The Data Aftershock: How Earthquake Records Distort Global Growth Estimates

The IMF paper “The Macroeconomic Impact of Earthquakes on Growth: A Tale from Two Datasets” finds that earthquakes significantly reduce GDP growth—especially in poorer countries, when measured with objective USGS seismic data, while traditional EM-DAT disaster records underestimate these impacts due to selective reporting. It concludes that credible, hazard-based data are essential for understanding true economic vulnerability and guiding better disaster and fiscal policy.

A new IMF Working Paper, “The Macroeconomic Impact of Earthquakes on Growth: A Tale from Two Datasets”, by Rabah Arezki, Youssouf Camara, Patrick Imam, and Kangni Kpodar, researchers from the International Monetary Fund (IMF), CNRS, McGill University, and FERDI, delivers a sharp rethink of how disasters shape economies. Supported by the UK’s Foreign, Commonwealth and Development Office, the study reveals a deceptively simple truth: the way we measure earthquakes determines how we understand their economic cost. Comparing the widely used Emergency Events Database (EM-DAT) and the United States Geological Survey (USGS) seismic records, the authors show that data choice can dramatically alter the story economists tell about recovery, resilience, and loss.

The Hidden Bias in Disaster Data

When an earthquake strikes, the devastation is immediate, but its economic aftershocks unfold slowly. For decades, economists have debated whether such events spur creative destruction or deliver lasting harm. Arezki and his co-authors argue that the divide stems largely from flawed data. The EM-DAT database, the standard source for many studies, only records disasters that exceed certain thresholds: ten deaths, a hundred people affected, or an official state of emergency. This design means countless smaller, yet economically disruptive, earthquakes go uncounted. The USGS dataset, on the other hand, systematically logs every measurable quake, regardless of casualties or damage. The difference, the paper contends, is more than technical; it’s epistemic. By overlooking frequent, moderate tremors, EM-DAT gives a skewed impression that economies are more resilient than they truly are.

Rethinking the Literature on Growth and Destruction

The study revisits decades of disaster economics. Early research, such as Skidmore and Toya (2002), suggested that disasters might fuel modernization by forcing capital renewal. But later work by Raddatz (2009) and Cavallo et al. (2013) found that catastrophic shocks often depress growth for years, especially in poor countries with limited fiscal capacity. Still, most of these analyses relied on EM-DAT, meaning they focused on extreme cases while ignoring the quieter but more pervasive shocks that strain developing economies over time. Recent critiques by Panwar and Sen (2020) and Joshi et al. (2024) exposed how EM-DAT systematically underreports events in wealthier or less transparent nations. The IMF paper builds on these insights, arguing that to grasp the true macroeconomic cost of earthquakes, researchers must turn to hazard-based datasets like USGS, which record physical phenomena rather than reported damages.

What the Data Really Say

To test this, the authors combine EM-DAT and USGS records for 2012–2022 and link them with macroeconomic indicators from the World Development Indicators. Their model examines 178 countries using fixed-effects panel regressions that control for geography and time. When earthquake exposure is measured using USGS data, the results are striking: a one-unit increase in earthquake magnitude reduces GDP per capita growth by about 0.5 percentage points, and this effect persists for up to three years, cumulating to a 1.3 percentage-point loss. By contrast, results based on EM-DAT show no statistically significant relationship between earthquakes and growth, a finding the authors attribute to data truncation rather than true resilience.

Further analysis reveals deep inequality in outcomes. In low-income countries, the negative growth effects are larger and longer-lasting, reflecting weaker infrastructure, limited fiscal buffers, and slower recovery. In high-income nations, where reconstruction capacity is stronger, the economic effects are milder and short-lived. The study also finds that even moderate earthquakes can have meaningful economic impacts, effects that EM-DAT, by design, fails to capture.

Counting Better to Build Better

The paper’s conclusion extends beyond earthquakes to the broader field of disaster economics. It warns that reliance on selective, outcome-based databases can lead policymakers to underestimate both exposure and vulnerability. If official data downplays the true frequency of seismic events, then governments may underprice disaster risk, underinvest in resilience, and misjudge fiscal vulnerability. For rapidly urbanizing regions in Asia, Africa, and Latin America, this data blind spot could translate into costly missteps. The authors urge policymakers to integrate objective, geophysical data like USGS records into national planning, sovereign risk modeling, and early warning systems.

Ultimately, the study reframes a methodological issue as a developmental one. By showing that earthquakes inflict persistent, measurable damage, especially in poorer nations, it argues that economic resilience cannot be built on incomplete evidence. As the authors conclude, while earthquakes are acts of nature, the stories economists tell about them are human constructions shaped by the choices made in interpreting data. Better measurement can uncover hidden vulnerabilities and lead to more effective policies. Extending this rigor to other disasters, floods, droughts, and storms could reshape how the world measures risk and plans for recovery.