Underwriters ‘better make the right automation choice’ or risk ‘serious systematic error’

Over the course of the Covid-19 pandemic, many insurance firms have opted to look inwardly at their own operational processes, seeking to drive digital efficiencies at a time when face-to-face interaction and manual activities stalled due to social distancing.

One key technological contender rising from the Covid-strewn ashes here is automation - and in particular, automated underwriting, which Re:infer co-founder and chief executive Edward Challis described as “helping your experts to be experts rather than glorified clerks”. 

According to Investopedia, automated underwriting “is a technology-driven underwriting process that uses algorithms instead of human beings to make underwriting decisions that are quicker and less prone to errors”.

The benefits of automated underwriting are multifaceted, according to Aventum’s chief digital distribution officer Tony Lawrence.

He explained: “Automating underwriting potentially facilitates insurance products to be bought anytime, anywhere, so the main benefit is capturing more opportunity, driven by customer needs.

“It means insurers [and] MGAs can potentially deliver value faster and more efficiently. It leads to increased sales and cost savings through innovation, whilst also increasing productivity. It ultimately frees up time for underwriters to actually underwrite the more complex risks, rather than [handling] administrative tasks.

“Automating underwriting will improve expense ratios and improve underwriting performance. Automation creates opportunities to link underwriting platforms to external data sets that can augment the underwriting at point of sale. Underwriters are therefore more informed when underwriting, which leads to better underwriting performance.”

For Challis, “automation improves underwriting in two main ways” following effective integration. “It enables faster underwriting through automated responses and [eliminates] time-consuming manual tasks,” he said.

“Automation also creates smarter underwriting through data-driven decision-making, freeing [up] underwriters to focus their expertise where it’s needed most.”

Chief executive of insurtech Concirrus Andrew Yeoman added that where automation has a “real big role to play” is around keying in data and handling mundane risks – this is because underwriters generally don’t see these tasks as “rewarding work”.

He continued: “Whether that be handling submissions or automatically pricing, quoting, binding risk – whatever level that takes, I think that’s absolutely a value in the process.”

Mike Scott, engineering manager at insurance platform Aventus, agreed: “Computers are amazing at repetitive tasks. Most underwriting tasks are repetitive - let computers do the bit they’re best at.”

Resistance to change

However, as in many industries, “overreliance on automation can create discrimination through machine driven bias [and] potential systematic decision error due to wrong coding design,” said Philippe Gouraud, chief executive of MGA Rising Edge.

Gouraud highlighted a specific example around credit analysis when selecting risks in directors’ and officers’ (D&O) liability insurance.

He explained: “If you were to use simple financial analysis metrics to automate risk selection, many ‘good risks’ would be eliminated due to ‘poor’ financial metrics.

“A startup that gets funding does not present a pretty balance sheet, yet it can have a robust operating model and [an] excellent disclosure practice – a credit agency model will not pick up these positive D&O risk attributes.

“In other words, if you will automate – you’d better make the right automation choice, or you will face a serious systematic error in underwriting.”

Another potential barrier to automation’s use in the industry is “the inherent human resistance to change”, added Gouraud. This sentiment is echoed by Scott, who has seen “hesitancy and a lot of patch protection” around automation.

Challis additionally noted: “People resist what they don’t understand and that goes for even the most skilled and senior underwriters.”

Avoiding wasted investment

Considering these adoption challenges, how can the insurance industry successfully integrate automation across underwriting?

“Automation can be deployed, but if underwriters don’t know how to use it then it’s a wasted investment – that’s why proper training and introducing automation gradually is so important,” Challis continued.

Automated underwriting also “has to be driven from the top down”, noted Lawrence. For example, the company’s board and the operating model of the business has to support automated underwriting processes. 

Lawrence added that businesses tend to prioritise solving internal problems with automation rather than improving the customer journey.

He said:  “Everything that is automated must be built around the customer first.

“Underwriting can be complex and to automate it, products need to be broken down to be digital native and operating models need to be challenged, radically changed and simplified.”

‘Newly refocused priorities’ 

The question that continually follows automation, however, is the impact on underwriters themselves - if a third of an underwriter’s day is automated, for example, what other responsibilities can they pick up?

Yeoman said: “The biggest con that people have to deal with is their fear of being made redundant or sidelined. For me, the key to this is developing a view of what are people going to do with the time that is freed up.”

To this point, Yeoman added that underwriters can look at their portfolios and begin to prospect new business to write with their newfound time. “I think that’s where you start to get the real value,” he said.

Gouraud agreed with Yeoman’s perspective, noting that while “there will always be a skill and an art to underwriting, in the future, the role of the underwriter will change”. 

“We anticipate it will be more skilled at intuitive risk analysis to challenge the machine’s output in return for less repetitive data entry or documentation production,” he added.

Challis highlighted that “insurers will also have to think carefully about how they measure the success of automation” because data aggregation can be an issue depending on the location and “scaling automation will be difficult in regions where publicly accessible data banks are unavailable”.

He continued: “Most [key performance indicators] will be expense loss ratios, but these will take time to bear fruit. 

“Interim measurements should be introduced to assess processes, underwriting efficiency and the value of underwriters’ newly refocused priorities.”