Technological innovation is disrupting the traditional insurance company business model and the professions that serve the industry. Not surprisingly, insurance professionals of all disciplines are finding they must adjust to the changes to remain relevant. Our profession is not immune from this disruption.
While the historic business model is based on intuition, experience and judgment, the emerging one is driven by coordinated intra-organization data and analytics made possible by three technological advancements.
The first is that data storage and computation costs are much more affordable than in the past. This is due to innovations such as cloud storage and computation.
Second, there’s more data availability than ever. Our digitally mediated lives leave breadcrumbs of data including: who we know; how we drive, exercise and sleep; and, what we buy, eat and read. These and other data bits are providing market segmentation insights, determining premium and more.
Third, the modern analytical applications and tools have become very powerful and handle a seemingly infinite amount of data, even on a transactional level. Gone are my old days, working with summary data sets, which was the best actuaries could do given the system limitations. In fact, the models are also changing from descriptive to predictive and ultimately, to prescriptive.
Analytics Changing the Actuarial Profession
Actuaries have been the insurance industry’s original data scientists and innovators of insurance predictive modeling since the profession’s beginning. The early predictive models developed by actuaries were, of course, constrained by available data and technologies. Here’s a quick timeline:
- In 1880, life actuaries realized that attained age was predictive of life expectancy. (While this seems obvious today, it was actually controversial at the time.)
- When states began adopting workers’ compensation laws more than 100 years ago, actuaries determined in 1915 that occupation was predictive of claim costs, developing work classifications and data systems to predict variations in costs.
- By 1963, age, gender, marital status and vehicle use were already reliable predictors of auto accident costs. Soon after, the first minimum bias multi-variable rating plan was developed in 1965.
- The American Insurance Association, in 1975, sponsored development of an industry financial database, including carriers that had failed; using that database, actuaries helped to develop a linear discriminant model to predict future insolvencies.
- During the early 1990s, innovative companies found a relationship between individual credit scores and accident predictability.
- Ten years later, predictive modeling had already taken root in other insurance lines including homeowners and small commercial.
From the timeline above, what is clear is that the actuarial role in building predictive models is a constant; what has changed is the technology.
While actuaries are the original architects of insurance predictive modeling, data scientists and statisticians have played a significant role in expanding modern predictive analytics. In a sense, at least in the context of insurance, the data scientists may be trying to reinvent the actuarial profession.
Actuarial employers have been telling CAS leadership that there is a talent shortage of professionals who offer three critical skills: data science, modern analytics skills and deep knowledge of the insurance industry. There are certainly some casualty actuaries who have all three skill sets. These individuals have often been working quietly behind the scenes to develop innovative pricing and risk selection models in auto and homeowners, and have been leading efforts to expand predictive modeling in other areas.
What is most important is that, candidly, employers are telling us that actuaries lack expertise in the first two skill sets. The advanced degree data scientists offer the first two skill sets but lack what actuaries have: the understanding necessary to apply models correctly.
For now, insurers are building teams from both disciplines to satisfy their needs, but there is a lag time in seeing results because it takes a lot of training to get both professions up to speed. Employers would prefer to hire professionals who embody all three skill sets — the center of the Venn diagram.
The sooner quantitative professionals satisfy their skill set deficiencies, the better positioned they will be for success in the emerging business model.
Responding to the Need
The CAS is responding to that need for all quantitative professionals to have these three skill sets.
First, we are aggressively pursuing changes to the actuarial curriculum to add topics on data management and contemporary analytics techniques to better align actuaries with current and future insurance company needs. The CAS has already introduced an entirely new statistics exam focused on the foundational material that underlies modern predictive analytics.
Further additions to the syllabus are under development, e.g., to address data management. By upgrading coursework and exams to be more focused on analytics, the CAS is already providing actuaries a path forward into the center of the Venn diagram.
Second, we have substantially increased our continuing education directed towards data and analytics. In 2015, the CAS offered more than 160 hours of continuing education directed towards data/analytics topics. In 2016, we are on track to exceed that figure.
Third, and perhaps most importantly, we announced the formation of a wholly owned subsidiary, The CAS Institute, at our annual meeting last November.
Informally called iCAS, the program will offer a separate credential in data science and predictive analytics. Curriculum development for this credential is nearly complete and will include requirements in the three skill sets later this year.
Employers verify the need for a credential such as the one to be offered by iCAS because it offers assurance that potential hires have a baseline level of the three skills sets. Unlike traditional actuarial credentials, the iCAS credential is highly recommended but optional.
Currently under development, the first credentials will focus on predictive analytics and data science. Over time, iCAS will be offering training in catastrophe model analytics, capital modeling and quantitative reinsurance analysis. To learn more, please visit http://bit.ly/266bO4W.
Since data scientists are also interested in being in the center of the Venn diagram, iCAS is available to them as well. While some believe that data scientists are a threat to the actuarial profession, I do not see a “turf war” between both professions. Instead, I believe iCAS is one way to invite data scientists into a bigger CAS tent.
I am excited about iCAS because it will provide an opportunity for actuaries to grow into the profession’s present and future. I hope you are too.
