Photograph: Chris Cone
Q: Tell us about your background. How did you decide to become an actuary?
A: For much of my career, I held chief actuary and/or CFO roles in the life and annuity segment of our industry. Thanks in part to finishing my FSA at a relatively young age and earning an MBA, I was on the fast track to senior management at a life insurance company. Then came the abrupt hairpin turn (some would say multicar pileup), and I emerged as a consultant in a small consulting firm where I remain to this day.
Keep in mind that if you are as old as I am, you did not grow up seeing actuary at the top of the desired job rankings. The profession was quite obscure. When I was in the eighth grade, I caddied at the local golf course for an insurance agent who asked me what I wanted to be when I grew up. I told him I didn’t know, but it would probably involve math, because I liked math. He said I should consider becoming an actuary. Like most eighth graders, I had no clue what an actuary was. My dad worked at a small life insurance company, and when I asked him about it, he became ecstatic—probably because he knew how rare actuaries were (this also shows my age). At that point, I began to gravitate toward the actuarial profession. With a few speed bumps on my path and some well-timed steering inputs, I eventually got there.
Q: How did your work history segue into your interest in predictive analytics?
A: When I graduated from Ball State’s actuarial program, I knew I was going back to school in two years to work on an MBA at Harvard, so my job search process was somewhat unusual. I really didn’t have a preference among life/health versus pension or company versus consulting, so I wanted to survey all of those different opportunities early on. I began in the individual life department of a large multiline carrier. After graduating from Harvard, I moved to pension consulting for a few years and then to life insurance consulting/investment banking. At that point, I committed to working at life insurance companies. Over the next 25 years, I worked as chief actuary and/or CFO for a number of life companies. Then in the early 2000s, I essentially worked myself out of a job and needed something to do. I started a consulting firm and have been an independent consultant ever since, except for a few years when I worked with an independent underwriter.
It was that underwriting work that brought predictive modeling to my attention. The independent life underwriting firm was accumulating data at a rapid rate. The firm hired me to help make heads or tails of it. At first, I followed a fairly traditional path of charting actual-to-expected (A/E) results by age, gender, smoking status and level of impairment, adjusting mortality assumptions to tighten the A/Es. Enlightenment came later, with the creation of a medical advisory board made up of prominent biostatisticians and epidemiologists who worked nearby. As our data grew, we enlisted their aid in furthering our analytical efforts. Under their tutelage, I learned about predictive analytics. I would say a collegial and/or mentoring environment is a key success factor in predictive analytics. This is analogous to having an experienced instructor riding with me on the racetrack to provide instantaneous feedback.
Q: As the CEO of your own company, what about your business fuels your fire? What keeps you up at night?
A: The most exciting element of my business is knowing that we (my firm) have never done the same thing the same way twice—ever. And I expect we never will. We look to improve with every iteration, even for repetitive tasks. Clients have come to expect us to raise the bar year after year, and I hope we are up to that challenge.
It’s like tracking my car at the Mid-Ohio Sports Car Course, where I have run well over 1,000 laps: From day-to-day and year-to-year, I want my last lap to be faster than my first.
There are many issues that vie for my attention at night, but they generally fall into two categories:
- Are we serving our clients’ interests in the best possible way?
- Am I providing my associates the best possible work experience?
Q: What skills positioned you for work in predictive analytics?
A: Back to my eighth-grade self: It’s important to enjoy the math. That doesn’t mean understand all the theory. Beyond that, the critical skills are curiosity, discipline (those who know me well are smirking), practicality and focus.
Q: What skills do actuaries bring to analytics that other professionals may not bring to the role?
A: Given the actuarial stereotypes, this may come as a surprise to some folks, but I would identify the actuary’s bias for action as a key element that differentiates us in predictive analytics. The academics who dominate the field are much more deliberate and theoretical, whether it be for research purposes or otherwise. My friends in academia have encouraged me to write papers solely focused on the relationship of atrial fibrillation and heart attacks, when that relationship, while important, was only a very small part of the project I was working on at the time.
Q: What advice do you have for people who may be interested in positions in predictive analytics?
A: Two things: Find that collegial/mentoring environment that I referenced, and learn on a real-world problem.
Q: What general advice would you give to someone who wants to start his or her own company?
A: First, embrace the internet. When I started my first company in 2003, the internet made it possible for me to leverage myself in ways I couldn’t have imagined in the 1980s or even the 1990s. Since then, it has become even more ubiquitous and enabling for entrepreneurs.
Second, let’s revisit that list of critical skills: curiosity, discipline, practicality and focus. There is a trap for companies, thinking every associate needs to possess all of the critical skills; not true! In an organization, all of the skills need to be covered, but expecting someone who is curious also to be disciplined is a tall order. On the track, the cars that excel speeding down the back straight are often not the fastest in the turns.
Third, you should be comfortable with never feeling comfortable—it is a binary life of either having too much business or not enough. You likely will not want to turn down any business because you don’t know when the next opportunity will come. Yes, it can be a real roller coaster existence.
When I am tracking my car, I cannot advance my driving skills unless I—carefully and in a controlled fashion—get to the edge of my comfort zone and maybe go a little further. It’s the price I pay if I want to get better.
Finally, because of all I have mentioned, it’s good to have the support of family, friends and fellow professionals. Thankfully, my wife of 34 years, Irene, has been a fantastic friend, counselor and confidant. Her patience and encouragement have contributed greatly to my success.
Q: How did you learn the tools and techniques of predictive modeling? What sparked your interest in this area?
A: It started with a real passion for statistics, which I developed while studying for the old, old Part Two of the fellowship syllabus. Once I started down the managerial track in my career, that passion became dormant as I looked to develop leadership skills to better equip myself to serve the associates in my chain of command. When I started my own practice, I was forced to become more hands-on and focus on developing skills that clients would value in the marketplace.
This theme continued as I became associated with the independent underwriter. As I looked at the traditional approach to fine-tuning underwriting debits and credits, I was not satisfied. We had reams of data specific to the individual impairment, but we only could find statistical significance when we aggregated the impairments and looked at the overall level of impairment. That meant we grouped all of the severely impaired folks together, regardless of whether they were severely impaired due to heart disease, cancer, cognitive issues or other things. Intuitively, that made no sense to me.
Thankfully, it made no sense to the epidemiologists and biostatisticians, either. They introduced us to the predictive modeling techniques they used to forecast life expectancy for kidney dialysis patients or survival rates of people who followed the Mediterranean diet. It was an involved, hands-on process that took some time. Fortunately, I also was able to learn from some talented, curious IT folks, who used their programming skills to build complex functions that allowed me to become much more efficient in my efforts.
On the track, there is no substitute for seat time, meaning actual driving versus classroom sessions (which are helpful, but never provide the equivalent experience). It’s the same for building a repertoire of predictive modeling skills.
Q: Tell our audience some things they may not know about predictive analytics.
A: I’ve learned so much about predictive modeling that goes against the grain of my instincts. It’s like the overwhelming urge to slam on the brakes when you feel the back end of the car start to break loose; in reality, the correct action generally is to apply more throttle.
First, please know that simple yet powerful and effective tools like odds ratios exist. It takes very little effort to master these tools, and they provide tremendous insights. Before doing any regression techniques, I often calculate odds ratios to help me get a feel for the data.
But there also are tools that produce complicated analyses. Although these require more effort to understand, they are very simple and quick to run, which can greatly leverage your efforts. For example, the Cox Proportional Hazards model is quite powerful and quick once it is mastered.
Successful use of predictive analytics requires you to start slow to finish fast. Data preparation is key. In fact, the preparation phase will or should take most of your time. The analogy on the racetrack is that it’s not the car that enters the turn fastest that ultimately wins; it’s the car that exits the turn fastest, which often means entering at a slower speed.
Finally, I was very pleased to discover that our industry data is superior! What we consider inadequate and statistically unreliable others are willing to write about in scholarly publications. This portends well for actuaries who want to expand the role of predictive analytics.
Q: Where do you see opportunities for actuaries in the predictive analytics arena?
A: Notice the excellent segue from the previous question. Because we have so much data to begin with and the available data universe is ever expanding, it’s up to actuaries to find ways to marry these two worlds and produce better forecasts. Life, health, pensions—the industry segment doesn’t matter, because the opportunities are all there.
Q: If you could turn back the clock, knowing all that you know now, would you choose the actuarial profession again? If yes, why?
A: There’s a loaded question! Yes, of course. The actuarial profession has been a great one for me. Looking back, there are some things I might consider a little differently than I did, however. For example, early on I was so concerned with career path and scoping out exactly the right career decisions; in reality, so much was outside of my control. We think we’re in the driver’s seat, but we may be in the passenger seat or even the back seat. I’d trade a well-crafted career plan for a deeper understanding at each stage of my development.
Because we are on the subject of the actuarial profession, let me add one more thing: There are great people in our profession—folks who go out of their way to help others. The Porsche Club of America has a saying—It’s not the cars; it’s the people. As far as I am concerned, regarding the actuarial profession—It’s not the numbers; it’s the people.
Q: How do you see the role that predictive analytics plays in the actuarial profession changing over the next five to 10 years?
A: We are entering the golden age of predictive analytics in the actuarial field. More and more data is becoming available, yet considerable judgment and discipline are required to properly analyze and interpret it. These kinds of challenges always have been in the purview of the actuary, and I see no reason why that should change. The old timers out there will remember one of my favorite quotes, attributable to John Ruskin, which was the SOA motto at one time: “The work of science is to substitute facts for appearances, and demonstrations for impressions.” To that, I will add, “Actuaries, start your engines, errr, computers!”