The data landscape has changed a great deal over the last decade, new technology, new concepts, new approaches, new techniques, new roles, new functions, new team names. Yet a lot of insurance pricing teams have not changed at all, using the same tools and approaches as they did a decade ago, if not longer.
Change for change’s sake is not the aim, but there are huge amounts of value that can be unlocked for pricing team’s by adopting some of the most relevant aspects.
In my blog post on comparing Pricing teams vs Data Science teams I describe the differences between how the two types of teams operate, but why is it that Insurance Pricing differs so much from other industries?
Domain Knowledge
When a pricing team is hiring, other than at the junior level, the role will very likely require n years experience in insurance pricing, the industry very rarely hires from others. Whereas in other industries, data engineers, scientists and analysts will be hired based on general skills with data science tools and technologies rather than domain knowledge, and so cross industry hiring is very common. This means that pricing isn’t hiring from the same pool of talent and knowledge that other industries do.
That’s not to say domain knowledge isn’t important; an analyst or data scientist in any industry will be able to build better solutions to more challenging problems if they have experience in that field than if they don’t. But the cost is that the insurance pricing isn’t hiring analysts with experience of modern approaches.
Upskilling through self-study
A key method of learning skills and technologies is learning them from other’s at work, however if none of the other analysts are using modern approaches or best practices, this won’t be the case.
Another key method to upskill is through self-study or personal projects. Data Engineers, Scientists and Analysts will generally work on personal projects using the skills they wish to learn in order to keep on top of advancements. Whilst plenty of pricing analysts are prepared to invest their time into upskilling, the default choice to pursue the Actuarial exams and qualifications, which don’t cover data tools, technologies or best practices.
Aversion to change
Anecdotally pricing analysts are generally reluctant to change their ways of working. Where the pricing profession has existed much longer than the term ‘data scientist’, many of the methodologies used in pricing have existed for several decades, and so many analysts are comfortable with these and see adopting new technologies as a risk, and therefore don’t explore the opportunities that new technology has to offer.
A key example of this is the use of machine learning, it is often seen as riskier to deploy a Gradient Boosting Machine rather than a hand crafted Generalised Linear Model, and so a pricing team won’t explore the data science field much further if not planning on utilising these – but this is just one part of the full model building process, the technology that can vastly improve the preparation of the modelling data, the validation, deployment, monitoring and refreshing the model is then missed out on.
Conclusion
The insurance industry is missing out on a lot of the potential value that advancements in the data landscape offers. Some potential explanations for why this is the case are that people with experience and knowledge of these advancements in other industries aren’t being hired by pricing teams, pricing analysts don’t tend to study new data tools and approaches in spare time, and experienced pricing analysts are often averse to changing their approaches and so don’t explore the data science field in depth.
