Subscribe to DSC Newsletter

I recently have been working for a client who develops predictive models for the property and casualty insurance industry. Is this really the next "frontier" of analytics? You know, like database marketing was 10 years ago?


There seems to be big demand for the quantitative master's or PhD level person who wants to do pricing, underwriting and claims models.

Would this be attractive to a good statistician? Is "an entrepreneurial statistician" an oxymoron? I appreciate everyone's thoughts

Tags: insurance, modeling, statstician

Views: 413

Reply to This

Replies to This Discussion

Marybeth:

I think there is a shift taking place where "pure" statistical models are no longer the state of the art of predictive models. Particularly in property and casuality weather and climate models factor now heavily given the fact that if global warming is a reality some very expensive ocean front homes are going to be uninsurable in the next 25 years. Similarly, in finance and business intelligence I am starting to see more causality based models that are validated by statistical models of empirical data or where the output of the models are treated statistically. So statisticians are still very much needed as team member but the move towards causality driven models favors engineers, physicists, and chemists. These all have the quantitative skills but also bring more of a theory based reasoning to the model making and validation.

I would be interested to know more about your experiences.
Cloud computing for large data sets is not easy to integrate in the current IT assets of a typical Fortune 500. First of all the data is highly proprietary and sits in OLTP or several datawarehouses. Secondly, lots of this data is contained in archival form. Getting this data into the cloud is nontrivial for reasons of security, accountability, and cost. For the real money makers, the cloud will sit behind the firewall, as it is the case for the leading crop of information processing folks like Mastercard or JP Morgan.

By the way, almost all clouds are based on PC servers: Google, Yahoo, Amazon to name the big three.
There are two parts here:
Demand for pricing models, automated underwriting and claims modeling in Insurance has been definitely on the rise in last few years. I am not sure this can be analogous to 'database marketing' movement 10 years ago especially when we consider the context that quantitative analytics always played an important role in the industry. Now the movement is to add more sophistication to their existing models. Underwriters are the most expensive assets an organization has, deploying them on the right kind of policies (Policies that involve human judgement, for example Australian Cricketer insured his thumb and forefinger) and automating regular policies results in significant gains to Insurance companies.

Second part:
My personal feeling is 'entrapreneurial statistician' is not an oxymoron - If we consider the word 'entrapreneurial' in the spirit rather than in the 'act' of becoming an entrapreneur. If a statistician, or anyone with an advanced degree in mathematics can think out of box and solve the current problems in a efficieint and effective way, he/she can definitely add a significant value.

As an example, recently we are engaged with a client who has rule-based engine for claims modeling, we used decision trees and data visualization techniques (in combination) to help investigators reduce false positives. This when added to their existing rules engine is driving a huge productivity gains.

My 2 cents contribution.

Cheers - Sri
I am seeing similar happening in the healthcare area, where health insurance companies are using predictive models to predict, based on a person's current health conditions, the future cost to service an individual. Anyone working in this area?
We have a predictive model that takes in health conditions / finacials and predicts the future cost of an individual. Very Accurate.

http://ergenomics.com/blog/can-financial-predictive-model-solve-hea...
http://blog.cequitysolutions.com

There are many posts related to predictive modelling, analytics, text mining and data analysis here.
There has been many enhancements in auto insurance rating, via the use of detailed encoding of vehicle id numbers (think SKU), zip code territory ratings, and the infamous use of credit scoring in order to predict the prob. and severity of a claim. There is also been a shift in focus of thinking in terms of WHO is being insured rather than WHAT is being insured. This of course leads to many more relationships to analyze. The end result is better pricing models. So, to answer your question, Yes, I believe it's a good area for a statistician to be in.

Ralph Winters
You are right that this is a so-called 'hot topic'. Our company is currently doing much work in this area and our real advantage is not so much our math or stats knowledge but our knowledge of the data environment and being able to manipulate this information at the individual level. Traditional actuarial training does not address this at all. In the future, we might expect that knowledge of data will be as equally important as knowledge of statistics/math.
I may be the voice of dissent here but it seems like anyone with an Actuarial background would fit the bill. I fail to see much that is new around that area as Actuarial specialists have been dealing with these kinds of problems for a while. Building predictive models over failure rates is hardly new.
Predictive Modelling is definitely going to play a vital role in Property and Casualty Insurance Industry.
There is increasing awareness of the benefits that predictive modeling can bring to all industries, including insurance.
There has been a growing recognition in cognitive psychology, behavioural economics, and business that predictive models across various segments like property/casualty insurance, help analysts make decisions more accurately, objectively, and economically.Predictive models have enabled insurers to build underwriting models with significant segmentation power and are increasingly being applied in such areas as claims modeling, agency analytics, customer segmentation and target marketing,price optimisation,market mix and Decision Theory as well.
Predictive Modelling would definitely provide an edge to the Insurance Industry
“Predictive modeling will increasingly be regarded as a core competency for all forward-thinking property/casualty companies around which they can fashion their competitive strategies,” he added.
Even Text Mining has a lot of opportunity in this phase.
And no one knows,it may strike Gold.....Hope for the best
The big problems with analysing insurance claims is that there is often a lack of structured data, the main issue being that claims are often in written format and read by an 'emotional' and inconsistent group of numerous humans. Text Mining or similar analysis of the claim text itself is required for speedy and impartial mass processing. This is certainly a fairly new avenue and offers some opportunity for insurance companies to optimise work force (yes, cut work force...) and reduce (or identify) frauduent insurance claims.

- Tim Manns
Congratulations J. Matthews - I think you're off to a very successful and lucrative career. As an update from my perspective - as a recruiter who specializes in analytics and statistics - the insurance field is still growing and many of the top tier (stable, no bad debt - NOT AIG) type companies are looking for predictive modeling talent.

Here's my commercial - CALL or Email me if this is an attractive career path for you! 312-265-6280 or [email protected]

RSS

On Data Science Central

© 2019   AnalyticBridge.com is a subsidiary and dedicated channel of Data Science Central LLC   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service