For many months I struggled with what to think about the Keystone pipeline. On the one hand, my friends in the Administration and elsewhere argued persuasively (at least on first blush) that the oil from the tar sands would be sold one way or another, so whether the pipeline was completed or not didn’t make any difference from a climate perspective. On the other hand, I was instinctively skeptical of building infrastructure to high carbon resources that we simply don’t have the luxury to burn if we hope to stabilize the climate.
What I came to realize was that the first point of view was based on a flawed framing of the problem, which led me to write the op-ed titled “The Fatal Flaw in the Case for Keystone”. The purpose of this short note is to explain the underlying intellectual roots of that framing (and the circular reasoning it engendered), so students of these matters can dig deeper and avoid making such mistakes in the future.
Neoclassical economics has taught us a lot about how economies work, but it is based on a set of assumptions that often don’t reflect economic decisionmaking in the real world. For example, most economic models assume perfect & costless information, perfect competition, no externalities, no transaction costs, and constant or decreasing returns to scale. In this world, it is sufficient to show that there’s a price difference between (for example) Alberta Tar Sands oil and similar heavy oil from other places, and to state that market forces won’t let that price difference persist. That’s, in essence, what it means to state that “the Alberta tar sands oil will be sold anyway”.
In the real world, however, information is imperfect and costly, transaction costs can be large, and increasing returns to scale are pervasive. These (and other) factors lead to what’s called “path dependence”, meaning that our choices now affect our options later. For example, if we invest in deploying mass produced technologies (like solar panels and wind turbines) we move down the learning curve, thus reducing the costs of those technologies five or ten years hence. If we deploy fewer of those devices, we don’t move as far down the learning curve and their costs in 2020 will be higher than they would be in the case where we more actively promote deployment of these technologies.
Another source of path dependence is the nature of the climate problem itself. Because the most important greenhouse gases stay in the atmosphere for a long time, it’s the cumulative emissions of greenhouse gases that matter. That means that we can emit only a fixed amount of carbon (our “carbon budget”) if we want to stay under the 2 Celsius degree warming limit that the US and other major nations accepted at Copenhagen in 2009. If we burn more high carbon fuels now, we commit ourselves to even faster reductions in emissions later (because the total carbon budget over the next century is fixed).
In the case of Keystone, path dependence matters a lot. Right now the heavy oils from the tar sands are “landlocked”, because pipeline capacity is limited. That means that the price of heavy oil from Alberta is much lower than comparable heavy oils (like those from Mexico, called Maya heavy oil). This discount can be tens of dollars per barrel, and it reflects the limited transportation options for tar sands producers to move their product to market. In the last nine months it has ranged between $20 and $40 per barrel. Building more pipelines will allow this differential to narrow and eventually close, but the rate at which it closes depends on how fast pipeline capacity is built (it is path dependent).
The idea that “tar sands oil will be sold anyway” assumes that adequate pipeline capacity will be built to allow this outcome to come to pass, so it’s circular to argue (as the State Department’s Environmental Impact Statement does) that approving Keystone XL will have no effect on the exploitation of the tar sands. Any one project will have a minimal effect, of course, but the cumulative effect of building enough pipelines for tar sands oil to make its way to market will be to allow greater exploitation of tar sands than would otherwise be possible.
The claim that tar sands oil will make its way to market one way or another is therefore dependent on the construction of additional pipeline capacity. If pipelines aren’t built, then the price differentials won’t narrow and less tar sands oil will be produced than otherwise (because the profitability of exploiting this resource would be substantially reduced). The logic in the State Department’s Environmental Impact statement about whether approving Keystone would increase exploitation of the tar sands is therefore invalid (because it’s circular).
The key issue from a climate perspective comes down to whether building more pipelines would affect the cumulative emissions from the tar sands. The State Department’s environmental impact statement comes to one conclusion, based on the circular reasoning I identify above, but the Canadian Oil Industry comes to the opposite conclusion (as I point out in the op-ed). If the construction of additional pipelines would affect the quantity of heavy oil extracted from the tar sands, then approving the Keystone XL pipeline (and any additional pipelines to the tar sands) is counter to the interests of climate protection, and the pipeline should therefore be rejected on that basis.
Addendum: Given what the President said in his climate speech today, the argument I make above should sink the Keystone XL pipeline.
Addendum #2: See these illuminating musings by Dave Roberts about President Obama’s statement about Keystone and this excellent piece by Jesse Jenkins summarizing the different possible scenarios related to Keystone. The key quote from Jenkins: "So can rail lines really scale up to ultimately handle a couple million barrels of new tar sands oil shipments per year? In many ways, the Keystone debate hinges on this question.” This quote echoes what Robert and I discuss in the comments below, which is why I’m going to delve more into the question of whether rail is truly a substitute for pipelines.