Missouri River Navigation Water Compelled Rates - Update

colored bar
Bottom of Page]


      Rail Rates and the Availability of Barge Transportation: The Missouri River Basin (February 1997), contains an analysis of the impact of available navigation on railroad rates to, from, and within the Missouri River region based on the Surface Transportation Board's 1992 Carload Waybill Sample (CWS).1 Even though these base data are less than six years old, a number of fundamental institutional changes have altered the competitive relationships both within and between surface transport modes.2 As a result, both the quantitative and qualitative relationships identified within the 1996 analysis may no longer be accurate. The uncertainty surrounding the validity of these results, in combination with a desire for temporal consistency, has motivated the U.S. Army Corps of Engineers' Missouri River Region, Northwest Division to seek an supdate to the 1996 study.

      The underlying structure and methodology employed in producing this update is very nearly identical to the methodology used within the 1996 analysis. Readers are, therefore, advised to refer to the original document for descriptions of underlying theory, data preparation, and estimation techniques. The primary focus of this text is to explain the few instances in which the methodology was modified and to present (and interpret) the updated estimation results. The remainder of this text is organized as follows: Section A2 provides a brief summary of methodological modifications; Section A3 contains estimation results; and summary remarks are offered in Section A4.


      The updated analysis contains only four significant methodological modifications and two of these are unique to the estimation of linking available navigation to rail rates for the movement of coal.

      In the case of all commodities, previously described measures of the distance to the nearest navigable waterway were abandoned in favor of new measures based on the use of Geographic Information Systems (GIS). First, the analysis identified the centroid of each county within the contiguous 48 states. Next, technical specialists calculated the Euclidean (straight-line) distance between county centroid and the nearest navigation facility. It is believed that this methodology provides a more accurate depiction of available navigation than past distance-to-water measures.3

      As with the 1996 analysis, the updated estimates are controlled to exclude the influence of available navigation on the upper reaches of the Mississippi River. However, because the Mississippi's influence over rail rates has also been affected by changes within the transportation sector, upper Mississippi estimates developed in the course of the recently completed lower Mississippi water-compelled railroad rate analysis were used in place of the upper Mississippi measures employed in the 1996 analysis.

      The remaining two methodological modifications relate specifically to estimates for coal. The reader will recall that in the analysis based on 1992 data, the estimated model contained two splines or points of discontinuity in its treatment of the relationship between railroad rates and available navigation. The first of these breaks was at a distance of 10 miles from the waterway and captured the strong influence of barge transportation within that 10 mile range. The second spline was set at a distance of 20 miles from the waterway and depicted a much less formidable navigation influence on railroad rates up to that distance. The latter of these influences was attributed to competition between eastern and western coal - competition that was extremely vigorous in 1992. In performing the update using 1995 data, TVA attempted to fit a similar double spline. However, repeated estimation attempts revealed that the second spline - the one partially capturing east-west competition - is no longer valid. Thus, it must be concluded that, by 1995, the battle ground where eastern coal and western coal met head-to-head had moved east of the Missouri River Valley. The current specification includes only a single spline set at a distance of 10 miles from the river.

      Finally, unlike previous estimations, the current specification for coal includes variables that reflect both the BTU and sulfur content of the coal being transported. All else being equal, one would expect that coal with higher relative BTU content or a lower relative sulfur content would provide a greater premium from which railroads could extract their share of supra-normal profits. Thus, one would expect the coefficient for the BTU variable to be positive and the coefficient for the sulfur variable to be negative if rail-served markets for the movement of coal are characterized by a lack of effective competition.


A3.1 General Results

      A full set of statistical outputs is included at the end of this appendix. Generally, however, these results are similar to those that were based on 1992 data. Rail rates decrease as the tons per carload, shipment size, and shipment distance increase. Interchange adversely affects available rail rates for grain, while improving rates for the movement of coal. Market concentration, where statistically significant, more often than not, leads to increased rail rates, and some commodities move in corridors where additional traffic density produces economies that allow for lower rates, while other commodities move in lanes where increased traffic leads to congestion, higher costs, and increased railroad rates.

      Interestingly, the two new variables included to reflect coal quality display signs that are opposite those anticipated. The variable measuring BTU content is negative and significant, while the variable measuring sulfur content is positive and significant. The result with respect to sulfur content is likely the result of the relatively low rates that Burlington Northern - Santa Fe and Union Pacific charge for the movement of low sulfur Powder River Basin (PRB) coal. The estimated negative relationship between BTU content and observed rail rates has no immediate explanation. A3.2 Commercial Navigation and Railroad Rates

      The 1996 analysis based on 1992 data produced results for seven separate commodities - corn, wheat, soybeans, coal, fertilizer materials, finished fertilizers, and scrap metal. The updated analysis considers these same commodities, except models for three fertilizer components (potash, phosphates, and liquid nitrogen solutions) are estimated separately. As with the original estimates, the updated results reflect a measurable relationship between available navigation and railroad rates. Nonetheless, the magnitude of the economic transfers between railroads and shippers was decidedly smaller in 1995.

      Table A1 provides a comparison of the updated results with those produced using 1992 data. For those commodities where direct comparisons are possible, the magnitude of water-compelled rate savings fell by roughly 60% over the three year period between 1992 and 1995. While the maximum effective range of waterborne competition did not change radically, the mean per ton-mile savings accruing to affected shippers fell by at least 90% and by as much as 99%. Clearly, in 1995, available barge transport did not play the same role in the formulation of railroad rates as it did in 1992. The two most relevant questions are: (1) is the observed reduction in water-compelled influence the result of a decline in navigation's ability to discipline rail rates or did some other competitive force usurp barge transportation's role; and (2) is the observed decline in the competitive influence of waterborne commerce in the Missouri Valley simply an anomaly or does it reflect a more pervasive intertemporal and cross-sectional trend? These questions will be addressed more fully in Section A3.3.

Table A1.

Year Corn Wheat Soybeans Coal Fertilizer Materials4 Fertilizer Finished5 I&S Scrap
Competitive Range of Water - Origin (miles)
1992 45 160 60 n/a No valid Comparison 30
1995 50 180 30 n/a No valid Comparison 30
Competitive Range of Water - Destination (miles)
1992 60 60 n/a 10 No valid Comparison 40
1995 50 n/a n/a 10 No valid Comparison 50
Affected Regional Railroad Tonnage (millions)6
1992 19.3 12.4 5.0 21.5 No valid Comparison 1.0
1995 17.2 9.8 4.9 43.4 No valid Comparison 1.0
Total Regional Railroad Tonnage (millions)
1992 32.4 16.8 6.5 66.3 No valid Comparison 1.4
1995 40.7 14.3 7.6 105.7 No valid Comparison 1.9
Mean Revenue per Ton-Mile Transfer (cents)
1992 0.425 0.740 1.951 0.287 No valid Comparison 2.303
1995 0.007 0.071 0.014 0.037 No valid Comparison 0.227
Total Regional Water-Compelled Transfers and Class I Revenues
1992 $45 $56 $35 $43 $6 $12 $6
1995 22.5 11.8 2.7 27.3 2.5 10.0 2.8
  -50% -78% -89% -36% No valid Comparison -53%


A3.3 Interpreting the Results

      Table A2 provides 1992 and 1995 summary statistics for five commodities where a direct comparison between the two periods is possible. Even a cursory glance at these data suggests there may not be one single explanation for the reduction in water- compelled rate effects in the Missouri River basin. Instead, it appears that causation may vary across commodities.

      In the case of corn, Table A2 indicates a 21% increase in shipment distance that probably owes to an increase in export shipments to the Pacific Northwest (PNW). Any shift in market conditions that favors the PNW simultaneously makes it somewhat easier for rail carriers pay less attention to barge competition. However, these data cannot make clear how much of the decline in the water-compelled rate effects for corn movements is attributable to increased PNW traffic nor can they indicate whether the shift in market conditions that produced this outcome are permanent or transitory in nature.

      Table A2 also reveals a sharp decline in shipment size for wheat and soybeans. The average shipment tonnage for wheat fell by 35% between 1992 and 1995 while the average shipment tonnage for soybeans fell by 31% during the same period. In both cases the mean tonnage is still well in excess of the 1,500 tons needed to fill a conventional covered hopper barge. It is clear, however, that a larger percentage of wheat and soybean movements from the Missouri River basin in 1995 may have been too small to effectively utilize barge transportation. Again, it is impossible to determine the degree to which the decline in shipment size affected the competitive influence of barge transport. It should, however, be noted that the lower shipment sizes for wheat and soybeans appear to be related to poor 1995 soybean and wheat yields - particularly in Kansas and Nebraska. Shipment sizes for both commodities rebounded to 1992 levels in 1996.7

Table A2.


Year Corn Wheat Soybeans Coal I&S Scrap
Mean Revenue per Ton-Mile (cents)
1992 2.70 2.73 3.45 1.72 6.26
1995 2.64 3.00 2.84 1.51 6.09
Mean Shipment Distance (miles)
1992 825 722 627 660 336
1995 998 752 741 717 418
Mean Shipment Tonnage
1992 4,581 3,367 4,072 10,913 133
1995 4,836 2,182 2,814 11,586 108
Mean Number of Carloads per Shipment
1992 46.8 34.1 42.5 106.6 1.7
1995 48.5 22.1 29.1 107.0 1.3


      In the case of coal, the percentage reduction in water-compelled rail rate effects is somewhat smaller than it is for corn, wheat, and soybeans. Nonetheless, in the aggregate, estimation results suggest that the reduction in barge-to-rail competition cost Missouri basin coal shippers $15 million in 1995. As with grains, the summary statistics in Table A2 provide some clues to those attempting to explain the observed reduction in water-compelled rail rate effects.

      Between 1992 and 1995 nominal rail rates for the movement of coal to, from, and within the Missouri River basin fell by more than 12% Hence, in the case of coal, it appears that changing market conditions may have been generating competitive forces that made the discipline offered by available barge unnecessary. This hypothesis is consistent with the nationally observed trend under which rail rates for the movement of high sulfur Illinois basin and northern Appalachian coal have steadily fallen. Essentially, as clean air standards have made coal from these regions increasingly less attractive, any previously available premiums have been squeezed out of observed prices. This is true for both mine-mouth prices and transportation rates. Eventually, if this trend continues unabated, coal from the Illinois basin and northern Appalachia will be entirely displaced by low sulfur coal from the Powder River basin, Rocky Mountain states, or southern Appalachia. It is worth noting that the decline in rail rates for coal is even more pronounced in Illinois, Wisconsin and Minnesota, again suggesting that the battle lines between eastern and western coal continue to shift east and north.

      Finally, it should be noted that, while the magnitude of water-compelled rate effects changed measurably, the distances over which available barge transportation affects rail rates changed very little between 1992 and 1995. This is, in fact, what economic theory predicts. While the magnitude of water-compelled rate effects is a function of railroad costs, barge costs and market demand conditions, the range over which these effects are observed is primarily determined by the costs of moving shipments from their actual origin to the point where they can be loaded onto barge. Thus, given that motor carrier rates and handling costs did not vary significantly over the 1992-1995 period, one would expect that the effective range of barge competition also did not vary.


      The empirical analysis based on 1992 data indicated that available Missouri River navigation resulted in railroad rate reductions of roughly $203 million for that year. A nearly identical analysis based on data from 1995 indicates that the water-compelled rate effects for that year are barely more than one-third of their 1992 level. Moreover, this large reduction in the competitive influence of commercial navigation on the Missouri would appear to have little to do with navigation conditions or the cost of providing barge transport. Instead, the reductions in the Missouri's competitive role appear to stem from exogenous economic changes in the markets for the dry-bulk commodities shipped to and from the Missouri River basin.

      Ultimately, it will be tempting for some to conclude that the observations from these two water-compelled rate investigations reflect a pervasive and non-transitory trend under which available Missouri River navigation plays an ever diminishing competitive role in surface transport markets. The reader should be advised, however, that these analyses do not support such a conclusion. Clearly, two points in time do not constitute the extensive time-series necessary to identify an intertemporal pattern. Moreover, even if 1992 and 1995 reflect a more pervasive trend, there is nothing to suggest that this trend is non-transitory. To the contrary, the causal factors that at least partially explain the 1995 reductions in the Missouri's competitive influence were largely exogenous and often temporary. It is, therefore, possible - even likely - that further exogenous changes have and will continue to redefine the Missouri's importance as a competitive alternative to rail transport.

      From a policy perspective, the erratic magnitude of water-compelled rail rate effects is, perhaps, confounding. It is difficult to say with confidence that either estimate is more valid than the other as a predictor of future economic importance. What remains clear, however, is the irrefutable conclusion that available commercial navigation on the Missouri River can provide necessary competition to rail carriage under a wide array of historically observed economic conditions.

  colored bar


1 The actual analysis was performed in the fall of 1996. At that time, 1995 CWS records were not as yet available. Given that 1993 and 1994 provided abnormal navigation seasons, 1992 was the most recent year on which the analysis could be based.
    Back to Text

2 These changes include the adoption of notably stricter air quality standards, the elimination of the Interstate Commerce Commission and the implementation of additional deregulation, and three major rail system consolidations.
    Back to Text

3 TVA is currently in the process of developing a third generation distance-to-water measure that will provide actual highway distances to the nearest appropriate dock facility based on preferred routings and specific commodity characteristics. These data should become available in the summer of 1998.
    Back to Text

4 For 1992 Fertilizer materials included potash and urea. For 1995, this category includes potash and phosphates.
    Back to Text

5 For 1992 Finished Fertilizers included liquid nitrogen solutions and dry fertilizers, NEC. For 1995, this category includes only the nitrogen solution.
    Back to Text

6 Readers may observe that the 1992 affected tonnages presented here differ from those indicated within the original analysis. These differences owe to a modification in the assumptions underlying the calculations. Specifically, the original analysis assumed that, in order to be affected the, tonnage must be within the competitive range of navigation at both origin and destination. In the current analysis, available navigation at either origin or destination is sufficient to produce some degree of water-compelled effects. It should be noted that this latter assumption was used to calculate the magnitude of shipper savings in both analyses.
    Back to Text

7 It is worth noting that the Burlington Northern responded to the poor yields by providing unit train rates for Kansas City wheat and soybean shipments as small as 25 carloads These unit train rates were not available at other locations and were discontinued the following year.
    Back to Text

[ Top of Page ]   [ Master Manual Technical Reports Page ]