Tracking Progress in the Governance of Critical Risks

Annex A. Technical notes 

Number of disasters and technological accidents per year (1985 – 2021) 

The number of disasters and technological accidents per year from 1985 – 2021 was calculated using data from EM-DAT for the Adherent countries.  

Average annual deaths per OECD country per 1 million inhabitants (2000-2020) 

The average annual deaths per million inhabitants was calculated using the sum of the total number of deaths caused by natural and technological hazards (n) for each year from 2000 to 2020. Then, the sum of the country’s population (POP) for each year from 2000 to 2020 was calculated. Finally, the result from the sum of deaths per million was divided by the sum of POP and converted to deaths per million inhabitants multiplying it by 1 million to get the average rate for the country over the specified period.  

$Average Annual Deaths per Million Inhabitants=\frac{{\mathrm{\Sigma }}_{j=2000 }^{2020}n}{{\mathrm{\Sigma }}_{j=2000}^{2020}\mathrm{P}\mathrm{O}\mathrm{P}}*\mathrm{1,000,000}$

Where:  

POP=Population estimates according to OECD population database 

We assume that "missing" values for the variable "total deaths" corresponds to 0 deaths. This implies that we will slightly underestimate the average value.  

The figure excludes deaths caused by extreme heatwaves and COVID-19, to avoid mixing different methodological approaches for counting deaths.   

Average annual damage as % of GDP (2000-2020)  

The data covers 33 out of 41 Adherent countries.  To calculate the amount of average annual damage as % of annual GDP, fist the sum of total damage (d) caused by all disasters for each year from 2000 to 2020 is determined. Then, the sum of the country’s GDP for each year from 2000 to 2020 is calculated. Finally, the total in damages is divided by the sum of GDP and converted to a percentage multiplying it by 100 to get the average rate for the county over the specified period.  

$Average Annual Damage \left(\% of GDP\right)=\frac{{\mathrm{\Sigma }}_{j=2000}^{2020} d}{{\mathrm{\Sigma }}_{j=2000}^{2020} GDP}*100$

We assume that "missing" values for the variable "total damage" correspond to 0 USD in damages for an event. Consequently, this methodology underestimates the average losses as a function of the percentage of missing values. The higher the percentage of missing values, the more the Adherent’s average annual damage is underestimated.  Figure 1.3 uses data on economic damages from EM-DAT, The Emergency Events Database, (www.emdat.be), and on GDP from OECD, “Gross domestic product (GDP)” (indicator), (http://dx.doi.org/10.1787/dc2f7aec-en).