External Costs of Transporting Petroleum Products: Evidence from Shipments of Crude Oil from North Dakota by Pipelines and Rail.

• Author: Clay, Karen
• Position: Report - Statistical table

1. INTRODUCTION

   This paper constructs new estimates of the air pollution, greenhouse gas, and spill and accident costs associated with the long-distance movement of petroleum products by rail and pipelines. Movements of petroleum products, particularly crude oil, have received enormous media attention. Almost all of the attention, including very recent coverage (Smith, 2017), has focused on spill and accident costs despite the fact that air pollution and greenhouse gas costs are also likely to be significant.

   Pollution emissions for pipelines and rail differ from one another in three important ways. First, while emissions from trains occur along the transportation route, emissions from pipelines manifest at the power plants that generate the electricity consumed by pumping stations. The distance between these power plants and the associated pumping stations can be quite large. Second, ground-level emissions, such as those from locomotives, tend to be more harmful than the same level of emissions released from tall smokestacks at power stations (Muller and Mendelsohn, 2009).

   Third, the existing railroad infrastructure moves goods through population centers. In contrast, power plants are typically located in less densely-populated areas. This final difference matters for pollution exposure, because the emissions from trains moving through cities are likely to affect many more people than those emitted at power plants.

   To construct our estimates, we use data on locomotive diesel consumption, pipeline pumping station electricity consumption, locomotive and power plant emission factors, and the AP3 integrated assessment model, which maps county level emissions to costs for counties affected by the emissions. Estimates are constructed for movements of crude oil from North Dakota in 2014, a year in which roughly half of said crude oil was shipped to refineries by rail and half was shipped by pipelines. Locomotive diesel consumption is estimated based on movements of crude oil from the Surface Transportation Board's Confidential Waybill Sample and industry data on average ton-miles per gallon. Our approach to estimating diesel consumption is similar to that taken by U.S. Department of State (2014). Pipeline pumping station electricity data is from Genscape (2014). The AP3 integrated assessment model estimates changes in county-level air pollution based on changes in emissions. To construct estimates of county-level damages, the AP3 model uses census data on population and other county characteristics together with peer-reviewed concentration-response functions and valuations of outcomes used by the EPA. We construct estimates of spill and accident costs from Pipeline and Hazardous Material Safety Administration (PHMSA) regulatory impact analyses. (1)

   Our analysis has three main findings. First, air pollution and greenhouse gas costs are substantially larger for rail than for pipelines. For shipments of crude oil from North Dakota to the Gulf Coast in 2014, the air pollution and greenhouse gas costs are nearly twice as large for rail as for pipelines. (2) We provide evidence that the higher air pollution and greenhouse gas costs for rail relative to pipelines generalize to other products, routes, and years in North America. Second, air pollution and greenhouse gas costs are three times as big as spill and accident costs for rail and ten times as big for pipelines. Thus, the policy debate surrounding crude oil transportation has likely put too much relative weight on accidents and spills, while overlooking air pollution and greenhouse gas emissions, which appear to be a far more serious source of external cost. Third, the air pollution and greenhouse gas costs of rail transportation of crude are about one-fifth of the air pollution and greenhouse gas costs from combusting their refined end-products in motor vehicles and about one-fifth the private cost to the shipper of rail transport. The costs for pipelines are one-tenth the costs of combusting fuel in motor vehicles and one-fifth of the private costs of shipping crude oil.

   This paper contributes to the literature on transportation of crude oil. In particular, it complements work by Covert and Kellogg (2017) on the flexibility of crude by rail and the effect of rail prices on optimal pipeline capacity investment. They calculate that if rail had to internalize pollution externalities of $2 per barrel, which is roughly the estimated value in this paper, the capacity of the Dakota Access Pipeline would have been 59,000 barrels per day above its current level of 470,000 barrels per day. Similarly, we connect to work on the broader impacts of increased shipments of crude by rail such as that by Bushnell et al (2017) which examines the effects of heightened crude oil traffic on wheat, corn, and soy prices in North Dakota and nearby states.

   This paper also contributes to the broader literature on air pollution associated with transportation. For a summary, see NRC (2009) on combustion costs. Estimates of air pollution and greenhouse gas damages from rail transport are largely absent from the recent literature. For example, perhaps the most recent estimates for rail are those of Forkenbrock (2001) which were derived from cost per ton estimates by Haling and Cohen (1995), which are in turn based on estimates from a 1993 study by National Economic Research Associates (NERA) that looked at electric utility resource selection in Nevada.

2. BACKGROUND

   Figure 2 shows that oil production in the United States increased tremendously beginning in 2008. One driving factor was the rise of production in the Bakken Field, which is primarily located in North Dakota. As a result of this rapid increase in production, in 2014 North Dakota was the third largest producer of oil in the United States after Texas and the federal offshore region in the Gulf Coast.

   Shippers send oil to the location that provides them with the largest revenue net of transportation cost (termed "netbacks"). Firms move oil to a rail or pipeline terminal using either truck or gathering pipelines. (3) Our analysis focuses on the long distance transportation of crude oil and excludes these 'first miles' primarily because of difficulties in obtaining detailed data on how oil was moved from the wellhead to a terminal. (4) Firms can ship crude oil to refineries using a range of modes of transportation, including rail, pipeline, or some combination of rail, pipeline and water. In this last case, oil is initially transported by rail or pipeline and is then offloaded to tankers or barges for shipment to refineries. In 2014, U.S. refineries reported receiving 78 percent of domestically produced crude oil by rail or pipeline. (5) They reported receiving the remaining 22 percent by tanker, barge, or truck.

   Firms shipping crude oil from the Bakken are likely to use pipelines to the extent that there is available capacity, because rail is significantly costlier than pipeline per barrel-mile shipped. For example, Frittelli et al (2014, p. 7) noted that: "Railroad transport reportedly costs in the neighborhood of $10 to $15 per barrel compared with $5 per barrel for pipeline." This is consistent with information from Genscape's Petrorail Report (various dates) and with the prices reported in Covert and Kellogg (2017). Additional oil is shipped by rail to the extent that it is economically attractive to do so. Rail offers some benefits over pipelines, because it serves a more flexible set of destinations with faster delivery times. Another factor in the continued use of rail--even when the economics appear unfavorable--is the fact that some shippers entered into multi-year contracts for rail shipments when the price of oil was high. These "take-or-pay" contracts require them to ship oil by railroad or pay for unused capacity.

   Roughly half of the oil shipped from North Dakota in 2014 went by pipeline to refineries and the vast majority of oil traveling by pipeline went to refineries on the Gulf Coast. Most of the oil moving by pipeline ends up in the Gulf Coast, because there is almost no crude oil pipeline infrastructure on the East and West Coasts. (6,7) The other half of the oil shipped from North Dakota in 2014 went to refineries by rail, with the largest share going to the East Coast. (8)

   At refineries, the increased availability and favorable pricing of domestic oil led to reduced use of foreign oil. For example, in the U.S. between 2011 and 2014, domestic crude oil received at refineries increased by about 1 billion barrels and foreign oil decreased by about 500 million barrels. In total, crude oil consumption by refineries increased by about 9 percent.

   A central issue in North Dakota is the extent to which trains carrying crude oil crowd out the rail transportation of other products. When evaluating the air pollution effects of crowding out, it is helpful to think about the two ends of the spectrum. The first extreme is that railroads are operating with excess capacity and so additional rail traffic has no congestion externality on the rail system. In this case, any additional crude-by-rail traffic increases the overall air pollution costs from transporting goods by rail. The other end of the spectrum is that railroads are already operating at full capacity. In this case, increased crude-by-rail traffic completely crowds out lower value products. There is no change in the total rail traffic due to crude-by-rail in this scenario, and thus there is no change in the overall air pollution costs from transporting goods by rail. In fact, the crowding out of other products may cause additional pollution, if these products are transported by other higher pollution modes such as trucks instead. Thus, there are three possible sources of air pollution: i) railroad-related pollution from increases in rail traffic; ii) any additional pollution related to system-wide railroad congestion, since slower speeds are associated with...