Mohsen Rahnama, Cihan Biyikoglu and Moe Khosravy of RMS look to 2029, consider the changes the (re)insurance industry will have undergone and explain why all roads lead to a platform
Over the last 30 years, catastrophe models have become an integral part of the insurance industry for portfolio risk management. During this time, the RMS model suite has evolved and expanded from the initial IRAS model — which covered California earthquake — to a comprehensive and diverse set of models covering over 100 peril-country combinations all over the world.
RMS Risk Intelligence™, an open and flexible platform, was recently launched, and it was built to enable better risk management and support profitable risk selection. Since the earliest versions of catastrophe models, significant advances have been made in both technology and computing power. These advances allow for a more comprehensive application of new science in risk modeling and make it possible for modelers to address key sources of model and loss uncertainty in a more systematic way.
These and other significant changes over the last decade are shaping the future of insurance. By 2029, the industry will be fully digitized, presenting even more opportunity for disruption in an era of technological advances. In what is likely to remain a highly competitive environment, market participants will need to differentiate based on the power of computing speed and the ability to mine and extract value from data to inform quick, risk-based decisions.
So how did we get here? Over the past few decades we have witnessed several major natural catastrophes including Hurricanes Andrew, Katrina and Sandy; the Northridge, Kobe, Maule, Tōhoku and Christchurch Earthquakes; and costly hurricanes and California wildfires in 2017 and 2018. Further, human-made catastrophes have included the terrorist attacks of 9/11 and major cyberattacks, such as WannaCry and NotPetya.
Each of these events has changed the landscape of risk assessment, underwriting and portfolio management. Combining the lessons learned from past events, including billions of dollars of loss data, with new technology has enhanced the risk modeling methodology, resulting in more robust models and a more effective way to quantify risk across diverse regions and perils.
The sophistication of catastrophe models has increased as technology has enabled a better understanding of root causes and behavior of events, and it has improved analysis of their impact. Technology has also equipped the industry with more sophisticated tools to harness larger datasets and run more computationally intensive analytics. These new models are designed to translate finer-grained data into deeper and more detailed insights. Consequently, we are creating better models while also ensuring model users can make better use of model results through more sophisticated tools and applications.
In the last decade, the pace at which technology has advanced is compelling. Emerging technology has caused the insurance industry to question if it is responding quickly and effectively to take advantage of new opportunities. In today’s digital world, many segments of the industry are leveraging the power and capacity enabled by Cloud-computing environments to conduct intensive data analysis using robust analytics.
Technology has also equipped the industry with more sophisticated tools to harness larger datasets
Such an approach empowers the industry by allowing information to be accessed quickly, whenever it is needed, to make effective, fully informed decisions. The development of a standardized, open platform creates smooth workflows and allows for rapid advancement, information sharing and collaboration in growing common applications.
The future of communication between various parties across the insurance value chain — insurers, brokers, reinsurers, supervisors and capital markets — will be vastly different from what it is today. By 2029, we anticipate the transfer of data, use of analytics and other collaborations will be taking place across a common platform. The benefits will include increased efficiency, more accurate data collection and improvements in underwriting workflow. A collaborative platform will also enable more robust and informed risk assessments, portfolio rollout processes and risk transfers. Further, as data is exchanged it will be enriched and augmented using new machine learning and AI techniques.
We continue to see technology evolve at a very rapid pace. Infrastructure continues to improve as the cost of storage declines and computational speed increases. Across the board, the incremental cost of computing technology has come down.
Software tools have evolved accordingly, with modern big data systems now capable of handling hundreds if not thousands of terabytes of data. Improved programming frameworks allow for more seamless parallel programming. User-interface components reveal data in ways that were not possible in the past. Furthermore, this collection of phenomenal advances is now available in the Cloud, with the added benefit that it is continuously self-improving to support growing commercial demands.
In addition to helping avoid built-in obsolescence, the Cloud offers “elasticity.” Elasticity means accessing many machines when you need them and fewer when you don’t. It means storage that can dynamically grow and shrink, and computing capacity that can follow the ebb and flow of demand.
In our world of insurance and data analytics, the macro cycles of renewal seasons and micromodeling demand bursts can both be accommodated through the elastic nature of the Cloud. In an elastic world, the actual cost of supercomputing goes down, and we can confidently guarantee fast response times.
A decade from now, the industry will look very different, not least due to changes within the workforce and the risk landscape. First-movers and fast-followers will be in a position of competitive advantage come 2029 in an industry where large incumbents are already partnering with more agile “insurtech” startups.
The role of the intermediary will continue to evolve, and at every stage of risk transfer — from insured to primary insurer, reinsurer and into the capital markets — data sharing and standardization will become key success factors. Over the next 10 years, as data becomes more standardized and more widely shared, the concept of blockchain, or distributed ledger technology, will move closer to becoming a reality.
This standardization, collaboration and use of advanced analytics are essential to the future of the industry. Machine learning and AI, highly sophisticated models and enhanced computational power will enable underwriters to improve their risk selection and make quick, highly informed decisions.
And this ability will enhance the role of the insurance industry in society, in a changing and altogether riskier world. The tremendous protection gap can only be tackled when there is more detailed insight and differentiation around each individual risk. When there is greater insight into the underlying risk, there is less need for conservatism, risks become more accurately and competitively priced, and (re)insurers are able to innovate to provide products and solutions for new and emerging exposures.
Over the coming decade, models will require advanced computing technology to fully harness the power of big data. Underwater robots are now probing previously unmapped ocean waters to detect changes in temperatures, currents, sea level and coastal flooding. Drones are surveying our built-up environment in fine detail. Artificial intelligence and machine learning algorithms are searching for patterns of climate change in these new datasets, and climate models are reconstructing the past and predicting the future at a resolution never before possible. These emerging technologies and datasets will help meet our industry’s insatiable demand for more robust risk assessment at the level of an individual asset.
This explosion of data will fundamentally change the way we think about model execution and development, as well as the end-to-end software infrastructure. Platforms will need to be dynamic and forward-looking verses static and historic in the way they acquire, train, and execute on data.
The industry has already transformed considerably over the past five years, despite traditionally being considered a laggard in terms of its technology adoption. The foundation is firmly in place for a further shift over the next decade where all roads are leading to a common, collaborative industry platform, where participants are willing to share data and insights and, as they do so, open up new markets and opportunities.
The analytical and computational power of the Risk Intelligence (RI) platform enables the RMS model development team to bring the latest science and research to the RMS catastrophe peril model suite and build the next generation of high-definition models. The functionality and high performance of RI allows the RMS team to assess elements of model and loss uncertainty in a more robust way than before.
The framework of RI is flexible, modular and scalable, allowing the rapid integration of future knowledge with a swifter implementation and update cycle. The open modeling platform allows model users to extract more value from their claims experience to develop vulnerability functions that represent a view of risk specific to their data or to use custom-built alternatives. This enables users to perform a wide range of sensitivity tests and take ownership of their view of risk.
Mohsen Rahnama is chief risk modeling officer and executive vice president, models and data, Cihan Biyikoglu is executive vice president, product and Moe Khosravy is executive vice president, software and platform at RMS