A critique of our Keystone paper is probably not precisely correct, either

We appreciate the lively debate that our recent Nature Climate Change paper has engendered, and especially the serious engagement of thoughtful commentators, such as Michael Levi.

The main point of our paper is that far too few assessments of fossil fuel infrastructure projects have considered impacts on markets, energy consumption, and associated carbon emissions. While acknowledging the deep uncertainties in assessing those dynamics, our paper presents one attempt to fill this gap, finding that in the case of Keystone XL, global emissions could increase by 0 to 110 million tonnes CO₂e annually, depending on the extent to which the pipeline leads to additional oil sands production.

In his critique of our work, titled “A New Keystone XL Paper is Probably Wrong”, Dr. Levi generally agrees with our core thesis, noting that the U.S. State Department’s final Environmental Impact Statement (EIS) understates, if not altogether ignores, this dynamic relationships among oil production, prices and demand.

Dr. Levi makes three critiques of our work, primarily directed at our analytic methods. We address each of these, in turn below. While we may dispute some of Dr. Levi’s assertions and interpretations, we fully agree that given uncertainty with regard to future global oil market dynamics or development of alternative oil transport routes, one cannot speak with precision. Our paper offers a transparent approach to assessing the emissions impact of new energy supply infrastructure, and analysis of Keystone XL that we believe is based on reasonable assumptions, drawn from widely used and respected sources. We hope that this work will continue to spur greater conversation and research.

Topic #1: On modeling the world oil market

The global oil model we present in our paper, while admittedly simple, is very similar to ones used by scores of other energy modelers in the literature. Dr. Levi critiques the simplicity of our model, arguing that it fails to fully account for political and economic possibilities such as cartel behavior.

We agree that a simple model has limitations, but it also offers clear benefits. In contrast to more complex “black box” models, such as those used for the State Department EIS, a simple model can be shared and understood easily, with assumptions clearly identified, replicated and changed in order to examine the implications of alternative assumptions.

(In fact, another commentator on our paper, Andrew Leach, put forth a simple model that functions very similarly to ours. He assumed that rail can fully substitute for Keystone XL, which automatically led a conclusion, with little need for the model, that the impact of Keystone XL would be close to zero. A more recent blog by Leach looks into rail in more detail.)

We agree with Dr. Levi that politics are hard to predict, as are market power and cartel behaviors. A more sophisticated oil market analysis could have different (but not necessarily correct) findings. We chose to address uncertainties in the global market not by making our model more complex, but by modeling a range of possible supply responses, as described in our response to the next critique. We would welcome other suggestions on how to better do so.

Topic #2: On the dynamics of oil supply

We model the dynamics of global oil supply based on a supply curve from the energy consultancy Rystad Energy that many others (including the International Energy Agency) also rely on. Like similar supply curves from other analysts, this curve starts with conventional oil production in lower-cost regions (such as the Middle East), followed by a more steeply rising segment of higher-cost and less-conventional resources.

Our assumed equilibrium point on the supply curve – from which we derive our central case supply elasticity – corresponds to the level of oil consumption forecast by the U.S. Energy Information Administration (96.62 million barrels per day) for the year 2020, the same source used in the U.S. State Department EIS. In its most recent World Energy Outlook, the IEA forecasts a very similar level (95 mbpd in the New Policies Scenario).

Dr. Levi critiques our choice of the equilibrium point and the implied price-elasticity of supply in our central case, noting that “the point on the curve at which you estimate supply elasticity” can have a very large impact on the result.

While there’s no reason to expect that world oil consumption will be exactly 96.62 mbpd in 2020 (and, by implication, that the expected price is exactly $101 and the elasticity of supply is precisely 0.13), we see little reason to use a wildly different assumption for a central case.

We agree with Dr. Levi that estimates of oil supply costs and elasticities have significant uncertainty. Our paper provides a range of results based on a range of elasticities drawn from an OECD review. The paper’s supplemental information (posted online) provides results for all elasticities considered. While some cases, similar to the case shown below, yield lower emissions impacts, some also yield higher impacts (e.g. if future oil prices are higher and supply elasticities lower, or if demand responses are greater.)

Topic #3: On whether Keystone XL will lead to expanded production of the oil sands

While the State Department concluded that Keystone XL will have no effect on oil sands production levels, as stated in our paper, “Other analysts suggest that the State Department may be overly optimistic, however, and that regulatory, environmental and local community barriers faced by other pipeline and rail options could ultimately restrict expansion of oil sands production.” As a result, we consider the possibility that Keystone XL could affect the level of oil sands production, and as a result express our results as a range from virtually no emissions (if production is unaffected) to 110 million tonnes CO₂e (if production increases by the full Keystone XL capacity), and also as a ratio of added oil consumption per added barrel produced.

A number of commentators appear to reject this uncertainty and simply assert that Keystone XL will have no effect on oil sands production. We view such an outcome as unlikely, just as we do the notion that production will increase by exactly 830,000 bpd, Keystone XL’s full capacity. To illustrate this uncertainty, we cite the State Department’s own findings, that if future oil prices were in the $65 to $75 range, Keystone XL “could have a substantial impact on oil sands production levels, possibly in excess of the capacity of the proposed Project.”

Dr. Levi critiques our work for citing this finding but not considering the other implications of a low-price scenario, i.e. higher supply elasticity. If prices were in this lower range, then the supply elasticity, as suggested by the curve shown, would be significantly higher, about 0.85 instead of the 0.13 assumed in our central case. At that supply elasticity (and our central demand elasticity of -0.2), our model yields an emissions impact of up to 54 million tonnes CO₂e annually. As noted above, at these prices of $65 to $75 per barrel, the State Department’s analysis suggests that all (or more) of Keystone XL’s capacity could represent added oil sands production.

Closing thoughts

Our analysis uses a simple model with transparent assumptions, allowing others to make their own calculations using different input parameters. We view this as a strength of the approach, in part because it allows and encourages debate, in contrast to alternative approaches where assumptions are hidden.

Furthermore, our key point is broader: that market effects should not be neglected when it comes to investments that can increase energy supply, and that these effects can, in many cases, be far greater than typically accounted for. By our reckoning, the State Department’s analysis of Keystone XL appears to be one such case.