Breaking the Mold

Photograph: Frank K. Aieollo

Q: Why did you become an actuary? What attracted you to the actuarial profession?

A: I have loved math my entire life. I pursued a degree in mathematics at the University of Pennsylvania, with the intent of becoming an actuary or heading to Wall Street. I decided to try the actuarial profession first because I thought the hours would be more manageable. As an actuarial consultant at a powerhouse firm in the 1990s, the hours weren’t much better—but I loved it. I worked hard, studied incessantly and developed lifelong friendships with my colleagues. We solved complex problems for our customers, while supporting each other completely. We learned quickly and deeply. I was surrounded by positive, strong leadership that fostered creativity, innovation, experimentation and a deep-seated work ethic.

I think being an actuary is the greatest profession in the world. Even with a background in life, I was able to succeed across the property and casualty (P&C) space. I firmly believe the opportunities for actuaries are vast and growing, and we can move into any area that needs bright, quantitatively-minded innovators to create solutions to challenging business problems.

Q: How did your professional experience lead you to a career that is somewhat less traditional?

A: I’ve been an actuary for more than two decades, yet I’ve often been told that I “don’t seem like an actuary.” While I find this curious, it’s true that I tend to push the boundaries of traditional actuarial work.

Early in my career, I discovered an Employee Retirement Income Security Act (ERISA) nondiscrimination issue that led to the building of a new defined benefit plan for a global oil company’s service station employees. Besides the actuarial work required to build the plan—formula development, valuation, minimum required and maximum tax-deductible contributions—I wrote the participant communications, the technical specifications and the Summary Plan Description. My cubicle was the “customer service center,” and I spent months taking calls from participants, while simultaneously completing my “real” actuarial work. This was a beautiful, simple lesson that actuaries need not be pigeonholed into a typical role. Find a problem? Well, solve it.

As my career has unfolded, I’ve found myself repeatedly revisiting this lesson. I often move beyond the typical role to “fix” nonactuarial challenges, like optimizing home inspection schedules through behavioral modeling, improving loss prevention efforts through operations management and reducing expenses with sophisticated staffing and claims models. Actuaries often are uniquely positioned to apply their quantitative acumen beyond the traditional pricing and valuation roles. You just need to see, and grab, the opportunities as they present.

Q: How did you segue into work in predictive analytics?

A: I had been hired to build out actuarial reporting at a large global insurance company. Shortly after I started, I learned the company was experiencing a difficult implementation of a new P&C underwriting risk selection predictive model. Namely, the underwriters refused to use it, believing the model to be flawed. Having outsourced the build, no one internally understood the underlying data and assumptions; how the scores were calculated; or how to interpret results, modify variables and revise business rules. The company was facing a big loss if this model failed.

I volunteered to fix it. I had no predictive modeling background whatsoever—I didn’t even know what a predictive model was. But my consulting experience taught me how to learn anything fast, and with supplemental training from expert consultants in the P&C modeling space, I quickly got to work.

Q: How did you learn the tools and techniques of modeling?

A: I credit my manager for believing in me, and my company for investing in me. They hired a well-known consulting firm to give me a personal crash course in modeling in just a few days. I drank it all in. When I returned to work the following week, I dug into that risk selection model. I read the specs, calculated formulae and crunched spreadsheets until I thoroughly understood the workings of the model. I talked to the underwriters about their concerns, the actuaries who provided the requested inputs and the consultants who built it. At that point, I could understand and explain the technical aspects of an existing model.

It turned out the underwriters were right—there was an assumption error in the underlying data that skewed the interpretation of results. I removed the questionable variable, recalibrated the model and worked cross-functionally to implement the now-acceptable solution (facilitating the building of the scoring engine and user interface, and developing business rules and outcome reports). It was a big win for our organization in a relatively short time frame.

Wanting to push this concept further, I campaigned to improve our auto book’s pricing with predictive modeling techniques. But I didn’t know how to build a model, and neither did my small team. I contracted a work-share arrangement with another top consulting firm, agreeing that they would teach my team and me while we built together. We learned how to select variables for inclusion, test for interactions, refine splines and bins, and validate the models. I learned how to understand competitive market positioning using predictive modeling techniques, and when to consult with subject-matter experts based on output.

The rest I learned through observation and execution, and often just used good sense.

Q: Where do you think the greatest opportunities exist for actuaries in predictive analytics?

A: We need leaders. Predictive analytics within the life space is in dire need of leaders who understand the complete modeling process, and can build capabilities and lead teams. Currently, there are many great modelers without the necessary leadership and visionary skills, and many great leaders without modeling skills. Our leaders don’t need to actually build models (personally, I haven’t built one myself since that original auto book model), but they need to understand the details and translate the language.

Q: Why do you think some companies struggle to develop predictive analytics capabilities?

A: Building a great model doesn’t impact a business; implementing the right model does. Some typical reasons that companies—and actuaries—may struggle with predictive analytics efforts are:

• Few leaders in the industry understand how to build the capability. Building a capability is more than hiring modelers. It also includes understanding techniques, technology and data privacy; identifying the right opportunities; and influencing change.
• Building models that don’t solve a problem. It is actually quite common to see models built because there is data, not necessarily because there is a problem. The problems to be solved should be agreed upon by the business, not simply supposed by the data scientists.
• Building models that are not implementable given the current technological environment. It is critical to consult your IT experts frequently. Once during a model build, my team was advised that using more than two external variables would freeze our systems. That was great advice to hear on the outset.
• Not seeing the big picture. I once encountered a profitability model (based on loss ratios) mid-build. I quickly suspended this model, as the actuaries simultaneously were rebuilding the underlying pricing formula. That profitability model would have been obsolete before it was ever implemented. Modeling activities should complement other business priorities, not exist in a vacuum.
• Engaging in a power struggle. Avoid the “us” and “them” when building a team. Solicit expert feedback from your business partners throughout the build. It will result in a better model, and your partners will be invested in its use.

We can build great models all day long, but if they aren’t actually implemented and used to improve decision-making, then we haven’t really made a dent in the solution after all.

Q: What kinds of challenges can actuaries solve using data analytics?

A: Actuaries can expand their reach and improve their work product by supplementing traditional methods with predictive analytics capabilities, and by solving problems not typically considered actuarial in nature.

The opportunities are vast: Price optimization. Lifetime value and retention models. Price elasticity. Risk selection optimization. Predictive underwriting. Enrollment optimization. Target marketing. Understanding drivers of policyholder behavior. Propensity to buy. Lapse and churn analysis. Likelihood to bind. Understanding market changes on customer behavior. Market segmentation. Broker segmentation. Finding hidden pockets of the population to develop niche products … I could keep going—the opportunities are endless.

As head of pricing, I want to do everything I can to price my risks accurately. If I can use nontraditional data sources and techniques to better define any or all of the above, then we’ll continue to keep the better risks and thank the competition for taking the rest.

Q: What advice do you have for actuaries who may wish to expand their reach?

A: Predictive analytics is an incredibly interesting area, but there are many opportunities beyond modeling in which innovative actuaries can apply their expertise and harness their skills.

• Always be a little bit uncomfortable. Stretch yourself every day. When exams are done, find other opportunities to learn. I recently completed my MBA at the Massachusetts Institute of Technology, while working full time. While completely life-consuming, I learned more than I ever knew was possible. It is like “drinking from a fire hose.”
• Learn to walk before you run. Start small. Build a model you can complete in a few months that doesn’t cost millions to implement. See modest results. Then go bigger.
• Relationships matter. Whenever possible, attend meetings in person instead of over the phone. Meet actuaries both inside and outside of your organization. The opportunities and knowledge you will gain by growing your network are vast.
• Believe in yourself and in your profession. Actuaries shouldn’t just be looking for a bigger piece of pie. We should be creating a bigger pie. We are uniquely positioned to do anything. Tell the world.
• Listen. Think. Take a risk. Remember, “risk is opportunity.”