Data Sources and Methodology

The Carload Waybill Sample is a stratified sample (i.e. classified by sub-categories) of select waybills submitted to the Surface Transportation Board (STB) by those rail carriers which terminate at least 4,500 revenue carloads per year. The sampling methodology is based exclusively on the number of carloads in the shipment. The sample is stratified in that each class of trains – characterized by the number of carloads contained in each train set – is sampled at a different rate. For shipments of 101 or more carloads, rail carriers must submit every other waybill for a given year, thereby generating a sample of 50 percent. As the number of carloads in a shipment decreases, rail carriers must submit a lower percentage of waybills. For instance, for shipments of 1 or 2 carloads, rail carriers must submit one out of every 40 waybills, thereby generating a sample of 2.5 percent of the population.

Unlike the public use version of the waybill data, which STB makes available on its website, EIA obtained the confidential version (referred to as "900-byte" by the STB) which includes greater detail such as rail carrier name; and Freight Station Accounting Codes for origin, destination and intermediary stations (the public use file groups shipments by region); and associated revenue, all of which are deleted or "masked" in the public use file. Caution must be used when relying on the revenue field. While that data includes actual contract revenues for each shipment, it also includes (where applicable) an amount of revenue associated with fuel surcharges. Recently, the STB ordered that fuel surcharges be shown as a separate revenue field in future waybill reporting, and EIA expects separate reporting of revenue and fuel surcharges for the 2009 dataset. In addition, contract revenues on the waybills do not necessarily represent the actual amount paid to the rail carrier due to possible rebates and other discounts that are settled periodically. Nonetheless, the revenues reported here through 2008 reflect EIA's best estimate of a shipment's transportation rate.

The purpose of the Carload Waybill Sample is to analyze shipments across the entire country as a whole and not along any particular route. However, EIA determined that it could use the Coal Waybill Data to reasonably approximate average coal transportation rates in the U.S. and on a State-to-State (and Coal Basin-to-State) basis for two reasons:

• Coal is most often shipped in trains of at least 100 carloads and thus subject to a 50 percent sampling rate (according to the Association of American Railroads (AAR), around 95 percent of coal transported by rail moves in such trains); and

• Coal is shipped by rail in such volume that the sampled waybills produce a particularly large number of shipments (on average, more than 35,000 waybills were sampled annually between 2001 and 2008, representing almost half of all coal shipped by rail in the United States).

While EIA's new Coal Transportation Rata Data is based primarily on the confidential Carload Waybill Sample obtained from the STB, the published results are also the result of additional adjustments and improvements. The biggest improvement entailed mapping rail carrier stations with actual mines and coal-burning electricity generating plants so that EIA could differentiate (and isolate) those coal shipments pertaining to electric generators from those going to/from various ports, junction stations, and coal-fired manufacturing facilities. EIA accomplished this by manually reviewing a variety of industry publications, reference books, and websites. While EIA was not able to identify all of the stations appearing in the Carload Waybill Sample (especially the smaller ones, which handle a small number of carloads per year), those stations that EIA did identify handle approximately 95 percent of all coal shipments by volume included in the Carload Waybill Sample. Moreover, EIA was able to successfully identify 187 of the 200 largest coal shippers and 194 of the 200 largest coal receivers by volume.

By limiting its analysis to only those shipments that originated at a known coal mine and terminated at a known coal-burning electricity generating plant, EIA was able to achieve two key results. First, EIA was able to isolate those shipments that directly affect the U.S. electricity industry; second, EIA was able to eliminate waybills terminating at interchange points or ports rather than at a coal-burning electricity generating plant. (Those shipments typically billed by one railroad for their "leg" of a through movement are known as the Association of American Railroads' Accounting Rule 11 shipments). By excluding such waybills from its analysis, EIA was able to ensure that its results were not influenced by those intermediate shipments that could skew results for a State with numerous such junction points, for example, Illinois (Chicago Gateway) or Missouri (St Louis). As the majority of shipments billed under Rule 11 originated in the Powder River Basin and terminated in the eastern United States, users should exercise caution when drawing conclusions about these shipments because the new Coal Waybill Data only captures waybills billed as "through rates" (i.e. the costs of both legs were combined in a single invoice rather than being billed separately under Rule 11).

EIA also strengthened the data by including additional data such as price deflators and coal basin definitions. EIA also adjusted the Carload Waybill Sample for changes which might have occurred since the record was submitted (e.g. changes to county names or sales of short-line rail carriers) and for errors such as duplicate waybills and obviously incorrect data.

In addition to strengthening the data itself, EIA also needed to determine whether it accurately represented coal transportation by rail in the United States. EIA conducted an extensive analysis by comparing the data to data it collects on EIA surveys and to publicly-available industry reports. EIA concluded that, subject to the limitations explained above, the data is a reliable estimation of rail transportation of coal in the United States for the years 2001 to 2008. Note that EIA also received from the U.S. Surface Transportation Board carload waybill samples for the year 2000 but determined that it did not accurately represent coal movements due to a significantly smaller sample size relative to the data received for the other years.

Once EIA had completed the Coal Waybill Data, EIA applied the suppression methodology stipulated by STB pursuant to its codified regulation to ensure that no confidential data was released. The STB has reviewed our results and has agreed to the release of the data in EIA's Coal Waybill Data. The methodology states that data can only be published for those aggregations where there are at least three different freight stations at the points of origin and destination as identified by the Freight Station Accounting Code (FSAC) on one railroad or there must be at least two or more FSACs than there are railroads present in the waybill data being aggregated at both the points of origin and destination.