How the media was misled about the UN’s disaster report
The Centre for Research on the Epidemiology of Disasters (CRED) should withdraw its false claims and admit that the UN disaster report is fatally flawed.
You will all recall the latest UN report, which claimed there had been a massive rise in “reported disasters” in the last two decades, compared to the previous two.
The Dutch newspaper, De Telegraaf, published an article in response complaining that the UN were not comparing like with like, because many smaller disasters were simply never recorded in the past. They also published this reply from Joris van Loenhout, researcher at the Belgian Centre for Research on the Epidemiology of Disasters (CRED):
From about 1960-1970 onward, the completeness of the data is much greater, and the share of missing disasters much smaller. We are constantly working to improve completeness, and this is also happening for previous years and decades. For this reason, statements made in 2004 and 2006 are now somewhat outdated, as the completeness of the database has since improved,”
I have now had time to analyse CRED’s database, EM-DAT, and have the figures to show that Loenhout has not been telling the truth.
In their 2004 report, “Thirty Years of Natural Disasters”, CRED included this table of the number of natural disasters:
Note that between 1979 and 1998, there was a total of 3973 disasters.
EM-DAT’s database now provides a tool to filter data. I have downloaded all natural disasters for 1979 to 1998, but excluding biological and extra-terrestrial. The 2004 report confirms that the same criteria has been used for the table above:
This is the spreadsheet from the download:
Note that this gives a total of 3986 disasters, just 13 more than were declared back in 2004.
This makes a nonsense of van Loenhout’s claims. An extra 13 events, out of a total of over 4000 in twenty years, is nothing more than a tidying up, and can in no way be seen as “improving completeness”. Neither does it justify the assertion that CRED’s own statements in 2004, to the effect that data was far to incomplete then to make any meaningful comparison with now, “are now somewhat outdated, as the completeness of the database has since improved”.