Figure 8-1: The costs of catastrophic weather events have exhibited a rapid upward trend in recent decades. Yearly economic losses from large events increased 10.3-fold from US$4 billion in the 1950s to US$40 billion per year in the 1990s (all in 1999 US$). The insured portion of these losses rose from a negligible level to US$9.2 billion annually during the same period, and the ratio of premiums to catastrophe losses fell by two-thirds. Notably, costs are larger by a factor of 2 when losses from ordinary, noncatastrophic weather-related events are included (e.g., as shown in Figure 8-6). The numbers generally include "captive" self-insurers but not the less-formal types of self-insurance (Munich Re, 2000).
Present-day impacts of weather events on financial services are caused mainly by extreme events. Differences in vulnerability exist, caused by geographical location, population distribution, and national wealth. In developing countries, there may be very high mortality from extreme weather but relatively small costs to the financial sector because of low insurance penetration. In developed nations, the loss of life may be much less but may have enormouseven catastrophiccosts to the insurance industry (see Section 8.3.1). Swiss Re (2000b) has compiled lists of the 40 worst catastrophes between 1970 and 1999 in terms of insurance losses and fatalities. These lists show that:
In contrast, of the 40 worst events in terms of fatalities, only 16 were weather related, of which 13 occurred in Asia. A list of natural disasters causing billion-dollar losses drawn up by Munich Re (2000; see Table 8-3) shows that, of 30 such disasters, 15 affected the United States and seven affected Europe. Eighteen were related to windstorm. With the exception of earthquakes, all were weather related.
In recent decades, economic and insured losses related to weather extremes have increased rapidly (see Figure 8-1). An important part of this trend is related to socioeconomic factors; another part may be explained by climatic factors. Where trends in climate variables do occur, there are two possible principle causes:
Whatever the cause, it is important to note that a relatively small change in the mean of a climate variable can lead to a large change in the occurrence of extremes. Meehl et al. (2000a) explore the implications for extremes of changes in the mean and/or variance; they show clearly that the relationship between a change in the mean and a change in the occurrence of extremes is nonlinear, as illustrated in Figure 8-2.
Other reports in this collection