|
|
|
Actuarial Roundtable Discussion— Part Two
The State of Casualty Actuarial Science Today
by Arthur J. SchwartzIn February, four distinguished actuaries and I got together by conference call to discuss the state of casualty actuarial science today. Our panelists include:
Glenn G. Meyers works for Insurance Services Office in New Jersey, has written many papers on risk loads, catastrophe ratemaking, and capital allocation, and many other topics. He has served the CAS on the Examination Committee for several years and on a number of CAS research committees.
Howard C. Mahler currently lives in Boston, teaches actuarial exam seminars, and consults. He has written papers on a variety of topics including workers compensation, credibility, experience rating, retrospective rating, and underwriting profit models. He served for a dozen years on the CAS Examination Committee, including three years as chairperson of the committee.
Sholom Feldblum works with corporate financial models for Liberty Mutual in Boston. He has written numerous papers explaining actuarial concepts, helping students learn these topics efficiently.
Stephen W. Philbrick is with Conning Asset Management, a division of Swiss Re, in Baltimore. His paper on credibility concepts won the Woodward-Fondiller Prize. He has been active on numerous CAS committees, including chairing the Committee on Principles, which seeks common principles with other actuarial organizations. He writes the "Brainstorms" column for The Actuarial Review, which discusses interesting new applications of actuarial science.
Schwartz: Are there any technical or business skills that the current CAS syllabus either does not cover or does not cover adequately? Are there any technical or business skills that the current CAS syllabus covers more thoroughly than necessary (out of proportion to their usefulness)?
Mahler: The CAS has been wrestling with the issue of allocating space on the syllabus at least since the early 1980's. Glenn and others are correct in saying that the process of selecting what students need to study is continuous and ongoing. It's not whether a reading is useless. It's whether a reading is less useful than something else we want to have on the syllabus. We need to improve the efficiency of our education process. A paper when written may contain important new ideas, but that does not make it necessarily the best means of conveying those ideas to our students.
If a paper can be read, and easily remembered, it's great. Take the Philbrick paper on credibility from the 1981 Proceedings. The target shooting analogy is insightful; not only does it help you understand the material while you are studying the paper, it will also stay with you long after. The Philbrick paper is what we should strive for in education, rather than our current approach. We don't have enough learning efficiency in our education process. We need more bang for the buck!
Meyers: Efficiency is critical. We have to do more with fewer papers. The syllabus as a whole should have less detail ontechnical methods. Instead it should focus on the business use of the technical methods.
Also, I'd like to comment on multiple-choice questions, which form the basis for testing on the first several exams. We cannot, through multiple-choice questions, really test whether a candidate has a deep understanding of a subject. What we can do is test if the student knows the language and has a basic understanding of the concepts.
For example, on loss distributions and fitting them, it's not necessary to learn every distribution or fitting technique. I'd want students to know how to use loss distributions to calculate the credit for deductibles or estimate increased limit factors.
Mahler: One of the often overlooked aspects of designing a syllabus is making sure you set the right level of comprehension at which a given subject is to be learned and tested. For example, the required level of comprehension for a life or pension actuary with life contingencies material is much greater than needed by most casualty actuaries. As another example, if you learn how to work with a Pareto distribution, do you also need to learn the Loglogistic distribution?
The concept of efficiency also involves reducing the number of examples, so a student gets the idea and has a few practical examples to study. It is inefficient to present loss reserving via 10 papers written by 10 different authors at 10 different times for 10 different purposes. One or two well-written summary study notes can compare and contrast the different methods and present the ideas in a manner such that the students learn more in less time. The CAS has been trying to add more well written study notes to the syllabus.
Philbrick: On loss distributions, I think learning one or two is enough. We have to think about the material on the syllabus versus the practicing actuary. The practitioner might need to know 88 distributions. For some people who write like that, they want to put it all on the exam. Maybe learning fewer is learning better, since once you've completed the exams and have a need for using this or that technique to solve a practical business-related problem, you can look up the details.
Mahler: The current textbook discusses about 20 loss distributions. We attach to the exam the underlying information on about 14 loss distributions. Restricting the syllabus readings and the exam questions to about half a dozen common loss distributions would more efficiently get the important ideas across. (Practicing actuaries want access to as big a list of distributions as they can get.) If it cost a substantial amount of money to put each extra distribution on the syllabus, the CAS would be more careful in how it allocates the scarce resource of student's time.
Philbrick: The marginal value of learning a 6th distribution, or more, grows less and less.
Meyers: I'd like students to learn three distributions, with one of those three being an empirical distribution.
Mahler: Another issue at the forefront right now is travel time. The CAS Board has recently stated that five to seven years is a good median travel time. That's measured from the time you take your first casualty actuarial job until Fellowship. Historically, median travel time has been more like ten years. In my opinion, it would be incorrect to focus solely on travel time. The CAS needs to also focus on how much was learned and retained. Also, of the many who start the education process, how many persevere through to the end? What percentage of promising candidates, who would make fine Fellows, stop or leave along the way because of the defects of our educational system, as opposed to economic, personal, or other reasons? Is an educational system with a median travel time of 6 years and a 30 percent success rate better than one with an 8-year median and a 60 percent success rate?
Feldblum: Candidates consider travel time and study time when deciding whether to pursue an actuarial career. We compete for the best candidates among a variety of other professions catering to the mathematically oriented student. We must choose passing ratios to accommodate the needs of students. The long travel time for the actuarial designation is one reason why the attraction of the actuarial profession is dropping.
Meyers: Should we cut down on the number of exams, have fewer of them?
Feldblum: The pass ratios make it very difficult to pass all the exams in five to seven years. A pass ratio of 25 to 30 percent reduces the attractiveness of the profession.
Meyers: A higher pass ratio, such as 50 percent, could possibly lead to a student passing, who had demonstrated a mastery of only 30 to 40 percent of the syllabus. I doubt that we would deem that acceptable.
Feldblum: The quantity of the material on the exams is not that different from years ago. However, I'd like to see us ask fewer questions that require memorization of details, of lists, or of formulas; and ask more questions that test the general understanding of a reading.
Meyers: I have looked at student performance on several past exams. One interesting finding—students "kill" the straight memorization questions; they do extremely well on them. Yet when it comes to a "thinking type" question, students don't do as well.
Feldblum: I don't mean the "thinking type" questions. I mean general questions that do not simply test the recollection of a specific formula.
Mahler: A couple of interesting ideas for education are being studied right now. For example, on DFA, rather than primarily teaching this material as part of a CAS exam, we are considering having a seminar. Another alternative to the traditional closed-book, timed exam would be open-book exams or even Internet courses. So the CAS may be moving away somewhat from the traditional approach as we go forward. I hope such changes will be tried carefully, one small step at a time.
Philbrick: I'm a fan of the open-book exam. Our traditional approach has been a closed-book exam with a strict time limit. In real life my boss might give me a project to complete within a limited time frame, but he doesn't say that I have to do it without looking at a book. An open-book exam may imply more questions for students that require them to think.
Mahler: Historically, questions requiring deep comprehension have been limited to the last few Fellowship exams. Even on these exams there would be at most two or three such questions. These questions are open-ended, with no single right answer—basically asking if you can write something on this scenario, which would be reasonable for a Fellow of the CAS to write. Such questions are difficult to answer and difficult to grade.
Meyers: When I look at people who make it through the exams and doing well as Fellows, they're good thinkers. Employers hire Fellows because of their ability to form a conceptual solution to a difficult problem. If they couldn't do this they would not be around long enough to take the Fellowship exams.
Philbrick: Employers are looking for that; people who conceptualize an issue well.
Mahler: Another issue with the exams right now is the international issue. Should students principally practicing outside the U.S. and Canada have an alternative to taking a U.S. 7 or a Canadian 7 Exam?
Meyers: Let's look at the SOA. In their new education system, at first they omitted all material that was nation-specific. Then in the last six months, they revolted and put nation-specific material back on the syllabus. The practitioners, rather than the academics, led the charge to put the nation-specific material on.
Mahler: One possibility would be to keep the U.S. and Canadian Exam 7, and then add as an alternative an exam with European (and maybe some U.S.) content.
Meyers: I'd like to see the nation-specific material limited to one or at most two exams.
Mahler: One issue is whether we'll get enough volume of candidates from a specific nation or region of the world to justify the work of creating an individual exam. For example, if we knew we'd get 50 people per year from India, we'd likely work with them to create a nation-specific exam. In some countries there may already be a nation-specific exam being given by another organization for which the CAS might want to give credit.
Philbrick: Let's consider what would be on a nation-specific exam. Workers compensation is not handled the U.S. way in any other country; other topics would include tax laws and accounting. For a European exam, perhaps it could be half-European and half-country-specific.
Schwartz: There's been some discussion of offering university credit for some exams. What is your quick view (then let's open it up to a fuller discussion)?
Meyers: No.
Mahler: Skeptical.
Philbrick: Positive; hopeful.
Feldblum: I'll pass.
Mahler: In theory, it sounds good. However, I see potentially serious problems. Presumably the CAS would continue to offer exams on these subjects also. Then you would have the problem of having these exams be of comparable difficulty to getting credit via a university course. Also, what is covered and the grading varies between the many different universities in North America. If a student gets a B+, that may mean very different things at different universities. We would lose control and uniformity.
Meyers: If you know the material, then passing an exam on that material is not that big of a factor.
Schwartz: To get around the problem of different grading standards, couldn't you have CAS standards, CAS recommended texts, or even CAS exams that the universities would simply administer?
Mahler: Then you're back to having a standardized test—a CAS exam or joint CAS/SOA exam.
Schwartz: Are there enough statistics on the early exams?
Mahler: It's true there's not enough about statistics on Exam 1 to prepare properly for Exams 3 and 4, yet I've been teaching an early exam (Course 4B before 2000 and courses 3 and 4 since then) for years now, and I have not noticed that much of a difference in students' preparedness with statistics since the introduction of the 2000 syllabus. It would probably be helpful to put statistics back on the exams and test it. In the 2005 syllabus, there may be some major restructuring of syllabus topics. It is a good idea for students to have a somewhat better foundation on statistics.
Schwartz: What ideas are being discussed to restructure the exams?
Mahler: I believe that one of the problems with our current exam structure, is the long time until most CAS students encounter material directly related to their jobs. Therefore, I would like to see the CAS consider having a short exam, perhaps two hours, and a relatively easy exam, early in the sequence of exams. The exam would introduce students to the basics of casualty actuarial science and property-casualty insurance. It would cover very basic ratemaking, very basic reserving, as well as some basic policy forms and underwriting
Meyers: I am with you. I'm on the Exam Committee. We reviewed the old 3b, which was a memory exam, yet it let the student learn the language of insurance.
Mahler: The material I am proposing be on the new introductory exam is currently on later exams. The policy forms and underwriting could be taken from CPCU books or other material. Everything else would be taken from the new edition of the CAS Foundations of Casualty Actuarial Science. Something from most chapters except credibility could be put on this exam. (I assume credibility would continue to be covered on another early exam.) In some or all cases only a portion of the chapter from Foundations would be on the exam. The most fundamental or important portions should be selected so as to keep the exam easy and the amount of reading manageable. Introducing basics on an early exam and following up with more advanced material on a later exam I believe is a good educational technique, as well as having practical value to students and to employers. If they chose to, a student could take this Foundations exam right before or right after getting their first casualty actuarial job.
Meyers: I am with you in concept.
Feldblum: The insurance material is more important to the practicing actuary than to the theoretical actuary. Exams 3 and 4 are too theoretical. They should be replaced by an early exam that teaches ratemaking, reserving, and reinsurance material. We have to be competitive with other professions; so the early exams should correlate closely with the kind of work that students are doing. Having highly theoretical material early in the syllabus is unwise; it dissuades the student from continuing on with the exams. Most of the students can not see the connection between the highly theoretical mathematics on the early exams and the work they are doing on the job.
Mahler: One impetus behind why the syllabus was created the way it currently is was to have the theoretical math up front right after you got out of college. The highly theoretical material does attract some people to our profession. A 26-year-old mathematics graduate student would be attracted to the profession by the mathematical and academic nature of the material on the early exams.
Meyers: I don't think we need to cover on the syllabus all the mathematical material, that only some actuaries use. Some of this advanced mathematical material can be covered elsewhere for those who wish to learn it. For example, Stuart Klugman is now offering an advanced seminar on loss distributions and it has had a healthy attendance over the years.
Mahler: Again you're back to the syllabus space being a scarce resource.
Schwartz: Thank you very much for a great discussion!