Mark Weatherhead, Head of Model Development - International
2016 marks the 60th year of the Monte Carlo Rendez-Vous. Since the first event in 1957 the insurance world has changed significantly, with economic and insured losses from natural catastrophes such as floods and hurricanes increasing dramatically.
This trend has in turn increased demand for improved measurability and risk modelling to allow carriers to better underwrite and manage exposures and to create a more reliable view of a company’s total risk exposure.
On a global basis, the key drivers of the increase in insured losses have been population growth, urbanization and increased insurance penetration. Other contributing factors, including climate change, may also have been an influence.
In Europe, population growth has been more modest than in other parts of the world, with the continent’s population increasing by 27 percent since 1957 from 580 million to 740 million. In comparison, the total world population increased by 161 percent from 2.8 billion to 7.4 billion during the same period.
But although its numbers have grown by less than the rest of the world, Europe has experienced a steady rise in urbanization that has led to increased concentration of insured risk.
While there is still uncertainty regarding the degree of impact from climate change, the likely consequences are subject to modelling. Increasing global temperatures are projected to impact the mid-latitude and polar regions more than the tropics.
This is expected to lead to two outcomes. First, the temperature gradient between the poles and the equator will drop, which will reduce storm formation. Second, the higher overall temperature will increase evaporation and the amount of water vapor in the atmosphere will lead to more intense storms. The projected impact of these two changes is increased frequency of both extreme rainfall events and extended dry periods in Europe.
As a result of these and other factors, the Intergovernmental Panel on Climate Change has identified three main areas of risk for Europe:
- That increases in sea levels, coastal erosion and peak river discharges, combined with increased urbanization, will lead to greater economic losses as a direct result of flooding.
- A reduction in water availability due to increased abstraction from river and groundwater and increased evaporative load will result in increased risk of drought conditions, especially in southern Europe.
- An increased occurrence of extreme heat events will impact human health, crops and the generation of wildfires.
In view of these projections, (re)insurers are working hard to gain a better understanding of their natural catastrophe risk through the use of catastrophe models.
There are two main lines of development. First, (re)insurers need to gain a deeper understanding of how the existing commercial vendor models perform for their book of business, how the models can be best modified or customized to better reflect their own loss history, and how the models compare to each other. Second, (re)insurers need to continue to address gaps in knowledge that are not adequately covered by existing models.
As this is an expensive, complex and time-consuming effort, the current trend is for companies to choose a single platform as their primary model and then complement where necessary with a secondary model.
We have seen greater interest in using the Oasis Loss Modelling Platform as a secondary modelling platform. It provides an open-source way of bringing models from many providers onto a common framework - including smaller commercial and academic teams and in-house developed models - and allows the insurer to do so without recourse to a third party.
As windstorms and earthquakes are modelled by more than one vendor, the biggest challenge for Europe is flood. Europe is somewhat unusual in that it has countries with small geographies and comparatively large rivers. Single flood events can hit more than one country and create correlation issues for insurers and reinsurers. The most notable example of this is the River Danube, which touches 10 countries as it makes its way from Germany to the Black Sea.
Flood poses particular challenges for modelling. In particular, the hydraulic modelling of flood extents requires a high-quality digital terrain model that has been processed specifically for the task by removing features like bridges.
The computational demand for hydraulic modelling is very high, especially as the size of the area to be modelled increases and flood is also one of the few perils that humans can directly influence through the construction of defense structures. Unfortunately, detailed data on the presence, construction standard and operational regimes of defenses is not universally available, and so considerable effort is needed by modellers to quantify this aspect.
Finally, as flood damage occurs in a fairly binary manner - either you are in the water or not - we need to have very accurate information on the location of the risks to be modelled, which is often lacking, especially in developing regions.
Despite these challenges the first models for flood in Europe began appearing in 2004. While commercial vendors have been slow to address the gap so far, others, including brokers, have been steadily producing models.
Guy Carpenter has produced a range of flood models for key countries and a pan-European hailstorm model based on detailed claims data.
The main challenge moving forward for companies is to bring the broad range of models available together into a single core platform, or possibly two, so that insurers can establish a comprehensive view of flood risk.