FIGURE 3.2 Trends in per-capita property and crop losses (adjusted to $2010) from natural hazards, 1960-2010. According to Gall et al. (2012), per-capita losses appear to be escalating in the United States, even when normalized by population, and have more than tripled per person since the 1960s. Source: S. Cutter; compiled from SHELDUS.
Another issue in analyzing loss data is that not all losses are counted and valued (Box 3.3). In the case of Munich RE, the NatCatSERVICE database provides property losses (total and insured) and insured business interruption losses, estimated from known insured losses. Because of the differences in loss estimation techniques, thresholds for inclusion in the database (large versus small events; insured versus uninsured losses), and data availability (public versus proprietary), natural-hazard loss databases are rarely comparable with one another. For example, comparisons among four publicly accessible databases show different total dollar loss estimates for the United States in 2010 attributed to weather perils such as winter storms, hurricanes, tornadoes, and flooding (Table 3.1). In the health arena, some losses of life and health may occur days or months after the disaster and thus may go uncounted.
Geographic Variation in Economic Losses
Long-term disaster loss data for specific geographic regions provide a baseline from which to measure improvements in resilience. The success of measures to reduce disaster risk and impacts are difficult to evaluate without this baseline information. A number of federal agencies compile separate data on disaster losses and costs including the National Oceanic and Atmospheric Administration (NOAA), FEMA, the U.S. Geological Survey (USGS), and the Department of Agriculture. These data serve quite specific and useful purposes, but in aggregate are incomplete, often incompatible with one another, have limited economic impact information, and are less useful for mapping the