Announcement: SOA releases passing candidate numbers for October 2024 Exam FM.

Fireside Chat with Edward W. Frees

By Liang Hong, Peng Shi and Tianyang Wang

Expanding Horizons, June 2022

ehn-2022-06-hong-hero.jpg

Dr. Edward W. “Jed” Frees is the Emeritus Hickman Larson Chair of Actuarial Science at the University of Wisconsin at Madison and a professor at the Australian National University (ANU). He is the only person to be a Fellow of both the American Statistical Association and the Society of Actuaries (SOA).

ehn-2022-06-hong-image-1.jpg

Jed grew up in southern Ohio as the eldest of four boys. He attended Miami University as a math major and then went to the University of Wisconsin at Madison (UW) for a master’s degree in actuarial science. His thesis adviser was James C. Hickman, who later became not only a mentor but also a colleague, boss and friend on Jed’s return to UW as a faculty member.

Upon graduation, Jed worked as an actuarial analyst for three years: first at an actuarial consulting firm in Seattle, then one in Wellington, New Zealand, and finally at the Government Actuaries Department in London, England. After this brief stint in the industry, he entered the University of North Carolina in Chapel Hill (UNC) and received his Ph.D. in mathematical statistics. Jed then began teaching at UW, where he remained for 35 years. After retiring in 2018, he accepted a fractional research appointment at ANU.

Jed was the co-chair of the organizing committee and the founding chairperson of the SOA Education and Research Section, a member of the SOA Board of Directors, a trustee of the Actuarial Foundation, the editor of the North American Actuarial Journal and an actuarial representative to the Social Security Advisory Board’s Technical Panel on Methods and Assumptions. He has written three books; was an editor of a two-volume series titled Predictive Modeling Applications in Actuarial Science; and is currently editing an online, open source book called Loss Data Analytics. He has won many research honors throughout his career, including the SOA’s Annual Prize for best paper published by the Society, the SOA’s Ed Lew Award for research in modeling, the Casualty Actuarial Society’s Hachmeister Award and the Halmstad Prize for best paper published in the actuarial literature (four times).

The following discussion took place virtually in March 2022.

Liang Hong (LH): How did you get into the profession?

Jed Frees (JF): I started as a math undergraduate in the United States, looking for applications. Because of some courses I took at the high school level, I had an interest in physics. However, there were too many smart people in physics, and the field was simply too competitive. I also took two quarters of computer science at Miami. Back then, we were adding zeros and ones together—a lot of binary arithmetic that I found to be tedious. Because of my interest in applications, the other options were either actuarial science or engineering. I chose actuarial science because it involved people. Upon completing my math undergrad major at Miami, I came to Wisconsin and got a master’s degree in actuarial science. Then I worked in the actuarial profession for a few years but always wanted to return to the freedom provided by academia. So, I came back [from London] and went to the University of North Carolina for a Ph.D. in statistics.

Tianyang Wang (TW): You mentioned applications. I assume you also focused on more applications in your research as well?

JF: Yes. Actually, I started my research in mathematical statistics and did my Ph.D. in sequential analysis. I stayed with that field for a couple years but quickly gave up on it. I learned that sequential analysis in business was not practical; at that time, people did not analyze data sequentially. One could not make interventions the way we typically think about in sequential analysis. Switching research fields so quickly after graduation was a big deal, and this switch was motivated by my interest in applications. I believe that researchers who focus on business contexts, such as actuarial science, should understand that their work must be interesting to the business community. The one caveat is that the work may eventually be interesting. It does not have to be interesting right away. It might be interesting 10 years down the road—that’s what we do in academia. However, applications should be the driving force in actuarial science research.

Peng Shi (PS): I’m not an expert in sequential analysis. Since you mentioned your research in those early years, I am thinking about the data we see nowadays. I can see a lot of new things, such as streaming data and telematics. Hypothetically, if you did your Ph.D. in sequential analysis in today’s world, would your research agenda or program be different? Would you find sequential analysis useful?

JF: Absolutely. If I were starting sequential analysis right now, I would have stayed with it. In today’s world, one does not need to think about hypothetical applications. Sequential analysis is interesting and useful in machine learning and optimization techniques. These are fields where you go through complicated processes and make decisions as you go about how to optimize things. In fact, some of the techniques in sequential analysis I was looking at a long time ago are now being used in this context. As you point out, we see more and more data that come in streaming where one makes decisions on an ongoing basis. As an example, think about motor insurance. Here, you are helping drivers decide whether to become alert and hit the brakes as they zoom along the highway. In this way, you are developing tools that not only monitor but also affect their behavior on a real-time basis. Sequential analysis is absolutely important now. The caveat there is, of course, mathematical statistics versus data-driven approaches: the mathematical statistics approach that I was doing was heavily influenced by probability theory. This tactic would not fare well with large-scale streaming data. Still, the idea of making sequential interventions has become more important.

TW: Last year, the SOA created an Early Career Award to support young researchers in academics. Do you have any advice for young folks who are just starting their academic careers?

JF: Absolutely. When you just get started, there is an incredible amount that you need to process and learn. As you know, we mostly talk about the research. However, other things need to happen in your daily job, things such as service to your university and the profession as well as teaching; these things actually take most of your time. In particular, first-year teaching is an endeavor where you need a lot of support because you have to get that right. Over the years, I have seen many colleagues who did not get it right. As a result, they didn’t have the option to stay in academia, even though they were very bright people. Those starting their careers have to make good decisions by choosing the right amount of service at the appropriate level and work hard on teaching. Historically, teaching was challenging because it was one of the things that you were not really well trained for in Ph.D. programs. It is much better now. Teaching training was nonexistent when I started my career.

Even though teaching and service take the most time for early career academics, naturally research is still the distinguishing factor at top-tier research institutions. My impression is that research support in actuarial science has improved at universities and at the professional level, such as the SOA Hickman Scholar Award. Nonetheless, the competition has become fiercer. For those interested in positions in top-tier research institutions, my advice is to make sure that you do a credible job at teaching and service and then devote your energies to the research domain.

TW: Peng, did you have to teach when you were a Ph.D. student at UW?

PS: Yeah. When I was a doctoral student, I did get some teaching experience. In the actuarial program at Wisconsin, students get various teaching opportunities during the Ph.D. study. Also, I think the business school in general offers more resources to help graduate students improve their teaching skills. However, you still want to focus on the research, so you have to look for your own optimal balance.

TW: Many schools have a journal list now for tenure and promotion. Jed, how was the situation when you were working for tenure?

JF: Back then, reviewers were expected to read at least a couple of representative papers by a tenure candidate, use this to evaluate the body of the candidate’s work, and then discuss the impact on the profession and the intellectual community. Nowadays it is common for people to use quantifiable metrics as such journal impact and grant activity, as well as presentations at selected conferences. So, you put up numbers there, and then you compare people using these metrics. As a statistician, I think that is a great place to start. Unfortunately, it is also where a lot of administrators stop.

TW: There are pros and cons.

JF: A benefit of focusing on research metrics is that this approach is very efficient and unbiased. So, there are good reasons for taking that perspective.

LH: You had a highly successful career. Did you ever encounter any challenges in your career?

JF: Almost always—there may have been a three-month period where I did not. But I worked almost too hard to make the challenges manageable. The whole point of research is you are supposed to be diving into the unknown. Further, taking chances is the point of the tenure system that we enjoy in the US. With tenure, you are supposed to be able to step outside your normal boundaries by attempting to do things that may not succeed and take risks. Sadly, I only realized that once I formally retired from Wisconsin. I was guilty of the same thing as everybody else: taking on projects that could reasonably be done in a year or two. Of course, there were many forces pushing me to this short-sightedness, in particular, our grant system where we need to propose projects that can produce quantifiable results in one or two years. (We often joked that one successful approach to grant writing is to write the paper first and then the proposal. In this way, you know that what you are proposing is doable.) Because of this and annual reporting requirements by institutions (to promote accountability), I didn’t take as many risks as I might have.

Fortunately for me, this has changed now that I am in the emeritus stage at Wisconsin, even though I have a fractional appointment with ANU. I am taking some big chances now with my research, working on long-term projects. As a consequence, I’m not ready to talk about results at a conference next year because it will take me a while to figure things out. This is a different mind-set for me, taking on big risks, and it feels like the right thing to do. I received tenure in 1988 and spent much of my career working on less risky projects, so I am learning to enjoy the uncertainty.

In addition to research, I have also been working on open book projects that I am delighted to talk about. Like the research, these involve a paradigm switch and so are very risky. It may be that in 10 years’ time, no one will be interested in the books and other resources that we are producing. Still, the idea of making free and open textbooks to support actuarial students around the globe is appealing to me. It just feels right; our work on open actuarial textbooks has the potential to influence the worldwide actuarial curriculum.

TW: It’s good to hear that you are now taking big risks. Among all the things you’ve done, what is the most interesting one to you?

JF: Nothing in particular. I have enjoyed my work throughout my career. Perhaps the most interesting to me is when I took opportunities to switch research fields. As I mentioned earlier, I started in sequential analysis. The next area of research I did was nonparametric statistics. That was a very hot area, and I worked in this field for about 15 years. I had an appointment with the statistics department in the first 14 years of my career. After that I came full time to the business school, and it was more natural to focus on statistical tools with applications to business, particularly regression, longitudinal and panel data, as well as dependence (copula) modeling. Switching research fields (and units within the university) may have slowed down my research productivity, but I have enjoyed the changes because they have given me opportunities to learn new things, and that’s probably what I like most. Research is just an excuse for learning, right? We all love learning, and I certainly do.

PS: Can you offer some advice to young researchers on how to identify impactful research questions?

JF: That’s a tough question. I can tell you some of the things not to do. Do not go for breadth early in your research career by working in several areas of research. At many universities, the requirement for a Ph.D. is three papers. This tempts one to try to contribute to different topics without an underlying theme. Especially at an early stage, you cannot become an expert in the field when working on disparate topics. When advising (and hiring), I know that someone is an independent researcher when they have developed an expertise and reputation in a certain area, so do not try to do a lot of different things.

The other thing not to do is to focus on extensions of others’ work. It is common in mathematics research to generalize a theoretical result such as a theorem. It is much more productive for society if you can take a problem, characterize it mathematically and develop a model that helps us understand the problem at hand. If you can do 90 percent of the work, that’s plenty. Let other people do the extensions (which is a good way for you to garner citations as well).

I often talk to young researchers who try to say, “Oh, this is what people in industry do,” and they forget that the people in industry—in particular, the actuarial community—are very smart. They know how to address problems and are very intelligent. One of the best experiences I had was when I had a part-time job with Insurance Services Office (ISO). At that time, they had some procedures that were working out pretty well. They were showing them to clients and were super excited, but they could not explain why the procedures worked. Would they work when applied to a situation that differed from the ISO demonstration? In collaboration with ISO analysts, I developed a mathematical framework that they used to provide intuition to their clients. As a nice bonus, this work resulted in several publications in top journals such as the Journal of the American Statistical Association and the Journal of Risk and Insurance. My contribution was simply explaining and documenting what was being done in industry—it was nothing I invented. Although industry gets some really smart people, these people typically do not have the background or the time to document all their work. So, if you can collaborate with industry analysts and understand their problems and work closely with them, sometimes you do not even have to invent anything new but just explain to them why something is working. That is the beauty of mathematical representations.

TW: You are not only a top researcher but also an excellent Ph.D. advisor. You produced several very successful doctoral students. Any secrets to share?

JF: In general, I like to think about Ph.D. advising as a type of teaching. It is an activity that advisers should take time and enjoy doing. As with other types of teaching, when people do not spend that much time on it, they are not good at it. They could be good at it, but they do not want to take the time and make the effort to do so. Take the time and enjoy it.

In general, I enjoy teaching at all levels: undergrad, master’s, and Ph.D. I think that particularly when you write textbooks, you really come to appreciate how much feedback you get with teaching. Textbooks are very hard to write. Even when students do not say anything, you have to explain material at a very deep level. This effort means that you have mastered the material, and that by itself is a huge reward.

At the undergrad and master’s levels, most of the students you know are gone after a year or so. You do not see them except maybe once every 10 years or so at a conference. In contrast, you see your doctoral students on a regular basis for your lifetime. To me, this long-term relationship is the reason that doctoral mentoring is especially rewarding.

In advising Ph.D. students, I should also mention an advantage I had. I started out in industry for a few years. So, I got used to the notion of treating everybody as colleagues and collaborators, and that includes students. Treat students like adults (instead of like children), and they usually behave that way (usually).

TW: Peng, do you have some stories about your Ph.D. time you want to share?

PS: One thing I appreciate a lot is that Jed was always generous in terms of the time he spent with the doctoral students. Now I have my own doctoral students, and I realize that it is a big commitment. Another thing I’d like to share is that Jed held students to a very high standard and developed them as independent thinkers and researchers. In retrospect, I find it to be beneficial to my career in the long term.

LH: Jed, do you have any other research advice for those young assistant professors in the US?

JF: I think my advice goes back to the choice of research topics that we discussed earlier. Don’t focus on incremental research. Don’t make a big switch, and stay within your field of expertise; become the expert of your generation in whatever subfield you select. Try to make your research practical. Don’t tell industry people what to do; work with them.

Also, look for contributions in the “mother disciplines.” I refer to the fact that actuarial science is interdisciplinary and relies on tools from other fields like economics, finance, probability and statistics. Earlier I opined that actuarial research should find its way into practice. However, some of the work that we do supports the foundation of our discipline and does not necessarily have direct practical implications. If you look at my vita, then you will see that I very much support this type of research effort. However, if it is truly fundamental, then it belongs in one of the mother disciplines that actuarial science draws upon. As a researcher, if you can publish some work in journals that are more general, then you get a broader readership (more citations). When you choose projects, look to problems that solve actuarial problems but also have the potential to be a contribution to a mother discipline.

TW: Great advice! In view of the rise of machine learning, what’s your vision about the future of actuarial science research?

JF: Machine learning techniques are great. Now we can do all kinds of things that traditional statistics could not handle before, such as text and video image recognition. These are extremely important in the insurance industry. Actuaries want to continue to focus on the fundamentals of risk management and insurance, such as principles of diversification of risks and things like that. These new techniques are fundamentally altering business right now, and academics should be part of that.

PS: Jed, what do you think is the best way for the professional societies, such as the Casualty Actuarial Society and the SOA, to collaborate with universities in terms of educating the next generation of actuaries?

JF: That’s a great question. It’s a tough one too.

When I started my career, the professional associations barely knew that academia existed. Then about midway through, academics became very, very in-line with the professional associations. In North America, there was a period of time when two University of Waterloo professors, Harry Panjer and Rob Brown, were presidents of the SOA. I served on the SOA Board soon after their terms, so at that time academics were very much involved in the professional associations. The pendulum seems to have swung in the other direction right now. Now there are not nearly as many conversations between academia and the professional associations, which I think is sad.

A little bit more history. I was the co-chair of the organizing committee and the founding chair of the SOA Education and Research Council. Peng mentioned that, as I serve as his role model in some sense. My role model at UW was Jim Hickman. Jim was also a former board member and dean of the business school, but he was the one who came up with the idea of sections for the Society of Actuaries modeled after the American Statistical Association.

LH: Some folks feel the rise of data science is a threat to the actuarial profession. What’s your opinion on this?

JF: Well, this is the one thing about being older; you appreciate history more and more. Just go back and rewind the tape to the discussions that we had about mother disciplines like finance. In the 1990s, people were asking the same question but with “finance” instead of “data science.” At that time, many felt that finance was a threat to the actuarial profession and that actuaries would become a subset of finance. To embrace financial economics, the SOA put together a highly influential monograph on the foundations of financial economics. Nowadays, actuaries discuss options and other derivatives as they would any other tool. In the same way, today the actuarial profession is slowly learning about machine learning techniques and the novel applications they can be used for. My opinion is that, with proper education, data science will simply be another resource in the actuary’s toolkit. Remember that actuaries are responsible for quantitative aspects of risk management and that the insurance industry by itself is about 5 percent of world GDP. That is a significant chunk, which deserves special attention. So, data science is not going to take over actuarial science or anything like that.

LH: We are running out of time. If you don’t mind, let me ask Jed one more question. What might someone be surprised to know about you?

JF: I don’t have that many secrets. If you see me now with my gray hair, it might surprise you to know that I used to be an amateur ballet dancer. A couple of years ago I did an interview article with a German statistics journal called Dependence Modeling. In that article is a picture of me (Figure 8) from a ballet performance when I was a doctoral student at the University of North Carolina. I suppose even back then I sought to have a good balance between work and outside activities.

Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries, the editors, or the respective authors’ employers.


Liang (Jason) Hong, Ph.D., FSA, is an associate professor in the Department of Mathematical Sciences at the University of Texas at Dallas. Liang can be reached at liang.hong@utdallas.edu.

Peng Shi, Ph.D., FSA, ACAS, is an associate professor in the Risk and Insurance Department at the Wisconsin School of Business and holds the Charles and Laura Albright Professorship in Business and Finance at the University of Wisconsin at Madison. Peng can be reached at pshi@bus.wisc.edu.

Tianyang Wang, Ph.D., ASA, CFA, FRM, is an associate professor in the Finance and Real Estate Academic Department at Colorado State University and is currently the editor of Expanding Horizons. Tian can be reached at tianyang.wang@colostate.edu.