Photograph: Riku Foto
Q: Tell us about your background. How did you make the decision to become an actuary?
A: I was studying applied mathematics at Tatung University in Taipei, Taiwan. I fell in love with probability and statistics during my junior year. After finishing college, I came to the United States to pursue a graduate degree in actuarial science at Boston University. I enjoyed learning how to quantify the unknown financial risk of a contingent event. However, I wanted to contribute on the actuarial research front. Therefore, I went on to pursue a Ph.D. in mathematics and statistics, with an emphasis on credibility theory.
I started my actuarial academic career as an assistant professor at the University of Central Florida (UCF) in 2002 and received tenure as an associate professor in 2008. Unfortunately, UCF decided to terminate a very successful actuarial science program in 2009. That’s when I officially started my actuarial career in the insurance industry. I switched to the health care insurance industry and led a risk identification research project in preparation for health care reform in the United States. I am a passionate advocate for the use of predictive analytics in this post-reform market. Meanwhile, I also serve as an adjunct professor at the University of North Florida and am helping the department of Mathematics and Statistics to establish an actuarial science program.
Q: What skills positioned you for work in predictive analytics?
A: My trainings in mathematics, actuarial science, statistics, data mining, and academic teaching and researching all come together and help me grow in the predictive analytics field. My mind is always filled with new ideas. More important, my passions to learn the unknowns and to compete in analytics really motivate me in pursuing this nontraditional role in the health care insurance industry.
Q: How did your professional experience lead you to a career in predictive analytics?
A: The actuarial training gave me the ticket to get on the bus, and my professional experience helped me do well in this field. My first industry job was with an auto and home insurance company, where pricing by risk is allowed. The use of predictive modeling is the norm in the non-life insurance industry. Because of my strong modeling background, it’s a natural fit for me. Predictive modeling is still relatively new to health actuaries. I hope it will become a regular part of actuarial functions in the near future.
Q: How do you use predictive analytics in your job?
A: I have been involved in providing support for different areas. These efforts include:
- Designing and selling the right product to the right people, at the right time, for the right price.
- Selecting the right members for the right interventions.
- Quantifying the risk and identifying opportunities under the influence of risk adjustment methodology.
- Coordinating cross-functional strategies and identifying optimized solutions at the enterprise level.
Q: How did you learn the tools and techniques of modeling? What sparked your interest in this area?
A: I learned the tools and techniques of modeling at different stages. I learned how to go from zero to one in my Ph.D. studies. The actuarial control cycle guides me through my work in analytics. My academic teaching and researching experience makes me a better thinker and researcher.
I joined the health care insurance industry in 2010 to be a part of U.S. health care reform. I had an opportunity to develop high-performance risk models using advanced predictive modeling and data mining techniques to quantify the risk and to identify new opportunities in the post-reform marketplace. This is the moment that makes me a matured predictive modeler. It has demonstrated that advanced analytics can be used effectively to predict cost and classify new prospects without having access to historical claims and utilization data. I consider myself a very lucky person because I am sitting in the front seat and experiencing this changing tide.
Q: Where do you see opportunities for actuaries in the predictive analytics arena?
A: Facing the challenges of the post-reform era, many health insurance carriers began using big data analytics on all aspects of their decision-making processes. The goal is to be able to compete on analytics in this quickly changing environment. However, becoming an analytics competitor is not easy. It requires the belief from the leadership and a group of dedicated analytical professionals to solve emergent business problems that existing strategies cannot address in an evolving market. These problems include marketing for acquisition and retention, coding accuracy, revenue growth, health care disparity, payment integrity, quality of care, health care cost reduction and, ultimately, personalized care management.
In addition, predictive analytics comes with costs. There is the direct cost of devoting resources to develop and implement analytical models properly. In addition, there are hidden indirect costs of relying on analytical models of making decisions, such as the possible adverse consequences (including financial loss) of decisions based on models that are incorrect and/or misused. These consequences should be addressed by developing a comprehensive model risk management plan. The lack of coordination for cross-functional strategies is going to reduce the competitiveness of a company. Therefore, it is extremely important to establish a modeling center within the organization to provide optimized solutions and to monitor and quantify the potential risk at the enterprise level at the same time. I believe actuaries are well-equipped for these challenges.
Q: What are some challenges in working in predictive analytics?
A: The U.S. health care industry has gone through a dramatic change during the past several years. The business model has moved from a risk selection model to a risk management model. People can shop online for their health care coverage. Everyone is welcome! Annual and lifetime dollar limits, underwriting and the pre-existing conditions exclusion are all in the history book now.
However, there are so many unknowns. Who can bring in more revenue, given that the revenue is now closely connected to “coded” health status? Who will cost more? Who will be more profitable? Most important, how can we help people and communities achieve better health utilizing the most efficient approach? We wish we could have a crystal ball to answer all of these questions, but we don’t. The best alternative is the use of predictive analytics to turn messy data into useful information and generate actionable rules.
Predictive analytics is about precision targeting and winning. It’s the winning part that makes predictive analytics extremely challenging. And the winning ingredient is the search for interaction and nonlinear terms. There are very few people willing to shut up, sit down and do it. However, there are a lot of unqualified people providing mediocre analytics that will damage the predictive analytics brand name.
In health care, there is great potential to positively influence patient outcomes. Nevertheless, influencing behavior isn’t as easy as convincing someone to click a few buttons. There is a need to address privacy concerns and HIPAA regulations on the use of big data analytics. Actuaries, who abide by the code of professional conduct and follow a set of protocols to guide their decisions, can serve as gatekeepers to protect the general public.
Q: What advice do you have for people who may be interested in positions in predictive analytics?
A: The health care insurance environment will continue to become a more competitive and customer-oriented business. The universal approach will not satisfy all customers, especially when they have increased freedom of choice in purchasing health coverage. Ultimately, a health insurance company needs to become a health solution company.
My advice to the future generation of actuaries is to learn statistics and data mining as much as possible, have solid hands-on programming skills, follow the actuarial education program and come join the party. We are going to have a lot of fun on this journey.
Q: What skills do actuaries bring to analytics that other professionals may not?
A: Predictive analytics is about foreseeing the future. Pricing and reserving are two classic actuarial functions. Actuaries have proven to be very good at turning historical data into adequate capital obligation and premium rates for the future. Strong attention to detail, creativity and patience are very critical when competing in analytics. In addition, actuaries are good communicators, able to explain what we know to people who don’t know as much. Actuaries need to build collaborative relationships with all partners, because a model is useless if no one uses it.
Q: What are some of your best professional memories/experiences as an actuary that may inspire others to explore different actuarial paths?
A: I strongly believe the use of advanced analytics will grow exponentially in this post-reform marketplace.
When it comes to predictive analytics positions in health care, actuaries are losing out to data scientists and statisticians. Having experience in all three areas, I say they all have a similar toolbox, but they apply these tools to different areas to accomplish different goals. Actuaries are experts on predicting how much members will cost in the future based on the data-driven traditional actuarial per member per month (PMPM) metric. Data scientists use these tools to recognize patterns hidden in the data, which typically is massive, dynamic and dirty. And, the principal aim of traditional statistical analysis is making an inference about the parameter using confidence intervals and hypothesis tests. I believe actuaries are capable of all three.
At a time when companies in the health insurance industry offer similar products, many of the previous bases for profitability are no longer available. What’s left as a basis for competition is making the smartest business decision possible. This is exactly what an actuary is trained for, isn’t it? I love it because it is very challenging.
Q: What are some things our audience may not know about predictive analytics?
A: Descriptive analytics proclaims “a zero is zero”; predictive analytics, on the other hand, dictates that “a zero is not zero.” A person’s health status changes over time. Additionally, his or her driving behavior could be very different a year from now. Predictive analytics is a methodology that allows us to increase the odds of guessing correctly for a future event so we can prevent it from happening. It is the analysis of all kinds of data using advanced quantitative techniques to gain insight that traditional query and reporting are unlikely to identify. For example, “How do I deliver personalized care to those at risk?” A wrong model is worse than no model. Predictive analytics is not a part-time job. We need more qualified, credentialed actuaries playing a leading role in this area.
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: Yes, I have no doubt in my mind I would choose the actuarial profession again. Looking back, it has been a wonderful journey for me. I came to the United States in 1995. While my parents wanted me to return to Taiwan after my studies—and they are still asking me when I will come home—they have been very supportive all these years. I thank them for believing in me and allowing me to pursue my dream.
Teaching and doing research on actuarial science at UCF was a great experience. It helped me grow as a researcher and led me to where I am today in the predictive analytics field. I thought life was good in academia for a while. The unexpected force pushed me into the industry. It turned out to be a great trip. I thank my wife and kids for supporting me during these years.
The U.S. health care reform gave me this once-in-a-lifetime opportunity to prove predictive analytics is a successful field. It is a great time to be a part of the health care insurance industry. I really enjoy it.