Key-indicator analysis, the Chinese virus and the climate scam

By Christopher Monckton of Brenchley

Models! Dontcha just wish your taxes didn’t have to pay for them?

First there was the Imperial College model that predicted 500,000 deaths in the UK and 40 million worldwide from the Chinese virus in the absence of control measures, by the end of this year. Control measures were introduced in the worst-affected countries, so we shall never know how credible that prediction was.

Then, in the other direction, there was the model from the Institute for Health Metrics and Evaluation, which had originally predicted 200,000 deaths in the USA, of which 55,000 have occurred at the time of writing.

On April 4, my good friend Willis Eschenbach, who has an enviable facility with and interest in data, published some predictions from the IHME model for how many people would have died of the infection in the four months to August 4 this year.

Willis pointed out that “The IHME model is … not worth too much trust – it’s been wrong too many times … The model historically has predicted numbers that were too high.”

Just over three weeks have passed since Willis got the model to make its predictions. He wrote: “The latest incarnation of the model is predicting 81,766 COVID-19 deaths in the US by August 4, 2020. That’s down from 93,000 in the previous incarnation of the model.”

By April 4, there had been 10,384 deaths in the United States. To yesterday, just 23 days later later, there had been 56,797 such deaths, a five-and-a-half-fold increase, giving a mean daily compound growth rate of 7.7% in deaths.

If that growth rate were to persist for just five days, there would be more than 82,000 dead in the U.S. alone, a whole three months before the model’s due date of August 4.  Fortunately, the daily growth rate in deaths in the U.S. averaged over the past week is down to 4.1% and will be likely to fall further. But even if it falls fast enough to average little more than 0.35% over the 103 days from today to August 4, there will indeed be 81,766 U.S. deaths by then.

At present, though, the daily growth rate in active cases is not falling much, which means that in two or three weeks the daily growth rate in deaths will not be falling much either.

Willis also looked at Italy. On April 4, there had been 15,362 deaths in Italy. The model predicted that this would rise to 20,300 deaths by August 4. In fact, that total was surpassed on April 13, just nine days after Willis wrote, with 20,465 deaths. By yesterday there had been 26,977 deaths in Italy. Lesson: there is no single reproduction rate. It varies from time to time and from place to place. Models do not capture such differences easily.

In Spain, there had been 11,947 deaths by April 4. The model predicted 19,200 deaths by August 4. That total was surpassed less than two weeks later, on April 17, with 19,478 deaths. By yesterday there had been 23,521 deaths in Spain. Same lesson.

Willis also looked at the model’s predictions for California, the world’s fifth-largest economy. There had been 289 deaths in the state by April 4. The model predicted 1783 deaths by August 4. Remarkably, that total was reached yesterday, April 27.

Lesson: in the early stages of a pandemic models are not a lot of use because there are insufficient data to inform them. The compound daily growth rates in cumulative cases and in cumulative deaths give a better indication than the models do. These are the key indicators.

As a policy advisor at 10 Downing Street, often asked to provide analysis of technical questions on which the “experts” were either divided or flat-out wrong, I would look for the key indicators – never more than two or three – and make my recommendations based on them. One has to be ruthlessly dispassionate, and the use of key indicators is a great help with that, because it is much easier to see when someone is tampering with those than when a modeller is tweaking a few parameters in his model to achieve whatever result is most profitable to him.

Why is key-indicator analysis so important? The reason is simple. If governments had been aware that in the early stages of a pandemic the growth rate in cumulative cases and thus in deaths is near-strictly exponential, they would have realized a great deal sooner than they did that early control measures work a great deal better than late ones, saving lives, buying time and very greatly reducing the eventual economic cost.

We are now passing into an opposite problem. Now that the daily growth rates in active cases and in deaths are declining, and now that it is known that only over-60s with comorbidities are at appreciable risk, on the basis of those key indicators lockdowns can be cautiously dismantled, beginning at once.

But governments are still not learning from the key indicators, so some – such as the UK, which has been spectacularly behind the curve at every stage – still refuse to countenance relaxation of the lockdowns, or even to announce what their plan is.

Mr Trump has sketched out a plan: Mr Johnson has not. The British people, rightly, are feeling left out of the loop and are becoming impatient. Travel increased 5% last week compared with the previous week. It will increase again the next time we see the sun.

Certainly, as the Hokkaido example demonstrates, ending lockdowns prematurely or precipitately can lead to a renewed spike in infections, requiring a second and fiercer lockdown. But proper attention to the key indicators day by day, and far less mucking about with models, will lead governments to better and timelier decision-making.

One can apply the same key-indicators analysis to the climate question.

First, define the question: How much warming will a doubling of CO₂ concentration eventually cause?

Next, find the key indicators. They are no more difficult to find than they are in a pandemic.

1. How much warming has actually been measured to occur up to a given date? From 1850-2011, the year to which data were updated for IPCC’s latest Assessment Report, just 0.75 degrees’ warming had occurred (HadCRUT4).
2. How many Watts per square meter of manmade radiative forcing drove the warming up to that date? Up to 2011 there had been about 2.5 Watts per square meter of anthropogenic forcing (IPCC 2013, Fig. SPM.5), from which the radiative imbalance of 0.6 Watts per square meter (Smith 2015) must be deducted, making 1.9.
3. What is the best estimate of the forcing in response to doubled CO₂? We can’t measure that, so we’re forced to rely on models. But it’s about 3.45 Watts per square meter (Andrews et al. 2011).

Just three key indicators. The warming to be expected from doubled CO₂ is simply the product of the 0.75 degrees’ warming from 1850 to 2011 and the ratio of the 3.45 Watts per square meter CO₂ forcing to the realized anthropogenic forcing of 1.9 Watts per square meter. And that’s about 1.4 degrees. The method is described in Lewis & Curry (2014).

One could argue that there has been quite a bit of warming since 2011, but one must also allow for more forcing since then as well. One could push up the equilibrium sensitivity to CO2 and make it around 1.5-1.6 K (Lewis & Curry 2018).

But the point is that with this simple analysis based on key indicators we are very likely to be somewhere inside the ballpark. But just look at the various profitable predictions of global warming made by the climate models:

The two scales –upper for doubled CO₂, lower one for warming from 1850-2011 – are aligned to each other so that they both start at zero and so that 3.45 Watts per square meter of CO₂ forcing is directly above the 2.5 Watts per square meter of radiative forcing to 2011.

Here’s how it works. We know how much warming would have been caused by 2011 if all of the 2.5 Watts per square meter of manmade forcing to that date had come through. It is the product of the 0.75 K warming to that date (the blue arrow) and the ratio of that 2.5 Watts per square meter total forcing to the realized forcing of 1.9 Watts per square meter: i.e., about 1 degree. Following the green arrow shows that 1 degree of warming to 2011 is equivalent to 1.4 degrees of warming in response to CO₂ doubling.

But just look at the predictions made by the wretched models. In 1990 IPCC, ignoring the importance of the leading indicators, predicted 3 degrees’ equilibrium warming; the CMIP5 models predicted 3.35 degrees; and the CMIP6 models predict 4.1 degrees, with an interval of 3 to a remarkable 5.2 degrees. All of these predictions are manifestly excessive. They are two to four times too big.

One can also calculate how much warming would have been observed by 2011 if each of these three wild predictions had been correct, simply by following the dotted arrows. Only 0.75 degrees of warming had been observed by 2011, but if the models’ predictions of equilibrium warming in response doubled CO₂ were correct the observed warming by now would have been somewhere between 1.7 and 2.25 degrees.

And it wasn’t. So we know the models are running hot.

We reached that conclusion simply by analysing the key indicators. Of course there are uncertainties in the climate data, just as there are with the pandemic. But on the basis of this simple calculation there is just not going to be anything like enough global warming caused by us over the 150 years or so that it will take to feel the eventual warming from doubled CO₂ at the present rate of increase in concentration to make it worthwhile to do anything at all to make global warming go away. There is a pandemic emergency, but there is no climate emergency.

Instead, let the trees and plants thrive on the extra CO₂. What a pretty paradox it is that those who call themselves “green” are so viscerally opposed to our returning to the atmosphere some insignificant and harmless fraction of the CO₂ that once resided there, for it is visibly greening the Earth.

Or is it that the Greens – the traffic-light tendency – are simply too yellow to admit they’re really Reds?

Fig. 1. Mean compound daily growth rates in estimated active cases of COVID-19 for the world excluding China (red) and for several individual nations averaged over the successive seven-day periods ending on all dates from April 1 to April 27, 2020.

Fig. 2. Mean compound daily growth rates in cumulative COVID-19 deaths for the world excluding China (red) and for several individual nations averaged over the successive seven-day periods ending on all dates from April 8 to April 26, 2020.