AGW Doesn’t Cointegrate: Beenstock’s Challenging Analysis Published

The Beenstock, Reingewertz, and Paldor paper on lack of cointegration of global temperature with CO2 has been accepted! This is a technical paper that we have been following since 2009 when an unpublished manuscript appeared, rebutting the statistical link between global temperature increase and anthropogenic factors like CO2, and so represents another nail in the coffin of CAGW. The editor praised the work as “challenging” and “needed in our field of work.”

Does the increase in CO2 concentration and global temperature over the past century, constitute a “warrant” for the anthropogenic global warming (AGW) theory? Such a relationship is a necessary for global warming, but not sufficient, as as range of other effects may make warming due to AGW trivial or less than catastrophic.

While climate models, or GCMs shows enhancement of the greenhouse effect can cause a temperature increase, the observed upward drift in global temperature could have other causes, such as high sensitivity to persistent warming from enhanced solar insolation (accumulation theory). There is also the urban heat island effect and natural cycles in operation.

In short, the CO2/temperature relationship may be spurious, have an independent cause, or temperature may cause CO2 increase, all of which falsify CAGW here and now.

Cointegration attempts to fit the random changes in drift of two or more series together to provide positive evidence of association where those variables are close to a random walk.
The form of time series process appropriate to this model is referred to as I(n), having the property that n is the number of differencing operations needed before the series has a finite mean (stationary, or does not drift far from the mean). A range of statistical tests, the Dickey-Fuller and Phillips-Perron procedures, identify the I(1) property.

Beenstock find that while temperature and solar irradiance series are I(1), anthropogenic greenhouse gas (GHG) series are I(2), requiring differencing twice to yield a stationary series.

This fact blocks any evidence for AGW from an analysis of the time series. The variable may still somehow be causally connected, but not in an obvious way. Previous studies using simple linear regression to make attribution claims must be discounted.

The authors also show evidence of a cointegrating relationship between the temperature (corrected for solar irradiance) and changes in the anthropogenic variables. This highlights what I have been saying in the accumulation that the dynamics relationships between these variables must be give due attention, lest spurious results are obtained.

While this paper does not debunk AGW, it does debunk naïve linear regression methods, and demonstrate the power of applying rigorous statistical methodologies to climate science.