Why the debate over Keystone and emissions comes down to rail

Econ 101 shows that blocking Keystone only matters to oil sands production if rail can’t pick up the slack

Stephen C. Host/CP

Stephen C. Host/CP

Yesterday, I wrote a long post on the basic economics of oil sands and Keystone XL in response to an article published this week in Nature Climate Change (article via Nature paywall). The article claimed that the U.S. State Department underestimated the emissions impact of the Keystone XL pipeline by up to a factor of four and my argument was that the authors failed to make the case that the change in global oil supply with, versus without, Keystone XL was actually affecting the marginal barrel into world oil markets; if you can’t show that, then you’re not going to have a global price effect from making oil sands more expensive. Further, even if you can show that, you’d have to show that the removal of Keystone XL as a transportation option does, in fact, make the marginal barrel of oil sands more expensive. If you can do both of those things, you can show that there’s a global price impact from Keystone XL and make the argument that those emissions should count in assessing the greenhouse-gas impact of the pipeline project.

As is usually the case, the discussion on Twitter after I published the piece led me to think about things slightly differently, and realize where I might have expanded some of the discussion, in particular, with respect to the second point above: What would removing Keystone XL mean to oil sands markets? To assess these issues, the consulting firms engaged to evaluate pipeline projects use large, computational models of the oil market, which solve for prices and trade flows with different assumptions with respect to infrastructure. For example, these models would predict that, if you have to move oil out of Alberta by rail to the Gulf Coast instead of by pipe, the price in Alberta will be lower relative to the price on the Gulf Coast than it would otherwise be, to account for the higher transportation costs. When you explain this using Economics 101, the various arguments made with respect to Keystone XL (and other pipelines) become very clear.

Rather than thinking about oil markets, let’s think about the market for oil transportation out of Western Canada. The supply curve would be the price-ordered list of transportation options, including existing and proposed pipelines and rail service, and other options including trucks, barges, etc. The demand curve would be a function of the existing and proposed oil sands production, and it would slope downward with higher volumes shipped as the cost of transport decreases, since higher transportation costs would erode profitability and lead projects to be left undeveloped. Obviously, there’s an important time dimension here, but let’s ignore that for now (this is Econ 101), so the basic market for transportation in 2020 might look something like this:


Now, there are various arguments made with respect to KXL, and let’s see if we can sum them up. The first is the oft-heard industry and government argument that “the oil will get to market” no matter what, so there is no greenhouse-gas impact of pipelines, and that would look something like this:

Slide2Now, I don’t think anyone really believes that the demand for oil transportation is completely inelastic to the toll costs. If you want to get industry to agree with that statement, ask them about the impact of greenhouse-gas policy costs on oil sands development, and you’ll hear pretty quickly about competitiveness concerns. The same thing would apply to transportation costs.

Where the dispute really lies in discussion over KXL is with respect to the elasticity of the supply curve of transportation options. If you talk to KXL proponents (or read the State Department report on the pipeline), the conclusion is that even though oil sands production would be responsive to increases in transportation costs, the impact of removing one pipeline (or perhaps even multiple pipelines) from consideration would be small in the impact on per-barrel costs of moving oil sands. In other words, they argue that the market looks something like this:


This supply structure implies that, yes, oil sands production would respond to transportation costs and, yes, oil transportation would at some point get more expensive, but there’s enough scalability in rail that the price of moving the marginal barrel of oil to market and, thus, the equilibrium price discount in the market, is not particularly sensitive to the removal of a pipeline or two from the plans. You could also make the same argument on the basis of the potential for other pipeline routes. Industry and pipeline proponents are happy to make this argument in the context of greenhouse-gas emissions, arguing that the oil will still get to market, and perhaps do so via slightly more emissions-intensive railways. However, they are less enthusiastic when one points out to them that this means pipelines don’t actually add any value to the oil, and won’t reduce the price discount relative to what would otherwise occur.

Opponents of Keystone XL and other pipelines tend to argue the opposite: that rail is not scalable and that the expansion of the oil sands is very dependent on the availability of pipelines. Their market looks like this, where there isn’t enough rail capacity to expand significantly without a large increase in tolls, and the oil sands industry is not sufficiently profitable to afford the tolls associated with rail expansion:


In this market, removing KXL and Northern Gateway, for example, would translate into a reduction in oil sands production and, thus, potentially a ripple effect through world oil markets leading to a reduction in total consumption and, thus, a reduction in total emissions.

So, the Keystone XL debate really is a debate about Economics 101, but it’s not about world oil supply and demand so much as it is about the supply and demand for oil transportation services out of Western Canada and, specifically, about the possibility of alternative pipelines and the scalability of rail. If you believe that rail is scalable at or near current costs to accommodate proposed production growth, cancelling KXL won’t have a large impact on production. If you believe that the supply curve for transportation is steeply upward-sloping because rail can’t be scaled easily, and new pipeline options aren’t likely to take the place of any which are cancelled or significantly delayed, then blocking a pipeline will have an impact.

I guess this is a good Econ 101 problem, after all.

Filed under:

Why the debate over Keystone and emissions comes down to rail

  1. I guess this is a good Econ 101 problem after all.

    Not resolvable at this level without first handing it over to the Operations Management 101 folks. Lots of logistical challenges transporting such large and growing volumes, especially if its not light crude but rather some form of diluted bitumen.

    I could see spending some serious $$$ doing a proper feasibility study, especially if the goal is to get AB bitumen to the east or west coast ports in Canada by rail for export. Heating for loading/unloading/storing/ etc in the cold Canadian winters, and transport through large population centres in Ont/Quebec/Lower Mainland BC or avalanche prone areas in Northern AB/BC just a few.

    Then there’s derailments…

    • Great. Sounds like you have your work cut out for you. Then, summarize it in 1000 words and submit it for publication. See what happens. Oh, you might have to sign your “real” name to it…one of them, anyway.

      • Sorry, did you not get enough special sauce on your Big Mac today? You seem grumpy. I’ll see what I can do…

  2. Re: TCPL’s plans to get into the rail business [huh?] My email comments to a G&M piece May 22, 2014 [I know it’s not twitter, so doesn’t count]

    Keystone indecision spurs TransCanada to study rail alternative

    Still, shipping large volumes of Canadian oil sands crude more than 2,000 kilometres from Hardisty, in central Alberta, to Steel City near the Nebraska-Kansas boundary and terminus of the southern and completed section of Keystone XL would also require large numbers of scarce tank cars.

    Keystone XL is intended to ship about 800,000 barrels of oil daily. At least 10 dedicated 100-tank-car oil trains a day would be required to deliver similar volumes.

    I think you may be grossly underestimating the number of tank cars required. Let me help.

    A rail car for shipping oil holds about 500-750 barrels, so a 100 car train holds about 50,000 – 75,000 barrels. But, because bitumen (if that is what they intend to ship) has a higher density than light crude – the capacity limit may be influenced by weight. Nevertheless 10 unit trains (of 100 – 120 cars ) is ballparkish.

    But because you are shipping 2000 km, you have to account for railcars in transit (going down and returning), and filling/unloading times.

    Say a railcar is allowed to travel at 80 km/h (I think new regs may limit it lower through towns etc), so it would take 2,000/80 or about 25 hrs just to get there with no stopping or slowing down. Double for returning empty, so there’s at least 2 x as many trains needed just in transit. Say, 20 dedicated 100 – train-cars.

    Then you have the time required to load/unload. Loading is less critical, because you can heat and pump from storage tanks. Still, you’ll need trains waiting to get filled.

    But, unloading is the real bottleneck, especially if you are shipping bitumen (diluted or raw) because it is very viscous, and may need to be heated using special cars with heating coils and insulation. May take up to a day to unload each heated car, depending upon outside temp and contents. And expensive multiple unloading facilities. Never mind the rail logistics/switching station. ETC.

    You may wish to revisit the feasibility of such large volumes with TCPL. Sounds fanciful to me.

    • Some quick follow-up calcs for future reference.

      Here’s the specs for a 30,000 US Gallon tank car (715 barrel) – non insulated or heated.

      Length coupling/coupling =~ 60 ft. So, a 100 unit train = 6,0000 ft plus locomotives.

      Say it is limited to 30 km/h (27 ft/s) through a controlled (road) crossing through an urban area. It would take 6,000/27 = 222 seconds to pass (say 4 minutes). Then time before for the crossing barricades to come down/come up. Say 1 minute on either side (give or take).

      So, as a first guesstimate, 6 minutes of delay every time an oil train crosses. And in the back of the envelope calcs – 20 trains (without slowing) per day – say one per hour.

      So, every hour, 10% of the time the road with the rail crossing is impassable – just due to oil shipping (6 minutes out of 60). 24 hours per day. Then there’s other freight.

      The p.r. would be interesting on this operation.

Sign in to comment.