Currently showing: Other

14 Jan 15 19:04

Think back 15 years. You were probably racing to meetings with your new flip phone or if you were tech savvy enough, maybe even a Palm Pilot. E-mail was just gaining traction, and if someone said "there's an app for that", you thought mozzarella sticks came with your meal. Between then and now, it seems as though we've taken an express train to the future.

My point is 15 years is a long time for any industry but particularly long when it comes to the insurance industry which is ever evolving. So why do we still rely so heavily on the past to price future risks?

Certainly, claims history and loss development give us insight on what was happening at a specific point in time. But, if we rely only on development from the past, we are applying past actions and decisions to today's evolving world which puts us at a disadvantaged; especially in Casualty lines. That’s why predictive modeling is getting so much attention, and why I was excited to join my industry peers at Advisen’s Predictive Modeling Insights Conference, where I took part in a panel discussion titled “Why Analytics?”

The conference was held at the St. John's University School of Risk Management in New York City, where the next generation of actuaries is being trained to respond to tomorrow’s risk modeling—which, by the way, looks far different than risk modeling most of us have come to know as the standard.

New iterations of risk models – like predictive modeling and Swiss Re's forward-looking modeling (FLM), are critical to success in our evolving and fast-paced world. I believe we have to wean ourselves from a reliance on retrospective modeling because it doesn’t reflect societal and economic changes that influence the "cause –effect" chain of loss development. The commonly used retrospective modeling relies heavily on historical information, which by the time it develops, tends to be countercyclical when applied; the insight it yields isn’t evolving as rapidly as the market.

Typically we have to wait up to 7 – 10+ years for the development of casualty losses, depending on the tail of a particular industry class, to understand its impact on our company and the industry. Imagine the game-changing nature of forward-looking modeling if we can become more nimble in grasping a trend and reacting, even anticipating it!

There are great opportunities in both directions, to identify more profitable business earlier and move away from things that are becoming less profitable. What’s more, FLM can help deliver better claims management and sharper broker segmentation, even more effective marketing and fraud detection.

The predictive model draws on data to isolate and examine a number of predictors, or variables that are indicative of future behavior or results. It’s nothing new in other industries; lenders use it in credit scoring and retailers use it to predict customer behavior. The big differentiator of FLM is that it acknowledges a structured cause-effect chain, identify a variety of factors / drivers that influence the outcome of a specific event; its frequency and severity providing a better understanding of the role the factor / driver is playing.

Personal lines carriers have used predictive modeling with great success; early adopters of FLM have gained a dramatic advantage over competitors. While FLM is more accepted in personal lines than in commercial, we can envision a day when it gains the same level of acceptance. It can’t be ignored.

Reform, regulation and legislation have a significant impact on commerce and collectively yield a wealth of data to be brought to bear on risk selection and pricing. Rigorous analysis of this type of data helps us discover relationships and patterns that are otherwise hidden and counterintuitive. The true art of underwriting will lie in our ability to evolve our analytical process; blending a view of the future with the retrospective analytical process.

Like anything transformational, it won’t always be easy. Resource-strapped companies will resist the investment and stick with what they already know. To invest in the evolution of actuarial science means one is investing in R&D and innovation,which is often likened to working in the unknown. Anxiety must be replaced by courage. It's going to take a major market influencer (or two or three) to make predictive and forward-looking modeling an accepted, even imperative, practice.

The adoption of FLM doesn’t minimize the importance of good underwriting, whose practitioners need to understand that the tools are here to supplement what they do; not take away the need for their expertise. As in other industries, insurers are coming to realize the importance of blending this type of science with good business judgment.

To be relevant in the future, we have to look beyond what we're doing today. As an industry we need to be investing in the development of the science and the talent behind it. Perhaps there were some St. John’s students in the audience who share my enthusiasm.

Category: Other

Location: St. John's University School of Risk Management, New York City


Krishna Burli - 15 Jan 2015, 7:43 a.m.

"Forward-looking perspective: the art and science of modeling" was an interesting reading. However, I am in Insurance Risk Management activities for the last 36+yrs, which is quite a time to express my views with firmness. I strongly feel that the decision makers, be it in the claims management and sharper broker segmentation, or effective marketing and fraud detection areas, I feel, are afraid to take a so called calculated or should I say "Forward-looking perspective" decisions on possible Risk Occurrences or Losses & reorient their actuarial calculations based on the same, with the fear of accountability of business practices & subsequent results in case of adverse probabilities.
I recollect the PML I calculated for a refinery during my days of Risk Engineering in one of the Insurer's company, where in I gave a cushion...(not even rigorous modelling techniques....based on the consistent higher LOSSES in all the previous PMLs calculated by worldwide experts with agreed formulas for Re-insurance purpose, associated by BIG brokers ).... Lo....., all Engineers of Brokers were flown down, discussed with the engineers of the Insurers, regulators..(Without involving me) & finally concluded, whatever formulas were used earlier alone should be used to work out & no "Cushion for the trends or experiences of the past" need be taken in to account...

"It needs guts to be in this racket"....

Nancy Bewlay - 16 Jan 2015, 1:18 p.m.

Mr Burli, I appreciate the perspective you shared. I can confirm that many agree it will take time for the industry to be comfortable with forward looking modeling and It appears as if you have experienced this first hand.

Implementing a new analytical process will take courage and conviction as it will not be perfect in the beginning.

There is a call for formal actuarial training to include the concepts of forward looking modeling. It is strongly believed that our next generation of actuaries will need a foundation that is open to the importance of recognizing and implementing "non traditional" future trends.

Being a "first mover" in an evolving industry is difficult, however, it can lead to competitive advantage. At Swiss Re, we believe creating the next generation of forward looking modeling is imperative to our success and that of the industry.

I agree; "it needs guts to be in this racket"...

Andrea Ferrario - 22 Jan 2015, 4:06 p.m.

Mrs Bewlay, thank you for the enjoyable reading and the nice discussion.
A small remark on the general approach to trends in analytics from my biased perspective.

Forward looking modeling is definitly *the* trend for modern analytics; it has already been succesfully applied in many business contexts and even in insurance, as Allstate proved by taking pioneering decisions.

On the other hand, it does not simply "replace" the already existing approaches to modeling like exploratory data analysis & mining and the use of experts' views (which are still fundamental!).
I believe that only the synergy of those 3 pillars can really provide the "first mover" with a sound approach to tackle innovation and finally enable a data & analytics driven enterprise model.

In this high level framework, culture, people ("As an industry we need to be investing in the development of the science and the talent behind it.") and governance would complete the transformation.

Every big challenge can be won by initiating the right sequence of little steps: this one makes no exception!

If you would like to leave a comment, please, log in.