Measuring the Tremor: How Data Bias Distorts the True Economic Toll of Quakes

The IMF paper reveals that the true economic cost of earthquakes has been underestimated because widely used disaster databases like EM-DAT omit smaller events, while comprehensive USGS data show significant, lasting GDP losses. It argues that credible, hazard-based data are essential for accurate disaster economics and effective resilience policies.

The International Monetary Fund’s working paper, prepared by researchers from the IMF’s Strategy, Policy and Review Department and the Institute for Capacity Development, together with experts from CNRS, FERDI, McGill University, and Harvard Kennedy School, redefines how economists should think about natural disasters. Authored by Rabah Arezki, Youssouf Camara, Patrick Imam, and Kangni Kpodar, it contends that much of what we believe about the economic impact of earthquakes depends not on the quakes themselves, but on the way data are gathered and reported. By comparing the Emergency Events Database (EM-DAT), a widely used, impact-based record of disasters, with the United States Geological Survey (USGS) dataset, which systematically measures all seismic activity, the authors reveal that global analyses have long underestimated the true cost of earthquakes.

When Data Shapes Economic Truth

The paper begins with an unsettling question: can the data we use distort reality? EM-DAT includes only events that surpass certain damage thresholds, such as ten or more deaths, 100 people affected, or an official emergency declaration. This means that countless small or moderate earthquakes, which may still destabilize economies, go unrecorded. Because of this, EM-DAT overrepresents catastrophic events, especially in poorer nations where weaker infrastructure makes even minor tremors more destructive. The USGS dataset, by contrast, records all seismic events detected by instruments worldwide, offering a complete and unbiased account. The difference between these two approaches, the paper argues, is not trivial, it determines whether earthquakes appear economically harmless or devastating.

From Creative Destruction to Persistent Losses

For decades, economists debated whether disasters spur recovery or destroy prosperity. Early research, including Skidmore and Toya (2002), used EM-DAT data to claim that natural disasters could encourage “creative destruction,” promoting modernization and productivity growth. But later studies, such as Raddatz (2009) and Cavallo et al. (2013), contradicted this optimism, showing that disasters often inflict long-term economic harm, particularly in countries with limited fiscal and institutional capacity. Building on this literature, the IMF paper demonstrates that these conflicting conclusions stem from data quality rather than economic reality. When relying on EM-DAT, smaller and unrecorded events vanish from analysis, giving a misleading impression of resilience. When using USGS data, earthquakes consistently show negative and persistent effects on GDP.

What the Numbers Reveal

The authors compile data from 178 countries between 2012 and 2022, matching earthquake records from EM-DAT and USGS with macroeconomic indicators from the World Development Indicators. They measure exposure using the maximum earthquake magnitude and total number of events per country-year. The contrasts are vivid. Figures in the paper show that the USGS captures a continuous distribution of magnitudes, while EM-DAT clusters only at high-intensity levels. EM-DAT’s entries are also skewed toward low-income countries, not because they face more quakes, but because moderate ones in fragile economies are more likely to cross reporting thresholds.

Through a fixed-effects panel regression model, the study estimates the relationship between earthquake intensity and GDP per capita growth. The results are telling. Using USGS data, each one-point rise in magnitude reduces GDP per capita growth by about 0.5 percentage points, with effects lasting up to three years and a cumulative five-year loss of roughly 1.3 percentage points. EM-DAT-based results, however, show no significant impact, reflecting the distortion caused by selective reporting. The study also reveals that low-income countries suffer the most pronounced effects, due to weak infrastructure, limited fiscal buffers, and fragile institutions, while high-income countries display greater resilience. Even low-intensity earthquakes have measurable economic costs, a finding only visible with comprehensive data.

Why Better Data Means Better Policy

The implications are far-reaching. By depending on incomplete, damage-based databases, policymakers may be underestimating both exposure and risk. This leads to the underpricing of disaster insurance, inadequate investment in resilience, and poor fiscal planning. It also means that aid and attention tend to favor large, headline-grabbing catastrophes over smaller but cumulatively damaging events. The authors argue that integrating geophysical, hazard-based data such as USGS into policy design would enable governments and financial institutions to assess disaster risk more accurately, strengthen early warning systems, and better allocate resources. The study’s message resonates strongly with ongoing debates about climate resilience and sustainable development finance: accurate measurement is a prerequisite for credible policy.

Rethinking the Fault Lines of Knowledge

The report is more than a technical exercise; it is a call for intellectual humility in the face of data bias. The authors show that the choice of dataset can alter not only econometric results but also the global narrative about resilience and vulnerability. The paper’s central lesson extends beyond earthquakes: in a world increasingly shaped by climate shocks, relying on incomplete data means misjudging the scale of risk. Earthquakes, they remind us, are acts of nature, but the stories we tell about them are human constructions. As the authors conclude, better data lead to better policy, and perhaps to a more resilient world that can withstand both geological and economic tremors.