Information overload

There is a breathtaking amount of market data available to insurers but does the industry know how to effectively use the information? Andrew Foxwell reports on a recent Post roundtable

Compared to a few years ago, the amount of data available to insurers today is breathtaking. From flood mapping and fraud profiling to predicting hurricanes and subsidence monitoring, data sources spring up on an almost daily basis - the question for insurers is, however, what to do with it and how to turn it into useful information.

How exactly do insurers use and apply this data responsibly when juggling privacy issues and the potential minefield of profiling? These were just some of the questions put to a panel of experts at a recent Post roundtable event in association with IBM Software.

The discussion was kicked off by Mark Humphreys, liability risk manager at QBE, who said that there was an "explosion of data" in terms of the amount that insurers can access but the biggest question was over the robustness of the data. "This means until insurers are aware that the data sources we are using are robust as well as accessible, then they are of relatively limited value," he said.

"At least with historical data - whether it is historical claims experience or loss history - we all know the issues. So you are almost able to build allowances into the modelling that you are using." He added that new data sources could lead to a "myriad of new problems".

Michael Lawrence, sales and marketing director at Highway Insurance, said that from a sales point of view a lot of data could be easily captured from the internet, allowing insurers to analyse those people who are not yet customers. "It presents you with a big opportunity to know which customers are insured and those that aren't," he said. "What is unique about them? What patterns can we find within them, what business can we take by altering our products just to fit those groups of customers?"

There was general agreement that there is a big difference between data and information - data may be flowing around the insurance industry in abundance but the key point is how can that be turned into useful information.

Nick Line, Markel International's chief actuary, said that from a Lloyd's point of view, there was not much data or information around but it was growing and starting from a particularly "low base".

Making data work

He added: "We often find we have lots of data and no information, and turning data into something that informs you is a real art. That is something that is a real challenge for us. We are trying to offer a product to customers to protect them and we are trying to get the most accurate feeling for them, and you can do that by knowing about them. So you are spotting patterns in data to produce information, which is something that we try to do a lot of."

Despite a belief in the wealth of data available, Budhi Singh, Arch Insurance's chief risk officer and actuary, said that perhaps this is merely the case for personal lines customers. He claimed that there were not many commercial arena data sources to choose from, and had only seen a couple recently come onto the market, one of which was Air Claims.

Mr Humphreys raised the question about data collection and analysis being merely reactive and in response to issues the insurance industry has experienced in certain sectors.

This was taken up by Brian Jewsbury, industry solution sales for IBM Software, who said: "Usually we find that many insurers pick up these types of opportunities as a result of major incidents like Buncefield where the companies that are selling data relating to geo-spatial stuff suddenly become a lot more active in the market."

Mr Jewsbury's colleague Bill Mew, IBM Financial Services' external relations manager, picked up on the point, saying it was important to differentiate between commercial data and proprietary data. He explained: "You have marine databases and other commercially available sources where it is in companies' vested interests to turn that data into useful information that they can sell as a service. However, with your own proprietary data, it is your responsibility to make sense of it and obtain value from it, and we are often not the best companies to look after and harness our own data."

A certain degree of reticence was identified by Robin Reames, claims director at online motor insurer Swiftcover, due to the data's expense and the fact that it was "continually being improved".

Graham Medcalf, UK and Ireland property manager at Chubb Insurance, wondered if the use of such data is inevitable in the current developing industry structure. Pointing out the obvious benefits of using data, such as giving insurers the opportunity to make more bespoke-targeted products, he said there was also an element of future compulsion.

Individual carriers

He said: "There is also the requirement of analysts and ratings companies to register how individual carriers are categorising themselves, such as how they are investing, how they are insuring, and how they are managing direct engagement. That is going to be a big change in the industry in the next 18 to 24 months, so the appetite for data and the manipulation of that information is going to increase.

"What will come through is that because ratings agencies are going to have far more interest in how companies are assessing their exposures, that method will become clearer and more transparent in the market. That is a factor more in our minds rather than how we can use it for innovation of the products and so on."

Mr Singh agreed with this point but said it was not just ratings agencies that were exerting pressure in a certain data-driven direction - it was regulators too. He said that regulators were starting to insist that insurers utilise certain models, for example, when evaluating catastrophes.

As such, data utilisation can be seen as one of those sink-or-swim issues for insurers, where those companies that adopt it earlier develop an insurmountable lead, while the laggards suffer immeasurable economic harm.

This point was raised and Mr Lawrence responded that perhaps, to avoid such a situation, data should be channelled through a cross-industry body like the Association of British Insurers. "As individual insurers," he said, "we should be using it through the ABI. It is there as our appointed spokesperson, and it should be taking data from all of us and saying that it can do something about this."

Mr Mew commented that there was potential for such an agreement to be implemented: "They can provide frameworks and models for the way in which the data is used, so that you can see some commonality - especially in areas like Carlisle where there is a cross-industry stake in seeking to make improvements and changes. If the frameworks for collecting data and asking questions about the materials used to rebuild becomes an issue for the industry, then there is possibly a role to be played there."

Third-party advances

Mr Humphreys concurred with this, saying the ABI had already made some progress in such fields, usually through third parties: "The ABI worked closely with the Health and Safety Executive on a corporate health and safety performance index, and the ABI was able to represent the industry in talking to them and saying these are the types of data sets that employers' liability insurers want to see." He added the difficulty had been communicating back to insurers how successful such schemes had been.

Returning to the issue of data utilisation, the panel considered whether the expense and scale of data handling could only be handled by an insurer of a certain size. Mr Reames commented: "It is going to have an impact only on the bigger players. I was reading somewhere that in the average supermarket, its RFID tagging generates something like seven terabytes of information just on what is on its shelves at any one time.

"It is such a phenomenal amount of information just on something that is fairly basic, and it is going to be extremely difficult for small players to be able to crunch that sort of data. I also think that historically some of the bigger insurers or those that maybe relied on older systems and on mainframe systems are going to find themselves challenged by the new players in the market, which are utilising technology and online capabilities. That is definitely going to pose a challenge to those more traditional players."

Niche roles

Mr Humphreys said that while smaller players would always have the ability to fill niche roles in the market, the onus will continue to be on technology and software providers to be flexible in their pricing: "Data should be available to everybody at reasonable prices."

He added that this would mean the market would avoid being dominated by the largest insurers just because they have the information. A counter-argument was then raised as to whether the process might have the opposite effect - whether larger players might suffer because they would be unable to upgrade their legacy systems, which cannot handle the data gathering and processing.

While data usage to evaluate information about customers can in principle be seen as a good thing, it should "carry a health warning", according to Mr Humphreys. He explained: "The moment you differentiate your top 10% or top 20% of your client base, you are also differentiating those who fall at the lower end. You are, in essence, building a framework that will penalise them, be that property insurers or employer liability insurers - that is the health warning, it is what you use the information for and how that is perceived to be used."

Mr Reames added there was a lot of data that simply was not being released by agencies: "There are enough challenges already in terms of the government agencies from the Department of Work and Pensions to Immigration and Customs and Excise - and those that are going to have access to the kind of information that may be held on identity cards - should they ever be introduced. You know we are seeing challenges from human rights, Charter 88 and others on the extent of that information and its availability, and is it likely to come out to the commercial sector? It is hard enough just to get information out of the Driver and Vehicle Licensing Agency."

Fraud convictions

He added: "There is a lot more data that we would love to get hold of, and it could be anything from previous driving convictions to convictions for fraud in general, which would help us in assessing the risks of certain individuals."

Mr Mew pointed out the tightrope that insurers would have to tread on with regard to personal liberty: "There is going to be a certain amount of data that we will never have that will always be personal, and there is going to be a certain amount of data that will always be available. Then there is going to be data in-between where we will be given access to that data under restricted circumstances. With that will come enormous responsibility and you will need to be a trusted brand in order to make use of and be trusted with that data - and woe betide any brands that misuse that data and lose that trusted status."


Chair: Stephanie Denton, supplements editor, Post

Mark Humphreys, liability risk manager, QBE

Brian Jewsbury, industry solutions sales, IBM software

Michael Lawrence, sales and marketing director, Highway Insurance

Nick Line, chief actuary, Markel International

Graham Medcalf, UK and Ireland property manager, Chubb Insurance

Bill Mew, external relations manager, IBM Financial Services

Robin Reames, claims director, Swiftcover

Budhi Singh, chief risk officer and actuary, Arch Insurance

  • LinkedIn  
  • Save this article
  • Print this page  

You need to sign in to use this feature. If you don’t have an Insurance Post account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an indvidual account here: