Transparency and technology are driving a revolution in catastrophe modelling, says James Lay.
A movement towards openness and transparency within catastrophe modelling is making (re)insurers more selective than ever before. Technological advances in this space now allow (re)insurers to trial and compare models, giving firms the tools to make more informed decisions about their risks.
The hurricane season in 2017 was one of the worst in decades, with global natural disasters causing total damage exceeding US$330bn - the second-highest recorded losses in history. Although catastrophe (cat) risk modelling plays a large part in preparing an insurer for risks such as this, client choice in catastrophe risk modelling has developed slowly during the past 20 years. This has been almost entirely dominated by a handful of vendors, which has resulted in reduced innovation and significant licensing costs. As a result, catastrophe risk models have acquired a reputation for being both costly and resource-heavy. Combined with the fact that, until recently, the industry lacked a simple option for evaluating models prior to purchase, this has made it difficult for (re)insurers to build a case internally for changing or adding new models. Consequently, the industry has simply accepted a limited view of risk - resulting in insurance not being offered, or made prohibitively expensive, in regions where (re)insurers do not have a view of risk. In a world where data is king and access to information is key, though, this is beginning to change, with the industry looking to make more informed decisions about risks.
Considering the pivotal role of insurance within the financial system and the damage caused by natural catastrophes, improving risk awareness should be a key focus for the industry. The issue is particularly important in emerging markets and parts of the world that are currently unmodelled and underinsured, making recovery following a natural disaster very difficult. There is a desire from both public and private companies to plug this protection gap in order to improve the development and adoption of catastrophe models in emerging markets. Without making it easier for firms to build a strong and compelling case for new model adoption, this will be difficult to do.
A movement towards greater transparency in the catastrophe modelling market signals a risk revolution, and modern hosting options and web-based technology are finally making it possible for insurers to trial models prior to purchase. This is enabling firms to compare and adopt models more easily. Historically, this has been prohibitively difficult because of intellectual property (IP) challenges from the model vendors' perspective and the significant IT and operational resources traditionally required to evaluate a model from the (re)insurers' perspective. However, new catastrophe modelling platforms are able to offer multi-model evaluations in one place without the need for any client-side installation, all while safekeeping the IP of the model vendor.
This kind of evaluation system enables (re)insurers to trial new models easily and with minimal resource requirements or costs.It also facilitates reciprocal feedback between model providers and (re)insurers to promote mutual benefits, promotes combined efforts and shared ownership for model validation and ensures model providers retain full control of their intellectual property.
The real value of this is the ability to compare two different vendor models in parallel. This has traditionally been challenging, since different vendor models operate on different platforms and are based on different financial calculation models, with varying abilities to deal with financial terms and conditions. Running two models of the same peril, on the same portfolio, can therefore produce different results simply because the underlying conditions are different. Being able to compare two models with the same underlying conditions enables (re)insurers to evaluate their model portfolio in a simplified way, making useful model comparisons to other models in the market to make sure they are using the best model for their needs, at the best cost point.
The actual process of model evaluations is straightforward. After a model evaluation agreement has been completed, the evaluating firm is given access to the platform via a secure web browser for global access. The firm can then log in and evaluate and compare model(s) at its own convenience for a set period of time. By leveraging evaluation systems like these, (re)insurers will be empowered to make strong business cases for new model adoption more effectively, which will play a vital role in validating risk models.
This was first demonstrated in 2015, with the Oasis evaluation of ARA's US hurricane model (Hurloss). Collaboration between ARA and the 26 participating Lloyd's managing agents meant that model validation questions could be shared across the companies taking part in the evaluation. The benefits of working in this way are twofold: first, the division of labour decreases the validation workload on an individual firm; second, it speeds up the model adoption process, which boosts investment into new models. This stimulates the demand for models in the market, encouraging new model development in both modelled and unmodelled regions.
If we take the Nordic region as an example, feedback from the Nordic Cat Risk Modelling Conference in 2017 highlighted the lack of flood models in the region, despite it being one of the costliest perils. Following this, JBA has started the development of such a model to fill this gap. We can also look to the Middle East, which until recently did not have an earthquake model, even though nearly 30 million people in the Middle East live in areas at risk of earthquakes. Commissioned by Lloyd's of London, CatRisk Solutions produced and released a Middle East earthquake model in 2017. A year on, the model is now being used by several global (re)insurers.
Model vendors can also benefit from offering model evaluations, given the opportunity they present for vendors to showcase their models and identify improvements based on client feedback. This helps strengthen their reputation in the market and therefore attract more demand for their models. It is anticipated that this should strengthen the quality of models in the market, making access to insurance more widely available and recovery from natural disasters more manageable - even in areas of elevated risk. Surprisingly, until recently, there was no unified source listing of the catastrophe models available in the market. Oasis Hub has made headway in compiling such a list, but without contribution from all the market leaders it cannot yet be considered comprehensive. This highlights another key issue in the marketplace around transparency - not just around how models are built, but around the information that is available about catastrophe models. As a result, the market is driven by demand, as there is limited data available on the global model availability, and by extension the actual model coverage gap.
The insurance industry, traditionally known for its conservatism, is developing, and technology is at the forefront of innovation. By leveraging modern technology, the insurance industry is making a fundamental shift away from legacy systems towards greater process and resource efficiencies. With natural disasters becoming increasingly frequent, it is essential that (re)insurers use the best tools available to assess their risks. And (re)insurers open to adopting new technology will benefit from greater choice than ever before.
Enabling (re)insurers to trial, compare and adopt alternative models more easily will stimulate the supply of catastrophe risk models in the industry, improving how it views and deals with catastrophic events in the future. In order to fill the protection gap and mitigate against future catastrophe losses as much as possible, the industry needs more views of risk, and model evaluations now provide a real commercial incentive for new and existing vendors to develop more models.
James Lay is commercial director of ModEx Simplitium.