idea is to develop a model that diagnoses COVID-19 from a cough. This illustrates that we’re now actively crowd-sourcing needed data in real time. I thought that was pretty interesting.
Heidrich: I think that’s fascinating, Christine. One of the other things I think is happening is that with the plethora of data that are being created like this, a lot of people are literally are moving themselves into the expert phase. We’re all becoming our own diagnosticians. I know this is an issue that medical professions have dealt with a lot, people coming to them with a list of symptoms they’ve pulled off a website. With online studies like Christine’s example of coughing data, it’s just going to accelerate. And we have to expect that there will be something analogous happening in the actuarial world. I think actuaries have to be aware that their roles as experts won’t necessarily just be assumed in this new world. There are going to be so many experts and so much data that finding a way to articulate the value is going to be critically important.
Hofbeck: I agree with that perspective. Historically, actuaries have asked, “What data do we have? What predictions can we make from the data that we have?” Our new thinking can be, “What data can we get? What data can we create? What data can we crowd-source to use in this way?” If we circle back to some of the points Tim was making about the ethics of it all, then the other thing actuaries ask is, “What data can we get or create that also fits within our current regulatory environment?” We’re not just creating a proxy to circumvent current regulations. We’re still keeping those ethical boundaries in mind when we’re acquiring and creating new forms of data.
Rozar: What I’ve taken away from all of this—with a feeling of surprise and some sadness—is how polarized people’s opinions of scientific models have become based on whatever preconceived narrative that they align with. It puts objective scientific professions in a difficult position. Even if you could remove all sorts of unintended biases from the process of selecting data, developing models, and making predictions and recommendations, it’s not good enough. People have such strong opinions on what the results should be, and if the model doesn’t comport with those opinions, they will immediately doubt the credibility of a model. I think we could potentially extrapolate that to any number of other types of predictions or algorithms as well. That isn’t to say that people shouldn’t question models because, as we’ve discussed, there are lots of problems with blindly following models, but it is important that in the process of challenging the models we are looking at things that are objective and data-driven. In this way, I view actuaries, along with epidemiologists, virologists and so forth in the current crisis, as being independent arbiters of truth. We always have to strive to stay objective and follow the evidence.
Haid: One of the outcomes of our research is that we are moving away from being “calculation agents.” In the new world, our value will be in figuring out if the predictions make sense, and how to use them. We need to be able to understand our own biases and take the perspective of multiple stakeholders to make decisions from a holistic frame of reference, instead of through the narrow lens of our own experience. This is a great opportunity for us to take a leadership role.
Rozar: One of the SOA’s big strengths is that we’re seen as objective and independent from any agenda; our brand and reputation have sort of a gold star of independence and that brings a cachet to anything we do, from education to public policy research. That’s a valuable resource, and it’s one we have to be very, very careful with and hold dear. When it comes to the research that the SOA is putting out right now on coronavirus, or even going back a few years to public pensions and other topics, the fact that something is the SOA’s opinion hopefully cuts through some of that political haze. Then people can say, “Okay, this information is coming out without any sort of agenda. It’s data-driven, it’s evidence-based, and it’s been developed by experts.”
Q: How has the Board been addressing these new issues of AI and technology and change. How can we prepare the profession for all sorts of changes, things like this pandemic, technological advancements and new challenges that are thrust on actuaries?
Rozar: A lot of this is work that’s a continuous evolution for the SOA to move the profession forward, whether that is in the research portfolio, the educational syllabus, or the content and delivery of professional development. To that end, many of the strategic initiatives that we’ve launched in the past few years, including predictive analytics, InsurTech, and international expansion, are already in execution mode and starting to bear fruit. But I think the Long-Term Growth Strategy we’ve been involved with is (1) a little more holistic in scope and (2) maybe more importantly, longer-term in focus. It’s not just about the next one to three years or even three to five years. It’s looking around the corner at all of the trends we’ve been talking about—artificial intelligence, the future nature of work, emerging risks, internationalization and so on. And it’s not just about how do we do what we’ve been doing better. It’s about challenging the very nature of what we do. It’s asking: What do these trends mean for the profession? What are the existential risks? Where are the untapped opportunities?
As part of this process, we’ve had interviews with staff, the Board, volunteers, external advisors, employers, regulators, and other stakeholders. Then we’ve synthesized all that information and overlaid it with research on all of these external trends, trying to come up with what they mean for the SOA and what they mean for the actuarial profession so we can develop a range of potential scenarios and responses that will inform SOA’s longer-term strategy.
One of the key insights from the work so far is the recognition that there’s strong competition for the work that actuaries can do and for the potential candidates who might become actuaries, so we have it on both sides. To be compelling to the top-level potential candidates coming out of university or even high school, who have to decide whether this is a profession they want explore, we will need to meet the evolving expectations of those young people in terms of what is important to them: What sort of career is fulfilling for them? What sort of work are they looking for? What sort of work environment? What sort of learning styles do they have? We have to build all that into our curriculum approach to education and into what employers expect from us.
On the other end, the competition is obviously, as Greg said, from both experts and hobbyists from different industries or professions who see actuaries making predictions and think, “I can make predictions. Let me have the data, and I can do something too.” In some ways that creates innovative, novel approaches that actuaries may not have thought of. In other cases, it may create unintended risks or naive approaches that bring risk or threats to the sustainability of the industry.
To wrap all that up, it’s about meeting the evolving expectations of candidates, employers and other stakeholders and building a profession that can respond to those needs.
Haid: Over the last few years we’ve known that our younger professionals want real-time education. They want it in bite-size formats, and they want it in ways they can easily consume at the time they need it. For example, do they need to understand a new set of regulations? Are they going to work in a new country? Are they operating in a new area of practice?
We also have a more geographically diverse membership than we’ve ever had before, and that comes with a responsibility to provide geographically appropriate learning content.
And now, as a result of the pandemic, we have to think about what we do in a world where we can’t actually provide in-person exams in one/several/all of our test centers. What happens if this persists for a year or more? That creates a sense of urgency behind the direction and momentum we already had (or knew we needed to have) because we are now trying to find a way to help our young professionals continue to make progress toward their credentials in a world where will be required to move online.
We need to quickly pivot to be able to provide just-in-time education to our members who are pivoting to model mortality or longevity in the context of a pandemic.
Before the current pandemic, we knew we needed to make changes to stay in step with the modernization of education delivery. Given the current challenge, the SOA is now committed to pivoting as quickly as is practical to continue to serve our members and our candidates.
Heidrich: One of the effects of COVID-19 so far is that it’s dramatically accelerating changes that were already happening in business models and ways of doing things. We knew long-term that we needed to move to much more online, modularized education. The fact that we’re trying to gather people together when every country in the world can effectively be shut down simultaneously highlights that need for change. We knew it was out there, but it’s coming a lot faster than expected.
Haid: I’ve honestly been really impressed with the speed at which the SOA has pivoted.
Hofbeck: And let’s remember that research is also a main component of the SOA. Our goal is to leverage research to create new value for current and new members and stakeholders. We’re always asking how we can lean on what we do best. We have actuaries working together from all areas to provide research that benefits society.
Haid: An example is the COVID-19 research report. It’s a multidisciplinary, thorough piece of work that is being updated weekly.
Q: We’ve talked about things being more virtual, and we’re doing more professional development offerings in virtual meetings. What can members and candidates expect in the near-term? What’s next for the SOA in terms of the long-term growth strategy?
Rallis: The environment has changed significantly in the recent past with both emerging technologies and the globalization of the profession, and that makes the consideration of a long-term growth strategy an imperative. Members should know that our Board is actively at work. We’re working with our stakeholders and outside advisors to gather input into what that strategy should look like, but the Board is actively considering it at this time, and we’ll be back to the membership as soon as we’ve finished.
Heidrich: I would characterize the work we’ve done over the past year as the deepest dive of an environmental scan than we’ve done in over a decade. We’ve gone out of our way to try to understand the changes that are happening in adult education, adult learning, AI and data analytics. We’ve tried to get a read on those changes, and we know they’re coming. The SOA and the Board are dedicated to understanding the environment in which our members are working. Also, we’re developing a set of programs and are making changes to the organization’s basic services that will respond to new developments. We may have a much more modularized education system that is responsive to members’ need to understand these trends, to develop their judgment and adaptability skills, and to improve their ability to compete and offer unique value in this environment. That’s in the basic education system, the pre-qualification education system, the continuing education system, and of course, research. So members should expect a variety of program changes and announcements that will carry us over a number of years in the future—all aimed at responding to these trends and developments we’re seeing.
We are convinced that actuaries have a very strong role to play in the future. It’s going to be a different role, in many cases, than they’ve played in the past, but it’s a very important role. We’re committed to preparing our members for the future.