The Magic of Models

A study recently published in the journal Nature Climate Change has gained a lot of attention in the environmental community and mainstream media—National Geographic for example.  According to the study, the Keystone XL pipeline could produce four times more greenhouse gases than the State Department estimated in its final review in January.

Since the study was conducted by the Stockholm Environmental Institute, the results were probably predictable.   This may seem like a harsh judgment but reviews of the study have been very critical with the reason being that the authors make assumptions that are questionable and inconsistent with how the real work actually works.

First, they assume that increased production will have no affect on other producers and the inevitable result is lower prices that will stimulate more consumption.  But, markets are dynamic and while lower prices do stimulate demand, they also deter investment in new production. Implicit is the authors’ assumption that if Keystone XL is not approved, Canadian production will be constrained.  That is patently absurd.  If the Canadian oil does not come to the US via the pipeline it will go elsewhere by means that have higher emissions.

The authors go through an elaborate process to predict oil prices in the range of $65-$75.  The price of oil going forward is not that easy to predict. The amount of production at a given price, changes in demand, producing countries reactions and the marginal cost of producing oil all come into play.  Presently, the cost of new production is higher than the authors assume because of the use of complex technology, the engineering costs involved in projects, and exploration efforts in more challenging areas.

There are two ways to conduct analysis.  One way is to collect data, formulate a hypothesis, and then construct a model to test the hypothesis with independent data.  That is the scientific process.  The second way is to construct a model and assumptions that will produce the desired result. Recently, the Wall Street Journal ran an opinion piece—Confessions of a Computer Modeler, July 8—that explained how that is done. The criticism of the study published in Nature Climate Change suggests that the Institute knew the answer it wanted and worked backward to construct the supporting analysis.

Even if the analysis is correct, which it is not, you have to ask, so what?

Canada is not going to abandon its production of oil sands unless the world price of oil collapses and it is believed that it will stay below Canada’s cost of production.  Anyone who believes that assumption simply does not understand oil markets and enlightened self-interest.  If the oil is not shipped by pipeline, it will be shipped by rail or by a Canadian pipeline west for export to Asia by tanker.  All of those alternatives produce higher emissions, as would consumption in Asian countries.

The other problem with the study, its fatal flaw, is the blind acceptance of the climate orthodoxy—more emissions of CO2 translate into higher global temperatures.  The past 17 year pause in warming shatters that assumption as well as the credibility of models that use the linear relationship as their foundation.  Charts showing model projections and actual global temperatures make clear that models grossly over predict temperature and are poor tools for policy making.

Credible analyses require a credible and objective process.  The study published in Nature Climate Change is not credible.