Big Compute and Big Data in Insurance
The insurance sector is undergoing substantial transformation, driven by regulators, InsurTech, and changes in customer behavior. One way to react to these challenges is to embrace digital transformation technologies. These help accelerate the rate of change, increase business agility, improve customer experience, and evolve existing business models. When implemented well, digital transformation technologies can help organizations adapt to the current economic climate and meet ever-increasing customer expectations.
In this piece we look at two technology trends disrupting the industry – “big data” and “big compute”.
Big Data in Insurance
The ability to store and process vast amounts of data is of particular interest to the insurance industry.
One obvious current example is the rise of telemetry for car insurance. Many younger drivers are happy to reduce the cost of their policy by allowing insurers to see their driving data. This means an increasing need to capture, store, and analyze driving telemetry in near real-time from onboard diagnostic systems or smartphones.
A 2017 report by Willis Towers Watson on US Property and Casualty looked at how insurers use predictive models and big data. They reported that 67% of insurers surveyed already use predictive models for underwriting and risk selection. When asked how they expect to use big data in the next two years, the vast majority reported it would help with pricing, improve loss control and claim management and help improve management decisions.
The same survey identified several significant challenges to bringing about this new analytics utopia:
- The lack of availability of people with the right skills and capabilities, which can be solved by upskilling existing employees
- Data capture and availability, which can be solved by hiring more employees with the right skills
- Cost consideration and funding, data quality and reliability, which can be addressed by outsourcing.
Ingesting and storing data is the first technical challenge in creating an advanced analytics program, and this can be solved with services like Azure Event Hubs and Azure Data Lake. But the real value comes from data mining and creating novel analytics for insights.
Insurers are already creating predictive models for risk and using anomaly detection models to identify fraudulent behavior. Insurers are using Machine Learning to examine data mined from social media, wearables, internet of things, and telematic devices, to produce more tailored risk assessments. In all of this, one trend is very apparent. Wherever decisions are made, there is an opportunity for predictive modelling to provide deeper insights, faster.
Big Compute in Insurance
Many of the popular Big Data technologies are geared towards real time or batch analytical workloads. But some financial modeling scenarios need a different approach to deal with their computational requirements.
For example, calculating weather risk, a typical catastrophe model might attempt to use 150 years’ worth of meteorological data. That represents ~ 30 terabytes of data, over 100M locations, while simulating approximately 100,000 atmospheric and hydraulic scenarios.
This yields 200 billion records to feed the financial models. These types of large scale, elastic compute, on demand scenarios are where the cloud excels.
Extremely large-scale compute jobs like these can be run on standard virtual machines. Additional performance and cost optimizations can be achieved by choosing different Azure VM SKUs such as the NC series which provide GPUs. These GPUs can be very useful in things like nested stochastic calculations. For example, in a static hedge simulation, containing over 680 million paths, a single GPU has a throughput of 600,000 paths per second. The complete simulation takes less than 2 minutes to run on 10 GPUs.
Looking further into the future, Quantum Computing promises to deliver computing power that can solve complex problems that would take current systems years to solve, in a matter of hours.
Digital transformation is redefining the insurance industry. By unlocking digital experiences for customers and employees, insurers can enable unique insight and engagement to drive measurable differences from their peers in underwriting performance, loss ratios, lapse rates, and churn.
Learn more by downloading Microsoft Perspectives on the Digital Insurer
