The world we live in is changing. Technology is developing at an ever-increasing rate and the world’s climate is creating an evolving landscape that is presenting new risks and new challenges across the globe.
This all means that property (re)insurers have had to adapt the way in which they manage and assess risks, with technology playing an increasingly important role.
A big area of growth in this regard over recent years has been the rise of geospatial data.
Jay Mullen, insurance team leader at geographic information system software provider Esri, says that such technology is giving insurance companies greater insight into their risk portfolios, particularly when combined with additional data sources.
“Geospatial technology and data is enabling insurance organizations to have a better view of their risks and the accumulation of those risks around the world,” he says. “When combining new earth observation data sources along with exposure locations and modeled hazard layers from natural perils to man-made perils like crime, a higher resolution prediction of risk can be established.
“This is a shift from the days of only evaluating historical losses to price risk exposure.”
It’s like card-counting for (re)insurance
Mullen adds that this makes geospatial data a real game changer for the property (re)insurance market.
“Geospatial technology and data is akin to counting cards in the business of property risk underwriting, and is the single highest ROI technology insurers can invest in today,” he says. “As our climate evolves and catastrophes become more devastating and more frequent, earth observation, or E/O, data sources will advance all while technology becomes more efficient & capable.
“The incumbents in the insurance industry will then be in a race to implement these systems of higher resolution insights to help them better understand the risk across all perils, across all lines of business and better respond when disasters do strike.”
One company that is amongst those leading the charge when it comes to geospatial data is HazardHub, now part of Guidewire.
HazardHub is the most comprehensive property risk database ever created, with more than 1,000 different variables covering every address in the US, all available to insurers through an API.
But Guidewire senior manager, operations John Siegman – who is one of the co-founders of HazardHub, says that what sets the company apart from their competitors is not just the scale of the data available, but also the way in which they provide it to their users.
“Unlike a lot of our competition, we provide all of the data that goes into all of our models, so there's no black box here,” he says. “That allows our customers to build their own models or pick a variable that's really important to them and score on that, as opposed to other things.”
But that doesn’t mean HazardHub’s models aren’t important, with the company having built a number of their own internal risk-rating models for things like floods and wildfires, as well as a number of their own proprietary datasets.
This includes the US’s first and only national database of fire hydrant,s for which Siegman has personally identified more than 200,000 individual fire hydrant locations.
A variety of use cases
The varied nature of HazardHub’s databases – which covers the whole gamut of variables from environmental to socio-political factors – means that the platform has a lot of different use cases for insurers.
“We've got customers that use us in a competitive rate or situation where they're pulling one or two, maybe three variables that they're really interested in, and then making the determination way up front if they want to talk to that prospect,” Siegman says. “Then we've got customers that will use us to pre-populate quotes after a customer inputs their address, or for things like risk-based pricing.
“And we also have customers who are using us in their marketing. For example, I score an F for wildfire, so unless you're really willing to take on that risk as an insurer, you could decide not to waste money sending me any marketing material.”
And on the reinsurance side, Siegman says it is all about understanding the risks that have been taken on by their insurer partners.
“They want to understand the risk that they're taking on,” he says. “And they want to be able to say to the companies that they work with: “Hey, you've got too much risk or not enough risk”.”
Remote sensors are another area that is having a big impact on the property (re)insurance market, providing data on everything from roof condition to water levels.
Jay Gentry, GIC director of member development at Vexcel Data Program, says this helps (re)insurers in three key ways.
“Number one, it helps them look at their current book of business and see if they have the right risk allocation across any geography that they're underwriting in,” he says. “Number two, it helps them manage costs by reducing the amount of cost it takes to actually write the policy. That’s because we're going to try to automate [using this data] and have people involved as little as possible.”
The third area is claims.
“From a claims perspective, if I have remote information or remote access to information, and I'm not going to have to send someone there, then two things happen,” Gentry says. “Number one, I can very quickly ascertain if it is a complete loss or not. If it is, write a check and move on.
“Number two is I can dissect fraud opportunities – is there a fraud opportunity in this type of claim or claim environment? So all these things mean that you're paying claims fast and figuring out if claims are actually real quickly. All of this helps to reduce costs dramatically.”
But all this new data raises the question of bandwidth, and it could’ve been easy for (re)insurers to end up lacking the capacity to deal with the volume and variety of data they now have access to.
Luckily, technology once again has the answer, and Donald Light, director North America property casualty practice at Celent, says that the answer has come in the form of artificial intelligence.
“It's great having all this data, but unless you can turn it into a digestible form or a form that's easily usable with either an automated system or a human underwriter or claim handler, then It's not that much good,” he says.
“These new kinds of analytics basically transform or process the data into a form that either a machine, a system, or a human can easily digest. This could be in terms of scores, or your red, amber, green light, or whether or not to make a referral, for example, if there's fraud.”
