Insurance Companies and Big Data

Forums › Off topic › Insurance Companies and Big Data

GregF

52 posts

Master Geek

#205143 31-Oct-2016 15:40

We all are aware of Facebook / Google etc tracking our web behaviour, and store cards like FlyBuys / CountDown etc tracking spend behaviour, in order to pitch advertising.

I've lamented a few times over a pint about how bad it will be when MasterCard / Visa start 'selling' consumer info - and one of the discussion points was what we thought Insurance companies would / could do with the data.

I can't recall when I first had this discussion - but essentially it's the premise that Insurance companies could scan your spending habits to determine if they were going to insure you. (Like - you eat bad food a lot, so up goes your Medical Insurance risk / cost).

Today, while scanning headlines, I see this exact idea is becoming talked about:

http://www.stuff.co.nz/business/85794331/wholesale-change-ahead-for-insurance-industry-report-warns

Now - I can get why they want to do this - the whole idea of the insurance industry is to sell risk mitigation, so if they have an accessible data source which allows them to better risk-mitigate (thus better profit for the company) they will do this.

What I cringed at was one line, and this is where the moral lines for me start getting blurred:

"...predicted the only group that would object would be those who thought sharing [their private data] would lead to higher premiums, but they are in turn would be classed as high-risk for not partaking."

So the Insurance industry is effectively saying you are guilty until *you* prove innocence - if you don't give up your data (and they're talking real time stuff here) then your premiums go up.

I shake my head - but I can't say I'm surprised. And as much as I hope it doesn't get to this - I can't help but think it's already well underway.

wasabi2k

2096 posts

Uber Geek

#1661545 31-Oct-2016 16:27

The technology exists today to do this - the issues are around data ownership, privacy and sharing - as well as the lowest common denominator (not everyone has wearables and connected cars).

Had a question as a presentation I gave asking if I really thought people would share all this info - I think if you can make a convincing benefit statement - people will do it. Combined with the current generation's utter disregard for protecting their information and privacy I don't doubt it will continue.

Some insurers already offer fitbits - not sure if they collect the data - but they should. This stuff only really becomes useful when you have a large enough data set to start doing analytics on it.

The scarier issue is the introduction of gender and race biases - but enabled by data. Most countries have enshrined protection from discrimination based on race, gender and sexuality - but if your analytics show that (for example) Maori Men die from lung cancer more than other demographics - you charge them more to "optimise" based on your data right? Look at my flashy computer model that proves it.

Granted it is a new type of Actuarial Modeling (car insurance based on age (http://www.contingencies.org/septoct07/age.pdf)) - but a decent analytics platform will let you segment however you like, easily and quickly. Once you push that capability out of the industries that have history and governance for handling it.... exciting times.

UHD

655 posts

Ultimate Geek
Inactive user

#1661578 31-Oct-2016 17:07

wasabi2k:

The technology exists today to do this - the issues are around data ownership, privacy and sharing - as well as the lowest common denominator (not everyone has wearables and connected cars).

Had a question as a presentation I gave asking if I really thought people would share all this info - I think if you can make a convincing benefit statement - people will do it. Combined with the current generation's utter disregard for protecting their information and privacy I don't doubt it will continue.

Some insurers already offer fitbits - not sure if they collect the data - but they should. This stuff only really becomes useful when you have a large enough data set to start doing analytics on it.

The scarier issue is the introduction of gender and race biases - but enabled by data. Most countries have enshrined protection from discrimination based on race, gender and sexuality - but if your analytics show that (for example) Maori Men die from lung cancer more than other demographics - you charge them more to "optimise" based on your data right? Look at my flashy computer model that proves it.

Granted it is a new type of Actuarial Modeling (car insurance based on age (http://www.contingencies.org/septoct07/age.pdf)) - but a decent analytics platform will let you segment however you like, easily and quickly. Once you push that capability out of the industries that have history and governance for handling it.... exciting times.

Wasn't the previous generation the one who built computer systems with no regard for security or privacy? At least this generation of engineers have some understanding of the need to secure and protect data from the ground up.

Playing the Devil's advocate here: I'm not entirely convinced by the wrongness of using data to categorise those with a higher risk of some ailment. I would absolutely understand if data showed that the incidence of cancerous melanoma was greater in those with lighter skin than those with darker skin which meant a higher premium on those with lighter skin for skin cancer treatments.

I think it is fair to be against unfounded prejudice but if you have research demonstrating some fact (women far more likely to be pregnant than men...) then I struggle to see the harm in using the data. Surely if it seems intuitively right rather than sexist that males aged 15-28 pay a much higher premium for car insurance (based on solid, tragic data) then it cannot be racist or prejudiced to argue similar positions based on race as long as data exists to back up the reasoning for doing so.

Fred99

13684 posts

Uber Geek

#1661713 31-Oct-2016 20:50

GregF:

Now - I can get why they want to do this - the whole idea of the insurance industry is to sell risk mitigation, so if they have an accessible data source which allows them to better risk-mitigate (thus better profit for the company) they will do this.

Not really. Despite the origins of the insurance "market" being individual risk mitigation through a shared risk model, I don't think there are any retail insurers like that left. The last in NZ was AMI - the "M" for mutual.
The whole idea of the insurance industry now is for the companies to return a profit to their shareholders.

sir1963

3260 posts

Uber Geek

Subscriber

#1661723 31-Oct-2016 21:03

Perhaps the tables can be rebalanced in that the commercial status should be made public.

After all, surely customers are just entitled to assess the risk the company poses

Financial stability

Payout %

Legal cases against them won/lost

% profit made

Investments made (customers may object to a company who invests in weapon manufacture)

etc etc etc etc, all legitimate concerns a customer may have.

Geektastic

17942 posts

Uber Geek

Trusted

Lifetime subscriber

#1661740 31-Oct-2016 21:35

I've had a few conversations with insurers about risk.

Sometimes I have been unable to get cover for something and they have productised their offerings so much that we have no 'free market' of insurance such as Lloyds in London, where you can propose a risk and get offered terms. The risk might mean a payout of $1000 and the terms may be a premium of $999 but they leave it up to the client to accept or decline whereas here they just won't cover it. It can be frustrating.