Welcoming the ‘bots’ into reserving

While re/insurers have opened the door to forms of so-called insurtech in various areas of their businesses, few have yet been willing to let these relatively new techniques and technologies loose on their all-important reserving departments where countless actuaries still toil doing jobs much as they would have a decade ago.

Yet tools are increasingly available that can make a big difference to re/insurers and help them in multiple ways, says Bill Miller, a managing director in the Advisory practice of KPMG in Bermuda, who argues that the process of what he calls ‘reserve modernisation’ is well under way and that re/insurers who embrace it first will gain a competitive advantage.

“Reserving is the holy grail when it comes to the possibilities of insurtech,” Miller says. “Companies want and need close control over their reserves but such is the squeeze on companies’ expenses and costs, given the lingering soft market driven by excess capacity, companies are increasingly open to anything that can improve efficiencies.

“There are some huge advantages if certain technologies are applied to reserving practices and companies that do that will gain a competitive edge. The fact is, most insurers’ approach to reserving has remained unchanged for decades despite the emergence of so many technologies—very little has changed.”

Miller describes ‘reserve modernisation’ as “a new vision that can drive reserving and enterprise performance as a result”. He adds: “It can enhance the efficiency of the process while maintaining accuracy. Reserves are a critical issue for any company and if there is a better way of doing it, re/insurers must be open to that.”

A big impact

He identifies three areas of technology that can enhance the reserving process in different ways; used in combination they can have a big impact. They are cognitive reserving, data visualisation and robotic process automation. The enterprise’s response, and what it then does with the information it has been given, can be the most important factor of all, he adds.

“Cognitive reserving is the application of machine learning techniques to claims valuation in a way that leverages new statistical software capable of processing both structured and unstructured data to identify trends in claims,” Miller explains.

To offer a real-world example, he says this means that the software can assess the notes (unstructured data) made by claim adjusters and identify patterns and trends.

“In one case I worked on, it found that the number of times the claims adjuster used the word ‘pain’ in her notes was directly correlated to the outcome of a claim—in this case in an adverse way.

“Clearly this was a trend the human mind was unable to spot but artificial intelligence (AI) could,” Miller says. “You find things you do not expect.”

He describes data visualisation as effectively using technology to better present and illustrate results, drilling down further into the data and the analysis of that in the process.

“This is important because actuaries can sometimes struggle to get their message across to the executives making business decisions, despite the importance of their data,” he says.

“Robotic process automation is the automation of tasks that actuaries have historically done manually but in a way that can also yield outcomes and analysis.

“These are not real robots but very sophisticated software or bots that can operate across multiple systems to identify relevant data, process it quickly and allow analysis. It is much more efficient and can mean that fewer actuaries are needed as a result,” Miller says.

Finally, he argues, the nature of the enterprise’s response to all this—how the data and analysis are then used to make business decisions in the real world—is the real key.

“If the decision-makers buy into all this, it can have a tangible effect on the way pricing is determined, underwriting completed and claims processed,” he says. “It can mean better coordination across multiple business units, linking what the reserving department is learning in real time to business and underwriting decisions being made. Better data means better risk selection and pricing.”

Overcoming the fear factor

Miller explains that the benefits to re/insurers embracing this system are threefold: they can cut costs, improve the efficiency of reserves (because they will be more accurate on each line of business), and also generate results that can improve the performance of the underwriting side of the business.

The biggest inhibitor to more companies embracing technology is the fear that it might get things wrong. “Yet the process at the moment is prone to all sorts of human error at many levels,” he says. “We suggest running any system in parallel with existing processes for a while so that differences can be explored and faith is built in the new system.”

KPMG has been through the process of applying all these techniques to the reserving department of one large, global insurer. He says the outcomes were “tangible and remarkable”. In just 10 weeks, some 18 percent of the work of analysts was automated, and it was forecast that up to 50 percent was possible in the long term.

“We configured a bot to complete eight of 20 manual tasks analysts were completing during the reserving process, allowing the information to be available more quickly with better analysis,” Miller says.

KPMG is advising companies wanting to invest in such technology. It can set up the most effective systems for insurers to run themselves in the long term and enjoy the advantages in-house.

He says companies with multiple long-tail lines would be the principal beneficiaries of such technologies but any company can benefit if the systems are used correctly.

Bill Miller is a managing director in the Advisory practice of KPMG in Bermuda. He can be reached at: billmiller@kpmg.bm