Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely
Publisher: In-Sight Publishing
Publisher Founding: March 1, 2014
Web Domain: http://www.in-sightpublishing.com
Location: Fort Langley, Township of Langley, British Columbia, Canada
Journal: In-Sight: Independent Interview-Based Journal
Journal Founding: August 2, 2012
Frequency: Three (3) Times Per Year
Review Status: Non-Peer-Reviewed
Access: Electronic/Digital & Open Access
Fees: None (Free)
Volume Numbering: 12
Issue Numbering: 3
Section: E
Theme Type: Idea
Theme Premise: “Outliers and Outsiders”
Theme Part: 31
Formal Sub-Theme: Actuarial Sciences
Individual Publication Date: August 15, 2024
Issue Publication Date: September 1, 2024
Author(s): Scott Douglas Jacobsen
Word Count: 1,134
Image Credits: Erik Haereid.
International Standard Serial Number (ISSN): 2369-6885
*Please see the footnotes, bibliography, and citations, after the publication.*
Abstract
Erik Haereid, born in 1963, grew up in Oslo, Norway. He studied mathematics, statistics and actuarial science at the University of Oslo in the 1980s and 90s, and is educated as an actuary. He has worked over thirty years as an actuary, in several insurance companies, as actuarial consultant, middle manager and broker. In addition, he has worked as an academic director (insurance) in a business school (BI). Now, he runs his own actuarial consulting company with two other actuaries. He is a former member of Mensa, and is a member of some high IQ societies (e.g., Olympiq, Glia, Generiq, VeNuS and WGD). He discusses: the data about risk assessment; new technologies; cases of limited data to make prediction of future events; artificially fill in the gaps; and inter-national collaboration.
Keywords: Artificial intelligence in probability modeling, Bayesian methods in risk assessment, international collaboration on actuarial data, intuitive experience in probability distribution, limited data in actuarial predictions, new technology impact on risk assessment, predictive validity in actuarial statistics, solvency structure in insurance risk models.
Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely
Scott Douglas Jacobsen: How do actuaries make the data about risk assessment make accurate predictions about future events?
Erik Haereid: When determining a probability distribution, or carrying out a risk assessment if you like, this usually happens through a combination of an intuitive experience, at one’s discretion, of what the probability distribution is like and experience corrections. Basically, the distribution is usually not known; one does not know what the probability of the single outcome in the sample space is. But many people have vague or more intuitive ideas about what such distributions are. The fewer empirical data one has, the greater emphasis is placed on the a priori intuitive ideas about the probabilities (e.g. Bayesian statistics).
As you collect relevant data, you normally place increasing emphasis on experience and less on the a priori distribution. This is mathematically more correct since any probability distribution of an arbitrary sample space becomes known with increasing numbers of observations. Cf. the dice rolling example; if you roll enough times, experience will show that there is a 1/6 probability for each of the outcomes 1 to 6, in contrast to if you roll the dice a few times.
When one has no a priori ideas about the probability distribution, and no experience (not empirical data), then it is natural to assume a uniform distribution (a la dice roll); all outcomes have an equal chance of occurring. But as a rule, one has a more intuitive understanding than that. Everyone intuitively understands that, for example, there is not an equal probability of dying in all age groups; we know that a 90-year-old has a higher mortality rate than a 20-year-old; here we can create an a priori probability distribution that is far more correct than the uniform one, even if we do not have a single empirical experience to base this on. In car insurance, we know that old and young men collide more often than the middle age ones. We hardly need data to be able to determine this with great certainty. We also intuitively know that the risk of collision is greater in urban than rural areas. In other words, there are many intuitive factors on which we can create a priori probability distributions in insurance, without a single empirical experience. When we then collect such experience, it is natural and right that we correct the a priori distribution on an ongoing basis, so that it approaches what the relevant data shows us.
In cases where it’s difficult or impossible to collect relevant empirical data, where other indescribable or unmeasurable phenomena affect the risk, it is natural to base the distribution on intuitive models, as mentioned above.
I don’t know if this is relevant to your question, but the data is collected both from their own statistics (the insurance companies record all incidents that happen to their insured) and general databases that exist for general use or on order (cf. surveys). In Norway, for example, we have Statistical research at Statistics Norway (SSB), an institution whose job it is to register a range of population data and other things that are useful for everyone, including the insurance industry. For example, mortality statistics.
Jacobsen: How can the introductions of new technologies change the predictive validity and landscape of actuarial statistics when looking to make such predictions about future events?
Haereid: If you are referring to new technology in the form of more modern analysis tools for processing data, then it must be artificial intelligence that can play a role here. The processing of the data and the outcome can give us new, AI-intuitive models, which we can initially test the durability of. It would be unjustifiable to rely on today’s AI algorithms in terms of probability modelling; it is essential to know what the error is. We don’t know that when we let a machine that we don’t know how arrive at the results and the conclusions, thinks. Therefore, it is important to use human intuition and experience when evaluating the AI processor’s conclusions.
New technology as a premise for changed risk structures provides us a future challenge. As an actuary specialized in life insurance, extended life is the first thing that strikes me. New technology in health and medicine leads to that we are living longer, and it’s difficult to determine probability models related to mortality and longevity. But in a world where technology is accelerating exponentially, we will encounter new insurance-relevant risk areas all the time.
New technology can lead to a changed distribution structure, but not least also greater total costs. It is therefore relevant both to establish a healthy a priori distribution function over the sample space AND a solvency structure. The latter occurs through reinsurance and shared risk with other companies. But, also by limiting the risk areas; you simply do not insure everything you want to insure in the future, until you know more about the risks. In addition, it happens by providing premiums that are too high to begin with, and thus builds up a solvency capital in the event that the payouts exceed the probability models.
What is certain is that new technology leads to more business areas, areas of activity and thus more areas of risk; there is an increased need to financially cover a constantly expanding risk repertoire. Thus, in the future, new business areas will increasingly arise for the insurance industry, with a need for new a priori probability models, increased recording of empirical data and hence flexible risk models based on experiences.
Jacobsen: What happens in cases of limited data to make prediction of future events more vague or less precise too?
Haereid: Then you create intuitive models, and change these in line with the increasing amount of available empirical data. Sometimes it could be difficult to obtain befitting data, and one has to use one’s intuition. Then one uses methods, like the Bayesian, which is constructed to gain some kind of certainty based on personal judgements.
Jacobsen: Is there a way to artificially fill in the gaps in missing data to add more fidelity to predictive actuarial models?
Haereid: Yes, as said by using statistical methods where one assumes something about the distributions without placing emphasis on empirical data. One issue with such models is that they are based on people’s intuitive judgements, which often turn out to be wrong. Even so, our intuition about probability models and the uncertainty associated with them is an essential part of statistics. As we collect data, we will correct the distribution so that it becomes increasingly correct.
Jacobsen: Is there any inter-national collaboration on actuarial data collection to see trends transnationally?
Haereid: Yes, I expect so, among other things with regard to global problems, such as the climate. But, I don’t know anything about that.
Bibliography
None
Footnotes
None
Citations
American Medical Association (AMA 11th Edition): Jacobsen S. Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely. August 2024; 12(3). http://www.in-sightpublishing.com/actuarial-sciences-4
American Psychological Association (APA 7th Edition): Jacobsen, S. (2024, August 15). Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely. In-Sight Publishing. 12(3).
Brazilian National Standards (ABNT): JACOBSEN, S. Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely. In-Sight: Independent Interview-Based Journal, Fort Langley, v. 12, n. 3, 2024.
Chicago/Turabian, Author-Date (17th Edition): Jacobsen, Scott. 2024. “Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely.” In-Sight: Independent Interview-Based Journal 12, no. 3 (Summer). http://www.in-sightpublishing.com/actuarial-sciences-4.
Chicago/Turabian, Notes & Bibliography (17th Edition): Jacobsen, S “Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely.” In-Sight: Independent Interview-Based Journal 12, no. 3 (August 2024).http://www.in-sightpublishing.com/actuarial-sciences-4.
Harvard: Jacobsen, S. (2024) ‘Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely’, In-Sight: Independent Interview-Based Journal, 12(3). <http://www.in-sightpublishing.com/actuarial-sciences-4>.
Harvard (Australian): Jacobsen, S 2024, ‘Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely’, In-Sight: Independent Interview-Based Journal, vol. 12, no. 3, <http://www.in-sightpublishing.com/actuarial-sciences-4>.
Modern Language Association (MLA, 9th Edition): Jacobsen, Scott. “Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely.” In-Sight: Independent Interview-Based Journal, vo.12, no. 3, 2024, http://www.in-sightpublishing.com/actuarial-sciences-4.
Vancouver/ICMJE: Scott J. Actuarial Sciences 4: Erik Haereid, M.Sc., on Predicting the Future Precisely [Internet]. 2024 Aug; 12(3). Available from: http://www.in-sightpublishing.com/actuarial-sciences-4.
License & Copyright
In-Sight Publishing by Scott Douglas Jacobsen is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. ©Scott Douglas Jacobsen and In-Sight Publishing 2012-Present. Unauthorized use or duplication of material without express permission from Scott Douglas Jacobsen strictly prohibited, excerpts and links must use full credit to Scott Douglas Jacobsen and In-Sight Publishing with direction to the original content.
In-Sight Publishing 