Describe Data analyses underwriting risk data for complex specialty risks (D&O, Cyber and Terrorism).
The company launched in August 2018 and was a member of the second Lloyd’s Lab cohort in April 2019. Its founders have deep insurance and IT experience including actuarial, data science, insurance IT management (RenRe) and a PhD in quantitative finance.
The idea for the business came while the team was consulting for insurers on complex risk analysis. They identified increasingly similar data and analysis issues among several clients and determined that they could build a product to address these problems.
The business’s risk engine uses data from public, industry and academic databases to supplement underwriting data. The team says they are strong proponents of ‘bionic underwriting’, combining the expertise and judgement of underwriters with enhanced technology.
Describe Data’s edge in the specialty space is that it uses advanced data analytics and Bayesian statistics to provide risk analysis where little information is available.
In one example, D&O underwriters at Lloyd’s wanted to see financial data and litigation history for specific companies and market segments. The team created a dashboard with company news, a key event timeline and current trend analysis. To supplement the available data, they used quant finance techniques to model frequency and severity, giving underwriters a more accurate view of the likely profitability of a deal.
The underwriters also wanted to see how a deal would fit into their existing portfolio so Describe Data built a fully customisable portfolio analysis tool that allowed them to heat-map their portfolio according to specific criteria. The team built and executed these tools within a few weeks without the use of any client-specific data.
THE OXBOW PATNERS VIEW
We have previously referenced a trend towards commercial and specialty insurance in technology activity. Describe Data’s proposition is similar to that of Pharm3r – an Oxbow Partners InsurTech Impact 25 member and healthcare analytics business frequently referenced on this blog – in that it seeks to provide a quantitative underpinning to a historically judgement-based discipline.
The challenge, of course, is linking signals of heightened risk to loss frequency or severity. In some areas this may be relatively straightforward (e.g. adverse drug trial outcomes must be reported to databases which Pharm3r can access) but in other areas such as D&O the existence or absence of a claim may never be known. For that reason, Describe Data’s predictive modelling approach is important, and echoes the approach taken by the catastrophe modelling businesses who are now increasingly looking at liability loss profiles (e.g. AIR’s Arium). The former D&O underwriters in our team believe this kind of insight is long overdue. Clearly it will be hard ever to build a ‘perfect’ solution for many specialty lines, but we like the way COO Gerard de Vere (also founder of InsurTech Ireland) describes the strategy: “better to solve 60% of underwriting issues today than waiting to solve 100% of them”.