Tornado and lightning

Fact check: Craig Kelly says UN lied about the number of extreme weather events


On 17 October, Craig Kelly said that the United Nations had lied that extreme weather was getting worse, when it was actually getting better.


Craig has ignored half the most important data, and so has drawn conclusions that are mostly mistaken or misleading.

There are some problems with the UN report, but Craig’s post misses them or overstates them.

Note: I modified this post on 25 October to reflect a revised assessment of the UN report. The Verdict remains the same.


The claims


His post, which referenced some Twitter posts by Bjorn Lomborg, said: “claims by the United Nations about the number of extreme weather events increasing is a complete fabrication, a fraudulent lie.”

The post went on to say: “the climate change the world has been experiencing over the past two decades has been a change FOR THE GOOD, a declining number of extreme weather events, a declining number of deaths from extreme weather, and a decline in insurance losses as a percentage of GDP.”

In this post I will examine these claims and the accuracy of his conclusions.

The UN report and the data

The UN reported on its findings in ‘Staggering’ rise in climate emergencies in last 20 years, new disaster research shows. The full report can be downloaded here.

The report compares climate disasters (drought, flood, wildfires and major storms) during the 20 year period 1980-1999 with those during the 20 years 2000-2019 and provides additional analysis for the period 2000-2019.

I downloaded the UN’s data and analysed it myself (see note).

Ignoring data and missing the obvious

When addressing the main claim of the report (increasing extreme weather events), Mr Kelly’s post uses a graph from the report showing the latter two decades (2000 to 2019), and ignores the report’s comparisons over 40 years (1980 to 2019). When the 40 year period is used, the UN’s claim is correct.

The rest of Mr Kelly’s conclusions use data that cannot be used to assess whether climate change is happening and so his conclusions either change or are thrown into doubt when the facts are properly considered.

Declining number of weather events?

The UN accurately reports that about 75% more extreme weather events were recorded in the later period.

Mr Kelly used a graph from the report that shows a decline in the number of extreme weather events but ignores data in the report showing the increase.

I analysed the full 40 years of data, which shows there is a clear increase.

20 year data used by Craig Kelly
40 year data used by UN

Craig Kelly has “cherry-picked” the data from the report. The graph shouldn’t be used in this way, it is too short a period to define a trend.

This is a common mistake among those who deny climate science. Because climate and weather are so variable, it is always possible to find periods which go against the trend. Proper analysis uses all the data and examines long term trends, not shorter term anomalies.

A declining number of deaths?

This statement is accurate. Severe famines in north Africa in the period 1980-1999 caused hundreds of thousands of deaths, whereas there was no such large scale deaths in the period 2000-2019.

Thus the deaths resulting from extreme weather have reduced whichever time period is used.

However climate change isn’t the only factor in the number of deaths by natural disaster. The size of the population, its vulnerability and preparedness are important factors. Clearly, the people of North Africa were very vulnerable to drought and famine.

Decline in insurance losses?

Mr Kelly doesn’t provide a graph to support this claim. When I analyse the UN data, I find insurance losses are rising with time, regardless of which time period I choose. However insurance losses as a percentage of GDP are falling.

Other data in the report

The report gives two other measures of the effects of extreme weather – the number of people affected and the total value of the damage. Craig Kelly doesn’t mention either of these, but both show an increase over the 40 year period (although again, total damage as a percentage of GDP is falling).

The Bjorn Lomborg connection

Craig Kelly’s claims are based on Twitter posts by Bjorn Lomborg. In this case, Lomborg admits that the database shows that extreme weather have increased. But he suggests that the dataset is incomplete and points out some problems with the report

Craig Kelly’s claims go way beyond the more moderate statements by Bjorn Lomborg.

Does the data demonstrate climate change?

The main claim of the UN report that I have seen is that the number of events has increased. This is a fair assessment of the data and agrees with the predictions of climate science.

All the other measures (impacts on people and property) are not so useful in assessing the reality of climate change because other factors such as population density, value of infrastructure, improved warning and preparedness, etc, impact on the measures.

In assessing all these impacts, the UN report is sometimes confusing by including non-climate natural disaster data (earthquakes and epidemics) in some statistics. The database of natural disaster is also incomplete, e.g. damage costs are incomplete. But the report’s basic claim is accurate.


Craig Kelly has ignored half the data for the only measure in the report (the number of extreme weather events) that directly measures climate change This data provides strong evidence that the climate is becoming more damaging and dangerous.

The remaining four measures give a mixed picture because they reflect factors other than climate alone (increasing population and value of assets, improved warning systems, and the particular vulnerability of the people of North Africa to severe drought).

On the basis of this cherry-picked and wrongly reported information, Mr Kelly accuses the UN of telling “a fraudulent lie”.

But it turns out, again, that his main claim is false and the UN report’s main claim is reliable, though the report is a little confusing, overstated, and some details can only be approximate.

Notes: My analysis of the data doesn’t give the exact numbers the UN has used, presumably because the dataset has changed since the UN’s analysis. But the numbers broadly agree.

Photo by Ralph W. lambrecht from Pexels


  1. Nice work Eric. I looked into this once also and concluded that the definition of a “disaster” is not really a fixed concept. For instance, a disaster may be subject to political spin; a government lacking in preparedness may deny a disaster has happened and prefer not to declare a state of emergency. Imagine if Craig Kelly were given the responsibility of reporting to the UN, he would claim that last years fires were nothing more than average just to justify his own political purposes.

  2. Thanks Tom.

    Yes, this was the most difficult fact check I’ve done because a record of natural disasters is very subjective and dependent on what information comes in. Then I had to analyse the data myself in an enormous spreadsheet. That is why I said the UN report had some problems.

    But Craig Kelly’s response was as uninformed as ever, just taking the cherry-picked information from Bjorn Lomborg and ramping up the rhetoric way beyond what was true or justified.

Leave a Reply

Your email address will not be published. Required fields are marked *