This is my first blog post in a while. I plan to pick up the pace during the latter part of June as I wind down the longest “vacation” in 28+ years. While I plan to resume blogging more often about coverage related issues (I’ve got some really good topics coming up), I felt compelled to do a quick post about another startup that is going to revolutionize the industry. I first commented about a recent interview with their CEO on my LinkedIn Discussion Board.
If you’re not aware, in addition to this blog, I also communicate about industry issues on LinkedIn and via Twitter…you can access them and subscribe from my Contact page:
This new startup claims that they can place homeowners insurance without an application of any kind. They basically need to know your name and address, then they get all the data they need from other sources. Presumably this process takes a couple of minutes and that’s it. I invite you to read the interview and then join the discussion. This “We can write your insurance in 2-3 minutes” mantra is becoming a common theme with many of these startups. The rationale is improving the “customer experience.” What about the customer experience of having a 6-figure (or greater) uncovered loss because nobody took the time to prompt the consumer for exposures?
I make this point in the LinkedIn discussion:
“If you were going to sky dive for the first time, would you pack your parachute yourself? Would you insist that someone else pack it as fast as possible with almost no attention to detail because you’re in a hurry? Or pack it with their eyes closed? My insurance program is my financial parachute. There are no shortcuts to security unless you’re buying insurance with no exclusions or policy limits and most of what is being sold on the internet is far from that.”
So what do you think? Are regulators who welcome these startups with open arms really doing their constituents a service by embracing “fast and cheap”? Or are they simply insuring (no pun intended) financial ruin for perhaps a great many of them? For more discussion, scroll down to the Comments section of this post and visit the LinkedIn discussion on this subject.
I wonder if the individuals often quoted in these InsurTech articles really, truly understand what insurance is all about and what they’re doing or think they’re doing. The coming years will be interesting. Either I’ll be remembered as a cautionary visionary or a cranky old guy living in the past and yelling at the big data kids to get off of his lawn.
Bill Wilson
Latest posts by Bill Wilson (see all)
- The Invisible But Potentially Catastrophic Homeowners Exclusion That’s Not An Exclusion - September 19, 2023
- Revisiting the Illusory Coverage Assertion Following a Claim Denial - September 19, 2023
- FREE Webcast: How to Survive and Thrive in a Hard Market - August 1, 2023
Similar to Slice, the CEO of British insurer Aviva says, “We’re going to go from asking hundreds of questions on insurance to asking none — zero. We’re piloting it now. We’re going to roll it out.” ….
First of all, a homeowners application doesn’t have HUNDREDS of questions and endless forms, so I wonder if the Aviva CEO has ever even looked at an application or filled one out. In fact, before “big data,” there were actually relatively few questions, most being about construction, occupancy, public and private fire protection, etc. Renewal applications are very simple to complete and, if processed by a good agent, they update and verify that the information is correct which is a good thing. If anything, it’s the increasing reliance on big data that has caused the expansion of information requested. The BIG issue with BIG data is accuracy, credibility and relevancy.
At the link below, the Aviva CEO says, “I can insure your house now without asking any questions because I use big data.” How can you not ask ANY (at least confirmational) questions about assets, income, values, exposures to loss, etc. in order to determine what policy(ies) and endorsements are warranted? I mean, ZERO questions? Just how reliable do they think this “big data” is? For example….
If you go to the tax rolls in my county and look at the size of my house, it’s off by over 1,000 sq. ft. because of all of the renovations over the years that were never reported anywhere in the big data rolls. I’ve had three valuations done on my home and two of the three relied on that data and came up with a value $150,000 too low. Thank you, big data.
Compare the information on homes for sale at realtor.com, Zillow.com, etc. They can’t even agree on the number of toilets in the house, much less square footage, age, condition, use, etc. There is currently a class action suit against Zillow alleging that their erroneous information has resulted in the undervaluation of homes to the detriment of sellers.
Data source credibility is a huge issue, along with the black box algorithms being used to weight the relevancy of the data to loss propensity. How can regulators enforce laws requiring that rates be adequate but not excessive nor unfairly discriminatory if they have no idea how this information is being used and whether the data is even credible, accurate or relevant?
In 2 of 3 years, my HO carrier came up with the wrong insurance/credit score for us to increase our premium by a total of $1,700 and it was only because of the law requiring reporting of adverse impact that allowed me to catch these errors. Would the typical consumer know this?
In the 1970s and earlier, the old rating bureaus/ISO audited insurance policie, the error rate was astronomical…a ‘good’ insurer’s error ratio was in the 40% range and many were far worse. In the case of workers comp experience rating, information error rates have been estimated to be as high as 40%.
When I discussed this with one individual, the counterpoint was whether consumers would be willing to give up some ‘accuracy’ in exchange for convenience. I’m not sure how many consumers would be willing to reduce their purchasing time from an hour to 2-3 minutes if they knew that the result could be that their premium could be off by $1,000. Most of my family and friends would consider it worth their time to simply answer some questions to minimize the chance that the “alternative facts” procured by the insurer could result in them being overcharged by hundreds of dollars.
The odds of having a flood loss in a 100-year flood plain is 1%, yet people buy flood insurance, sometimes admittedly at the insistence of a lender. Who would buy a policy if the odds were 40% that you’d be overcharged AND that you could have claims that aren’t covered but might have been covered if you had spent a half hour longer answering questions?
One of the problems I have with “big data” is that, for the most part, it is sold for the benefits that allegedly accrue to INSURERS, not insureds. If big data is good and really works, then it’s possible that overall premiums could be a little lower. But losses still happen and, on average and admittedly a generalization, half of consumers are going to pay more and half are going to pay less under big data. And the vast majority of what you read touting big data and predictive modeling is how it benefits insurers, not consumers.
BTW, construction type is pretty easy…either your home is frame/brick veneer or it isn’t. Historically, that’s about all that was needed for the construction component of the premium. If you’re in an urban or suburban area, you almost certainly are within 1,000 ft. of a fire hydrant and that’s all that matters, along with the rating of your fire protection district. And that’s a good example of the use of big data…with an address, you can identify the public protection class for rating with a very high degree of accuracy. Credit reports, on the other hand, are notoriously inaccurate but there can be a HUGE component in premium determination…literally, someone with a DUI, all other factors being equal, can have a lower auto premium than someone with an inferior credit score, a number than be based on bad data or which is more correlation than prediction.
“Big data” is just another level of the actuarial examination of information to determine price that has been used for many, many decades. It’s not inherently bad or evil. Neither are hand guns and nuclear energy, but we know that they can be abused. Most of what we read that favors “big data” is from people with a vested interest in selling the data or the analytics and those writing the news stories are too often not asking tough questions and merely repurposing press releases they blindly accept as gospel. These things can be great tools for actuaries and underwriters, but overreliance is dangerous.
I wonder if the individuals often quoted in these InsurTech articles really, truly understand what insurance is all about and what they’re doing or think they’re doing. The coming years will be interesting. Either I’ll be remembered as a cautionary visionary or a cranky old guy living in the past and yelling at the big data kids to get off of his lawn.
Thanks for your continued vigilance on this issue. However what seems missing from this discussion is what insurance company/s is willing to write risks with such slim information? Here’s where the industry must be careful not to yield up its underwriting obligation in exchange for a flow of new business.
I’m reminded of the old adage, “be careful for what you ask for, you might just get it”. If carriers align themselves with organizations who model themselves after the “sharing economy” then most of their clients and investors will be sharing in more than just the heartache of climbing combined loss ratios and low investment income yields. Our Good Neighbors recently lost $7B by asking a lot more questions than “name, rank and serial number”.
As Tom Peters popularized in his 1982 book, “In Search of Excellence”, the industry needs to “stick to the knitting” of measuring risk and price accordingly.
I think the problem might be that they don’t believe it’s “slim” information. They believe voluminous data is a better indication of claims traffic than traditional measures. Or perhaps some believe that it might not be as good but that the cost is so much less than traditional underwriting methods, that it’s cost effective. No doubt some believe that their predictive models are better than others and that will give them a competitive advantage. Virtually everything you read about data analytics and predictive modeling is oriented towards how its use is allegedly better for the insurer than traditional models.
What about the INSURED? Are these models truly fairly discriminatory? How can a regulator affirm that? What credibility does data from hundreds of third-party sources have? Is it really predictive or just correlative? As I alluded to in my Comment, I know for a fact that an insurance valuation on my house used tax records rather than measurements for the size of my house. These records are off by at least 25%? So is my house underinsured? If that data is used, how does this erroneous information impact my premium?
And how can someone really identify all of my unique exposures to property and liability losses by giving up the insurance application in exchange for a name and address and some suspect data that is fed into a black box algorithm? Is someone other than an algorithm looking at this data? Do they know I live on a lake? Do they know I have a boat dock? Do they know that, because my boat dock is on Army Corps of Engineers property and not my “residence premises” that I have no HO Coverage B on that boat dock? When it’s destroyed by windstorm, who is responsible for that uncovered $30,000 loss?
Seems to me that insurers are assuming greater risk by not obtaining information from the insured via the representations on an insurance application when it comes to exposures and property values. Will the many coverage and limit gaps that are going to occur lead to greater litigation of inadequately covered claims? I suspect that the trial bar is even more excited than insurers about big data. Big data could mean big lawsuits.
Of course, I could be entirely wrong about all of this and perhaps I should join in the orgasmic frenzy that is big data.
Another article of interest:
https://insnerds.com/data-not-better-better-data-better-debunking-myth-data-technology-will-obsolete-insurance/
Be sure to check out the feedback where I posted this article on LinkedIn:
https://www.linkedin.com/pulse/matter-how-you-slice-its-wrong-bill-wilson
Do Insurance Agents Matter?
https://www.linkedin.com/groups/7036591/7036591-6282582230573342720
Great article from Don Riggin:
http://insurancethoughtleadership.com/can-we-really-disrupt-insurance-4-ideas/