Why are reinsurers interested in climate research?(Formerly Poster JP6.3)

Over the past 35 years extreme weather events have killed hundreds of thousands of people and produced 25 of the top 30 insured losses for the property catastrophe reinsurance industry. Clearly, a better understanding of how extreme weather events vary with climate would benefit society as well as the reinsurance industry. While most people have no problem agreeing with the previous statement, there can be a problem with demonstrating how a “better understanding” can be used to benefit society or other entities such as the reinsurance industry. Below I make a small attempt at showing what type of information is needed for the reinsurance industry to benefit from a better understanding of how extreme weather events vary with climate. I provide some background on the reinsurance industry and how it operates to provide a context for understanding the type of information that would be of interest and when the information would be of use. I then discuss some of the climate-related research that is supported by companies in the (re)insurance industry that sponsor the Risk Prediction Initiative (RPI). I end by discussing other research topics that would be of interest to the (re)insurance industry.

Landfalling hurricanes in the United States and typhoons in Japan were responsible for 13 of the 25 most costly extreme weather events. Hurricane Andrew in 1992 caused the largest single property catastrophe (cat) insured loss from weather-related events, nearly $21 billion in 2003 dollars. Wind storms in Europe accounted for 4 of the 25 losses, but of these storms three were in the top ten insured losses. Other extreme weather events, in particular tornadoes, hail storms, and severe winter storms, can also cause large insured losses. Tornadoes and hail storms can be extremely intense and damaging but insured losses tend to be limited by their small scale relative to tropical cyclones and wind storms. Flooding also causes large insured losses, particularly in Europe.

The largest insured loss (over $21 billion for property and business interruption losses) was caused by the 9/11 terrorist attacks. The 1994 Northridge earthquake caused over $17 billion in insured losses and was the third most costly property cat insured loss. There is nearly a $10 billion difference between the losses from the Northridge earthquake and Typhoon Mireille, the fourth most costly loss ($7.6 billion). Relatively “small” differences (less than ≈$1 billion) separate the remaining events. The 30th largest event (tornadoes in the U.S.) produced an insured loss of $1.7 billion.

Insurance companies are designed to cope with the usual “random” events that are a catastrophe for an individual. Common examples are the loss of your car in an accident or of your home in a fire. If significant numbers of these events are correlated then insurers can be responsible for paying out large amounts of money. Extreme events such as earthquakes of landfalling hurricanes cause multiple houses and cars to be damaged and lost at one time. The losses then become catastrophe to an insurer and a company will often require their own type of insurance: reinsurance. Thus, reinsurance is essentially insurance for insurance companies. Property cat reinsurers operate on a global scale and diversify their portfolio by hazard and location.

A reinsurance contract, or treaty, generally lasts for a single year. Most contracts start on January 1. This business cycle imposes a significant hurdle for the use of seasonal forecasts. If a seasonal forecast is going to be used by most reinsurers it must be available by the start of December. This hurdle is not so large for European winter storms, but it is significant for hurricane wind reinsurance in the U.S.

Most (re)insurance companies use risk models for their business decisions. Essentially all risk models are based on climatology and don't account for climate variability that might be a significant factor for a given hazard. Consider, for example, how the El Niño-Southern Oscillation (ENSO) influences the frequency of hurricanes in the Atlantic basin. El Niño events are associated with lower hurricane frequencies, in part because of increased atmospheric shear in the Atlantic. Climate science will be of greater use to the insurance industry when risk models can account for climate variability. Nevertheless, the business cycle could still overwhelm the roll of climate variability in pricing decisions.

The RPI funds a variety of research related to landfalling tropical cyclones and European wind storms. The research focuses on the most intense storms and those that affect land because the companies that sponsor the RPI don't lose money from weak storms or storms that remain at sea. The RPI's interests are also U.S.-centric because the U.S is the largest insurance market. The RPI has yet to support research related to earthquakes in part because our small funding signal would be lost within the large amount of federal support.

The reinsurance industry operates on an annual time scale. Thus, climate variability on this time scale is of greatest interest. ENSO is certainly of interest but there are concerns about predictability. The Quasi-Biennial Oscillation (QBO) is Another mode of atmospheric variability that fits time scale of the reinsurance industry. The QBO changes on an almost annual basis and it appears to have a significant amount of predictability. In addition, the QBO seems to be correlated to varying degrees to a number of atmospheric phenomena, and most importantly for reinsurers, to hurricanes. The challenge to the scientific community is to find the physical connections between the QBO and the phenomenon of interest. Other modes of climate variability such as the North Atlantic Oscillation/Arctic Oscillation/Northern Annular Mode are of potential interest. However, their time scales differ from the typical reinsurance product and so other industries or new reinsurance products will be needed to take advantage of this information.