The Underwriter’s Gambit: Algorithmic Boosts in Complex and Specialty Risks
In Part 1, underwriter workbenches were described as platforms to support underwriters in making complex decisions at speed and with confidence. Workbenches come equipped with relevant insights about clients and the risk being underwritten, along with a modern user interface to enable frictionless data flow.
Historically, underwriting has been a nuanced aspect of insurance, seldom straightforward and hence important to get right. Pricing a policy to offer good customer value without compromising on profitability, makes or breaks an insurance business. With automated underwriting, efficiency of carriers’ decision making in risk selection and policy pricing, improved significantly. Owing to the extent of quantitative data available in underwriting, algorithms today play an important part. Algorithmic underwriting(AUM) uses computational algorithms, external data sources and big data to inform underwriting decisions.
However, the use of algorithms in underwriting should not be construed as absence of human judgment, but instead as an elevation of its role. Judicious underwriting will still be as vital as ever. The surge in algorithmic underwriting models in the last decade, inevitably raised new risks and uncertainties, such as the potential for unlawful discrimination and need for transparency. It is important to distinguish AUM from traditional actuarial models (e.g. valuation, pricing and ALM/hedging). AUMs rely on latest data science methods.
In complex risk, algorithmic underwriting (called smart follow / lead underwriting) is relatively new. The biggest opportunity appears to be in follow business as it is most amenable to being algorithm-led, followed by MGA and lead business. Though lead business remains expertise-led, a sizeable portion is expected to be subsumed by algorithms. Ki, an algorithmically-driven Lloyd’s syndicate launched in collaboration with Brit and Google Cloud, allows broker trading partners to access its algorithmic underwriting for follow capacity and write business. Ki’s algorithms evaluate Lloyd’s policies and automatically quote for business through a digital platform, which brokers can access directly. A new syndicate, Ki has plans to deliver additional platform capability and expand non-Brit Lloyd’s lead markets into new classes. A Portfolio Management function exists to monitor business written, to ensure a balanced portfolio with controlled aggregations.
Algorithmic trading has been successful in other financial markets and offers strong opportunity for insurance markets. Insurers can automatically analyze risk, offer personalized products and serve customers globally, at digital speeds.
In marine insurance, Concirrus offers algorithmically-driven underwriting, with which carriers can quote business instantly from anywhere, anytime. Its Quest Marine platform targets high-volume, transactional business so conventional underwriters focus on more complicated risks. The application combines submissions, analytics, pricing and live monitoring into automated workflows.
Risk appetite criteria is configurable, while risk assessment processes use behavioral data and market insights to calculate expected losses relative to line size. Existing data is augmented with feeds from Vessel Automatic Identification System, an internet-connected system for real-time telematics data — identification, position, speed, destination, ETA, navigation status. Information on weather, shipping route, geography, vessel and insurance further enrich the mix, for better risk understanding at highly granular levels.
Stepping into the distribution side, an interesting implementation of such underwriting capabilities is RGAX, subsidiary of life/health reinsurer RGA, which launched cloud-based engine UWPal. This SaaS engine allows underwriting rules to be continuously optimized for performance while incorporating latest medical insights. It has partners such as Pandora (digital fintech platform), that utilize UWPal to deliver go-to-market products faster without in-house underwriting.
The technology underling algorithmic underwriting while restricted by the amount and quality of data is advancing towards more complex algorithms, as digital sharing becomes widespread. The more the information underwriters receive, the more powerful will be the algorithms. Algorithmic underwriting will eventually become table stakes for maintaining competitive advantage.
In Part 3 of this series, the impact from alternative data sources is elaborated.