EM-DAT Disaster Database Analysis

An event is entered if it meets at least one of four criteria: ten or more reported deaths, one hundred or more people affected, the declaration of a state of emergency, or a call for international assistance. These thresholds mean EM-DAT records significant disasters rather than every hazard occurrence. The practical implication is that the database is excellent for comparing notable disasters but cannot speak to the many small events that fall below the criteria, and analyses must be framed accordingly.

Because the way disasters are recorded has changed over time. Reporting infrastructure, media coverage, and data-sharing have improved markedly over the twentieth and twenty-first centuries, so more events are captured now than before. An apparent increase in recorded disasters can therefore partly reflect better reporting rather than more hazards. Rigorous analysis addresses this by focusing on recent, well-reported periods, on high-impact events that were reliably captured throughout, or by modeling reporting explicitly, and by normalizing losses for inflation, population, and exposure.

With care, because the economic-damage field is often incomplete, particularly for lower-income countries. Missing values should never be treated as zero, since that biases totals and trends downward. Recorded damage should be deflated to constant prices and, for trend analysis, normalized for the growth of population and exposed assets so that increases in losses are not mistaken for increases in hazard. Where loss completeness is essential, EM-DAT is best combined with dedicated national loss databases and reinsurance datasets.