Economic studies of earthquakes often rely on incomplete data, excluding lower-intensity events that can still cause disruptions. This column identifies the effects of earthquakes on aggregate economic outcomes using two datasets: one that captures only earthquakes surpassing certain thresholds and another that records every earthquake. The latter hazard-based dataset shows that moderate earthquakes can still significantly reduce growth and have lasting effects, especially in low- and lower-middle-income economies. These results highlight the importance of systematic data collection of natural disasters.

Most empirical studies of the economic effect of earthquakes rely on the Emergency Events Database (EM-DAT), a damage-based disaster database that captures only events surpassing specified thresholds of human or economic impact. As a result, EM-DAT excludes a vast number of lower-intensity earthquakes that may nonetheless disrupt economic activity, especially in vulnerable settings.

Early cross-country studies offered reasons for optimism on the long-term effects of earthquakes. Using the EM-DAT database, Skidmore and Toya (2002) found that natural disasters had little lasting effect on growth and might even spur renewal through reconstruction. Crespo Cuaresma et al. (2008) reached similar conclusions, suggesting that disasters could accelerate modernisation by replacing old capital. But this evidence may have been shaped more by data limitations than by genuine resilience.

The problem lies in how disasters are counted. EM-DAT, the standard global database, records events only when they cross specific humanitarian thresholds: at least ten deaths, one hundred people affected, or a declared state of emergency. These criteria make sense for emergency response but leave out countless smaller earthquakes that damage infrastructure, disrupt commerce, and strain public budgets.

By contrast, the US Geological Survey (USGS) records every instance of earthquake detected by seismographs, regardless of casualties or political response. It is a hazard-based rather than outcome-based dataset. The earthquake assessment changes dramatically depending on which dataset is used.

Figure 1 Earthquake frequency by dataset: EM-DAT versus USGS

Notes: This figure shows the density functions of earthquake intensity from two different sources: Emergency Events Database (EM-DAT) and US Geological Survey (USGS). USGS records vastly more moderate quakes, capturing the true distribution of seismic activity.

Using a systematic account of earthquakes

When researchers measure exposure using the full USGS data, results are widely different. GDP per capita typically falls in the year of an earthquake and remains below trend for several years. The cumulative income loss over a five-year horizon is large. Yet studies based on EM-DAT data show little or no impact on output. Economies appear to recover almost immediately. What looked like resilience turns out to be an artefact of selective measurement.

This pattern mirrors what happened in climate economics. Early studies that relied on ‘climate disaster’ databases found only mild effects on growth. But when researchers began using physical indicators, including temperature, precipitation, or cyclone exposure, the economic damage became evident, especially in poorer regions (Dell et al. 2012, Hsiang and Jina 2014, Burke et al. 2015). The same holds for earthquakes: once we observe the hazard directly, the damage becomes visible.

In a recent paper (Arezki et al. 2025), we show that earthquakes of similar magnitude inflict far greater and longer-lasting losses in low- and lower-middle-income economies than in advanced ones. The difference lies in state capacity, infrastructure quality, and fiscal space. Where building codes are weak, insurance coverage scarce, and budgets tight, even moderate earthquakes can derail growth for years. In richer economies, stronger enforcement, deeper financial markets, and effective emergency systems cushion the blow.

The implication is stark. Global disaster data are biased toward visibility, not vulnerability. The earthquakes most likely to be recorded occur in places capable of reporting them. The silent disasters, in poorer regions less able to document their effects, remain invisible in the record and in global risk assessments.

Figure 2 GDP per capita around major earthquake years (hazard-based measure)

Note: GDP declines sharply in the earthquake year and remains below pre-shock trends for several years.

Accounting for the full risk spectrum

Underestimating the frequency and cost of earthquakes has serious policy implications. Governments that rely on incomplete data may underinvest in mitigation, underprice contingent fiscal liabilities, and overestimate resilience in debt-sustainability analyses. Financial markets and insurers may also misjudge sovereign risk, especially where small but frequent shocks steadily erode fiscal space.

Incomplete data distorts not only policies but also perceptions of fairness. Countries that appear resilient on paper may simply be undercounted. Research by Noy (2012) on the economic consequences of natural catastrophes and by Todo and Inoue (2019) on the propagation of shocks through supply chains shows that overlooked linkages can magnify disaster costs far beyond what headline figures suggest.

A particularly important insight is that the cumulative effect of repeated moderate earthquakes can rival or exceed that of rare catastrophic ones. Many countries experience multiple tremors each decade, none large enough to trigger an international appeal, but together damaging infrastructure, discouraging investment, and lowering potential output. Yet fiscal frameworks and resilience plans typically focus on the ‘big one’, neglecting the slow erosion caused by smaller but more frequent shocks.

If policymakers use a more comprehensive dataset, they would see that the ‘middle’ of the risk distribution – the moderate events – deserves just as much attention. A modest increase in investment in resilient infrastructure, especially in low-income, earthquake-prone countries, could deliver large long-term payoffs.

Towards better policy through better measurement

Good disaster policy begins with good disaster data. Strengthening seismic monitoring networks, expanding reporting capacity, and integrating hazard data into economic analysis can materially improve the accuracy of forecasts and the design of resilience strategies. Governments and development institutions should work together to ensure that comprehensive data informs fiscal frameworks, debt-sustainability assessments, and macroeconomic surveillance.

Accurate data also help clarify where investments deliver the greatest payoff. Our findings underscore that resilience investments yield the highest returns in countries with low institutional capacity and high seismic exposure, precisely where earthquake impacts are greatest. A dollar spent on strengthening building codes or retrofitting critical infrastructure in these settings can avert significant future losses.

Earthquakes cannot be prevented, but their economic impact can be mitigated. Getting the full picture is the first step toward preparing for it.

Source : VOXeu