60 seconds with Luzi Hitz

When Windstorm Kyrill caused widespread damage across Europe in January 2007, affecting the UK, Belgium, the Netherlands, Germany, Austria, Poland and the Czech Republic it caused insurance market losses of more than €3bn.

Delegates at the Chief Risk Officer (CRO) Forum – held shortly afterwards – called for an independent reporting agency to issue event loss data following major pan-European events. Thus the PERILS index – aimed at collecting and sharing catastrophe exposure and loss data across Europe – was born. Two and a half years into the initiative, Luzi Hitz, Chief Executive of PERILS AG, considers its progress.

What is the PERILS index and who is it for?

The idea was to create more data transparency for the entire market and that this would lead ultimately to a better understanding of the risk. We have focused on the major European markets which are likely to be impacted by a major storm event crossing Europe beginning with nine countries: Belgium, Denmark, France, Germany, Ireland, Luxembourg, the Netherlands, Switzerland, the UK and recently adding Norway and Sweden. 

How can European windstorm risk be better understood and modelled using industry loss data?

It’s important to note that PERILS collects two kinds of insurance data from the market. It collects sums insured (the data showing what assets are exposed to catastrophe perils) and the event loss data post a big event. In combination, these two data sets represent really big data pools which can be used to calibrate models. So catastrophe modellers can use this market data to benchmark the assumptions they have in their models, which helps to make the risk assessment more reliable over time.

In addition, insurance companies can measure their own portfolio data against the market as a whole, not just in terms of premium but also in terms of sums insured. After an event they can measure their own loss against the market event loss and this can lead to improved portfolio steering over time.

Have you been able to use real events to test the index?

Since our foundation in January 2009 there have been two big windstorm events – Windstorm Klaus in 2009 and Windstorm Xynthia in February last year. We captured these two events and produced the industry loss data at the full granularity – not just producing one number for an event but breaking down the industry losses by country and CRESTA zone (these are the two-digit zip-codes across Europe). So it was a significant geographical resolution. The industry losses were also broken down into property lines of business, such as private property or commercial lines property. We have done the same analysis for five older events: Anatol (1999), Lothar (1999), Martin (1999), Jeanett (2002) and Kyrill (2007).

How do your results for these events compare to those from other industry bodies?

We did a comparison with organisations that use a similar approach to us, those which base their industry loss estimate on collected data and not on modelled data – such , such as the data from national insurance associations, Munich Re Nat Cat and Swiss Re Sigma. We concluded that there was a very good agreement in individual event loss numbers, so we were very pleased with the results.

How does your approach compare with similar initiatives in the US?

There is really only one other organisation in the world that does similar work to us and that is Property Claim Services (PCS). The difference is that the PCS only collects event loss data whereas we collect event losses and sums insured. We also collect data at a higher granularity – at the CRESTA zone resolution. This granularity allows the insurance market to use our industry loss data in insurance risk transactions that are based on weighted industry losses per country or CRESTA, thereby reducing the basis risk significantly.

How can the index be used to facilitate risk transfer to the capital markets?

One original motivation was to facilitate the industry loss-based risk transfer via collateralised reinsurance such as industry loss warranties ( ILW) s [industry loss warranties] or ILS [insurance-linked securities] (ILS) and this has taken off very well over the past two years. At the moment we have more than $2bn of limits being placed using PERILS data as a trigger, including six catastrophe bonds. Prior to our existence there really was no independent industry loss data that could be used in these kinds of transaction.

Are there any other types of hazard that could be covered by similar insurance industry loss and exposure indices?

The world is a big place and there are certainly other areas where you could do similar things. The need is always greatest where it hurts most, by which I mean where there is a lot of insured loss potential. If you look at the areas in the world where the industry has a lot at stake obviously the global map is dominated by the US and PCS covers this market. Then we have Europe with the biggest insurance loss potential coming from windstorms, which we are covering now.

There are major exposures in other geographical areas, such as Japan, Australia and New Zealand and within Europe there is also flood and earthquake loss potential. In all these territories the work we do is probably highly beneficial to managing the insured risk. Ultimately what we provide is more transparency and this facilitates a better understanding of the risk and more liquidity and stability in managing these risks.