O’Rourke, Laurence 1 The Impact of Differential Pricing on Barge Freight Transportation Prepared for the Transportation Research Board Submitted July 31, 2002 Word Count 5,649 + 1 Table Laurence O’Rourke Senior Associate ICF Consulting 9300 Lee Highway Fairfax, VA 22031 lorourke@icfconsulting.com Phone: 703-934-3186 Fax: 703-934-3740 TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 2 ABSTRACT Through the Staggers Act (1980) and the Railroad Revitalization and Regulatory Reform Act (1976), Congress deregulated railroad pricing to improve the financial health of the industry. Deregulation legalized differential pricing; the policy of charging customers different prices according to their willingness to pay. While the railroads have returned to profitability, shippers have been angered by railroad pricing strategies that are seen as abusive. Railroads have refused to quote rates to competing transportation facilities or have set prices to divert traffic onto the rail network. This study seeks to measure the impact of differential pricing of rail services on barge transportation in the Ohio River Basin. It provides an economic measurement of this effect by constructing a model to predict freight traffic volumes at barge terminals in the Ohio River Basin. Keywords: Differential Pricing, Intermodal, Modal Diversion, Barge Freight, Railroad Market Power TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 3 INTRODUCTION In 1980, Congress passed the Staggers Rail Act. The act mandated the end to rigid railroad rate regulation by the Interstate Commerce Commission (ICC). Differential pricing policies legalized by the Staggers Act have been a source of significant political debate. In freight markets where competition exists from other transportation modes, rail rates are set close to variable costs. For captive shippers who are served by only one railroad, rates are often 300% of variable cost. (Kahn, 1999) Shippers have alleged that railroad rates to competing modes of transportation have been increased to divert traffic onto the rail network. For instance, in the Ohio River Basin, railroads may be increasing rates on rail movements to intermodal barge terminals in order to divert freight traffic to rail lines running parallel to the river. Industry proponents have argued that such diversion of traffic benefits all shippers by increasing the traffic on the rail network and allowing railroads to spread fixed infrastructure costs over more freight shipments, thus lowering costs and prices for everyone (Cunningham and Jenkins, 1997). This research conducts a measurement of the impact of differential pricing on the transportation of freight by barge in the Ohio River Basin. The pricing and service policies of railroads are currently issues of great contention. The railroads have argued that shippers have access to regulatory remedies to abusive pricing practices through the rate reasonableness and competitive access provisions of the Staggers Act. Shippers have insisted that abuse of railroad market power is having a substantial economic impact. Public policy makers have also been concerned about the size of the freight traffic diversion because of the potential environmental impact. Barge freight movements are less energy intensive and cleaner per ton-kilometer for some air pollutants. This research tests the hypothesis that differential pricing has diverted traffic from barge to rail. The paper is divided into six sections. Section 1 describes the theoretical and historical background of the differential pricing issue. Section 2 describes the data and methodology employed. Section 3 presents the results of the analysis. Section 4 discusses these results and places them in the context of other research. Section 5 provides some closing comments and conclusions from this research. HISTORY AND THEORY OF DIFFERENTIAL PRICING The regulation of railroad market power is a policy issue with much history. The Interstate Commerce Commission (ICC) was created in 1887 to restrain railroad market power and pricing practices. On local lines where the railroads held monopoly positions, they frequently charged small shippers higher prices for moving freight on short hauls than where charged for competitive long haul movements. The Commerce Act, which created the ICC, specifically forbade such long-haul/short-haul pricing practices. The law also stated that rates should be "just and reasonable". The law established the ICC as the agency that would hear shipper complaints and required the railroads to file public rates. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 4 Over time, the extent of railroad regulation grew. In the Transportation Act of 1940, the use of umbrella rate making was encouraged. Under this policy, the ICC considered the impact that railroad rates would have on competing modes of transportation. Umbrella rate making tended to favor other modes of transportation. ICC regulation was slow to accommodate changes in the economy following World War II. Less regulated transportation modes were able to capture freight from the railroads. Between 1947 and 1977, while the freight market grew by 91%, the volume of tonnage hauled by railroads decreased. (Congressional and Administrative News, 1980) Truck and barge operators captured this freight traffic. By the mid-1970's, several major railroads where on the verge of bankruptcy and the industry as a whole was experiencing poor financial performance which impeded its ability to attract new capital. Several pieces of legislation where enacted to improve the profitability of the industry. The Railroad Revitalization and Regulatory Reform Act of 1976 limited the scope of ICC ratemaking authority to those situations where a railroad had "market dominance". The use of umbrella rate making was abolished. The profitability of railroad capital became an explicit goal pursued through the regulatory process. The law also made provision for expedited merger proceedings. In 1980 the Staggers Act deregulated the industry further. The purpose of the act was to rely on market forces and the shipper demand for transportation services to set prices. The act formally addressed the issue of differential pricing by establishing a range of differential pricing which was considered reasonable on its face. Additionally, the law placed the burden of proof for challenging the reasonableness of a rate on the shipper. Private contracts where legalized, allowing railroads to negotiate separate and unpublished rates with shippers. During the 1980's, the railroads underwent fundamental restructuring. Through downsizing, consolidation and the use of differential pricing, the railroads increased their profitability and productivity. The result of deregulation was the concentration of market power in the railroad industry. In 1976 there were 63 Class I railroads in the U.S, by 2002 there were only 8. Four megacarriers currently control 95 percent of the gross ton-kilometers moved on the system. With the ICC focused on returning the industry to profitability, railroads were given free rein to consolidate their market positions. Many shippers believe that the lack of competition has removed the incentives for railroads to provide quality service. A GAO report found that a majority of shippers believed that their service has deteriorated since 1990. (GAO, April 1999) The ICC Termination Act of 1995 carried deregulation in a new direction by abolishing the ICC and transferring some of its functions to the Surface Transportation Board (STB). The STB is located in the Department of Transportation. One of the motivations behind the law was to obtain a greater balance between shipper and railroad interests. Many lawmakers felt that the ICC was beholden to the interests of the railroad industry. By placing the STB in the Department of Transportation, and requiring that it be re-authorized every two years, lawmakers sought to make the agency more accountable. The act also required that a simplified procedure for rate complaints be established which would expedite shipper complaints. In 1999 a number of legislative proposals were introduced to address shipper concerns. The "Railroad Competition and Service Improvement Act of 1999" (S.621) proposed to expand the power of the STB to consider competitive access solutions in a range of cases where the market dominance of an existing railroad had not been established. The law called for the development of measures of railroad service quality and proposed to allow the STB to mandate competitive access solutions where rail service falls below a certain threshold. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 5 The legislation endorsed several different competitive access remedies. It proposed to allow the STB to mandate trackage rights in a more expansive set of circumstances. Trackage rights are legal agreements that allow the trains from one railroad to run over the lines of another. The law proposed to overturn the STB's "bottleneck decision". Railroads are currently not required to quote rates from every origin and destination point on their system. Railroads often refuse to quote rates at certain choke points on their network to restrict access of shippers to competing transportation facilities. S.621 sought to overturn these practices and promote competition by requiring railroads to quote rates to all switching points on their network. S.621 also sought to cap freight rates for small agricultural shippers at 180% of variable cost. In short, the law was a direct attack on differential pricing and the philosophy of the Staggers Act. These congressional proposals proved to be too politically contentious and did not become law. The recently introduced “Railroad Competition, Arbitration and Service Acts of 2002” has proposed a more limited set of reforms. The act seeks to offer more service choices to shippers by promoting the use of trackage rights for unit trains and haulage agreements for carload traffic. Additionally the law seeks to promote competition by eliminating paper barriers to smaller carriers competing in new markets. Lastly, the law seeks to create a clearly defined final offer arbitration system that will create a more efficient and streamlined dispute resolution process for rate cases. The proponents of competitive access solutions for rail service argue that it is the analog of the de-regulation that has occurred in the gas pipeline, electric utility and the telecommunications industries. In these cases, monopolists were required to make their networks available to competitors at a rate that allowed them recover their costs. (Faucett, 1999) The competition that ensued spurred innovation and resulted in cost savings for the consumer. Opponents of a more competitive regime for railroads argue that differential pricing is economically efficient and that most shippers would be worse off if railroads where unable to price differentially. Railroads have high fixed costs to maintain their network. These costs are not assignable to any particular shipment, but are the overhead of running a railroad. The marginal cost of moving a given train over this network, the "variable cost", is often very low. Railroads face the dilemma of how to allocate their fixed costs to specific shippers. Their solution is to price differentially. Those shippers who are most price sensitive, such as those who can switch to truck or barge, are charged lower prices. These prices recover the variable costs of the movement and allow the railroad to spread a small fraction of its fixed costs onto these shippers as well. For those "captive shippers", who do not have any other transportation options, the railroad charges higher prices, thus allocating the majority of the fixed costs of the system to them. Some economists argue that differential pricing benefits even captive shippers. If competition or regulation forced prices to be equalized, those shippers who are the most price sensitive would exit the system. The remaining captive shippers would then pay the full fixed cost of the rail network. (Erickson, 1999) Differential pricing is thus a rather slippery issue. Those who favor competition are really making an argument for a more extensive regulatory regime to guarantee competition. The history of industry regulation at the ICC is not one that should instill one with confidence. The current regulatory regime, managed by the STB, is expensive, cumbersome and little utilized by shippers. They find the agency to be responsive primarily to the interests of the railroads. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 6 Those who support the policies embodied in the Staggers Act and the reliance on demand based pricing are making the case for the free exercise of market power by a small number of quasi-monopolists. This can also be a rather economically costly policy, as recent service and system failures have shown. This paper contributes to this debate by conducting an economic measurement of the impact of railroad market power on barge transportation. Defining and measuring the magnitude of impacts across different modes of transportation is an important first step in addressing the issue of railroad pricing. Rail pricing strategies that shift freight away from barge have a wide variety of impacts. While some academics have argued that the pricing practices of railroads are not inherently inefficient (Kleit, 1990), the impacts of the exercise of rail market power may have environmental externalities associated with them. (USACE, Rock Island, 1996) Additionally, the regulations governing rail service pricing provide a mechanism to transfer wealth from one set of private agents to another. Even if this transfer does not reduce efficiency, it is a substantial concern of policy makers. DATA AND METHODOLOGY Several data sources were used for this research. These include the U.S. Army Corps of Engineers (USACE) Waterborne Commerce Statistics, the Navigation Data Center Publications and Waterway Data CD, the Inland River Guide, the Association of American Railroads revenue data and the Bureau of Economic Analysis (BEA) industry earnings data. Each of these data sources is described in more detail below. Waterborne Commerce Statistics The U.S. Army Corps of Engineers collects waterborne commerce statistics on all barge freight movements on the inland waterway system. This information includes dock of origin, dock of destination, four-digit commodity code description and tons moved. The data are gathered from freight tariffs submitted by barge operators. The data contain fields describing other levels of geographic detail, including port equivalent origin and destination, county origin/destination, state origin/destination, and BEA origin/destination. Every barge freight movement on the inland waterway system is represented in the data. A subset of this data consisting of all commodity movements on the Ohio River System was used for this analysis. The Ohio River System includes the Ohio River and all of its tributaries. The river system encompasses the Tennessee River, the Cumberland River, the Green River, the Kentucky River, the Big Sandy River, the Kanawha River, the Monongahela River, and the Allegheny River, among others. Every freight movement with an origin or a destination on the river system was represented in the data for the years 1970-1996. Each record in the data file represents the total annual movements, in tons, of a volume of one commodity from a specific dock of origin to a specific dock of destination. The database contains 182,804 records. To facilitate the analysis, the four digit commodity codes were aggregated into seven commodity groups. These were agricultural chemicals, coal, coke, grain, industrial chemicals, petroleum products and all other commodities. These categories were chosen to focus on major waterway commodities. Analysis of grain, coal and industrial chemicals was particularly important since these commodities are the ones for which shippers have been the most active in seeking rate relief. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 7 Navigation Data Center Publications and Waterway Data CD The Army Corps of Engineers conducts periodic surveys of barge terminal operators and their facilities. An electronic database with information on all of the terminals on the inland waterway system is available to the public from the Navigation Data Center. Information compiled includes port and dock code, a description of the terminal facility, types of equipment on site and a field describing the rail access available at the terminal. About 1,200 terminals on the Ohio River System were reviewed and coded for rail access. Those terminals served by a rail line not connected to the rail network where coded as not having rail access. Inland River Guide The Inland River Guide, published by the Waterways Journal, contains information on the terminals, barge operators, boat stores and shipyards on the inland waterway system. Of particular relevance to this project was the designation of terminals as "private" or "public". Public terminals are those which serve numerous shippers. Private terminals are docks that have been constructed by a factory or electric utility to meet its own needs. It was felt that differential pricing would likely have a greater impact on public facilities. Commodities shipped to private terminals are most likely to be used at the site, and not require any further transportation. Commodities shipped to public terminals are likely to require additional transportation (truck or rail). Public terminals are thus much more likely to be effected by railroad pricing policies. Of the approximately 1,200 barge terminals in the study region, 85 were coded as public terminals. Association of American Railroads Revenue Data Data on average annual revenue per ton-kilometer from 1983 to 1996 was obtained from the Research Department at the Association of American Railroads. Revenue per ton-kilometer reflects the total cost charged to shippers divided by the total ton kilometers moved by the railroads. This data was employed as a proxy for railroad productivity improvements. Over time, revenue per ton-kilometer has declined. BEA Region Earnings Data This data consists of earnings data by industry for each Bureau Economic Analysis region. Three components comprise earnings - wage and salary disbursements, other labor income, and proprietors’ income. A negative value is rare but can occur when a negative proprietor’s income outweighs wage and salary disbursements. In some cases, earnings may also be set to zero to prevent the disclosure of information that could be attributed to specific firms. The BEA level economic proxies used in the analysis were manufacturing earnings (400) and agricultural service earnings (110). Agricultural service earnings were used to predict destinations of grain and agricultural chemicals. Manufacturing earnings were used to predict terminating commodity flows for all other commodity groups. Economic proxies were chosen in the belief that industrial activity in certain sectors was associated with the demand for commodity movements. In some cases, it was necessary to use aggregated industry categories to avoid missing data in the more detailed industry categories. For example, it would be desirable to use electric generation services earnings as a proxy for coal movements. This would have been logical since the majority of coal moved on the river is used in utilities. For a number of BEA's, electric generation services earnings were not disclosed to protect confidential industry information. Thus the aggregate manufacturing earnings category was used as an economic proxy for all non-agricultural commodities. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 8 DATABASE CONSTRUCTION The original intent of this research was to run the model on the disaggregated dock level database. It was thought that since the database contained over 182,000 records, the large N would facilitate obtaining statistically significant results. At the dock level there is a lot of "noise" in the data. The flow of commodities shifts between docks and varies greatly by year. For instance, if a particular dock was being repaired, the volume of commodity moved might be zero for one year. When the facility was reopened, thousands of tons might be delivered there. Data was aggregated by BEA region, commodity group and year to smooth out these fluctuations. The percent of tonnage flowing through docks with rail access and the percent of tonnage flowing through public terminals was calculated. These percentages were used as the proxies for rail access and destination at public ports. Choosing the relevant time period to model was another important consideration. The data set contained commodity movements for the years 1970 to 1996. Again, it was initially thought that using a greater number of years would increase the amount of data available to fit the model, and thus increase the precision of the estimates. Initial regression results using the full time series were not significant. A number of features of the pre-1980's data made it unlikely to reveal linear relationships between the independent variables and the dependent variable (terminating tonnage). Oil price shocks, the decline of the integrated steel mills and the bubble in the farm economy all had rather drastic effects on the volume of commodities being shipped on the Ohio River. The decline of the integrated steel mills (which use coke as an input to the steel making process) shows up in the data as a rapid drop in coke shipments during the 1970's. Coal movements also display a similar pattern because a substantial portion of coal is converted into coke. Demand for these commodities recovers during the 1980's and is less variable thereafter. Shipments of petroleum products were also radically effected by the oil shocks. The volumes of these commodities moved by barge show highly variable behavior, sometimes changing by a factor of five within a few years. Changes in the price of oil radically shifted the patterns of supply and demand. It is interesting to note that the highly unstable nature of the economy during the 1970s and the difficulty which government planners had in predicting future economic activity based on current trends was one of the arguments for rate de-regulation. Data from 1983 onward was used to estimate the model since a visual inspection of the data suggested that the model variables where more likely to have a linear form in these years. FORM OF THE MODEL The model estimated used average revenue per ton kilometer, industry earnings, percent of tonnage moved through public terminals and percent of tonnage moved through terminals with rail access to predict the volume of barge freight terminating in a particular BEA region. A separate model was estimated for each of the seven different commodity groups. The form of the model, including error term, is as follows. Yi = C + b1X1 + b2X2+ b3X3 + b4X4 + µi Yi = Terminating Tonnage The dependent variable in the model is total tons of a particular commodity, transported by barge and delivered to a BEA region. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 9 X1 = Revenue per ton-kilometer This variable controls for the impact of productivity increases in the railroad industry. One would expect that there would be a positive relationship between rail revenue per ton and barge shipments. As the cost of rail transportation falls, shippers will transport fewer goods by barge. X2 = Industry Earnings This variable controls for the impact of the economy on the demand for freight transportation services. One would expect that industry earnings would be positively associated with barge freight shipments. As industry earnings increase in a particular BEA region, barge shipments of freight also increase because they are an input to this economic activity. X3 = Percent of Tonnage Delivered to Public Terminals One would expect this variable to have a negative relationship to the tonnage of barge shipments. Public terminals are the places where freight will most likely be transferred to another mode. Thus, public terminals are the places at which railroads may employ differential pricing to capture freight. If a higher percentage of tonnage flows through public ports, we expect this to reduce the overall volume of barge tonnage terminating in a BEA region. X4 = Percent of Tonnage Delivered to Terminals With Access to Rail One might expect that rail access would have an inverse relationship to the volume of barge shipments over time. If a terminal has rail access, railroads may differentially price their services to capture freight traffic volume. MODEL ESTIMATION Tests where run on the data to check for multicollinearity and heteroskedasticity. To test for multicollinearity, auxiliary regression where used to test for linear relationships between the dependent variables. Multicollinearity was not found, indeed as the results will show, the amount of noise in the data made it difficult to identify strong relationships between any of the variables. Heteroskedasticity was identified as a problem. By graphing the predicted values of the model against the residual error terms, it was clear that the variance of the error term was not constant. As the predicted volume of tonnage increased, the size of the error terms also increased. White-Huber heteroskedastic consistent standard errors where used to correct for heteroskedasticity. PRESENTATION OF RESULTS The model results are shown in Table 1. While the F statistic on all of the models was significant, not all the coefficients were significant. The sign of the public terminal coefficient was the one hypothesized for every model. It was significant on all but one of the models. Each of the models are discussed separately below: Agricultural Chemicals The model for agricultural chemicals was the best among the models estimated. The R2 was .21, the highest among all of the commodity models. All of the coefficients are significant. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 10 Coal The model for coal has significant coefficients for average revenue per ton-kilometer and public terminal percentage. The percentage of variation in the data explained by the model (R2) is only 8%. The sign on the coefficient for average revenue per ton kilometer is not the one hypothesized. Coke The model for coke shows significant coefficients for only the public terminal variable. The earnings coefficient is close to significance at the 95% level, but its sign is not the one hypothesized. This may be due to the fact that the decline of the integrated steel producers, the main users of coking coal, reduced demand for coke while total manufacturing earnings were still growing. Grain The sign on the earnings variable is not the one hypothesized. Variability in the market for grain may explain this. Both the public terminal and rail access coefficients are significant. Industrial Chemicals Both earnings and the rail access coefficients are close to significance, although their signs are not the ones hypothesized. The public terminal variable is significant and shows the hypothesized relationship. Petroleum Products With an R2 of .03, the results are not very meaningful for this model. All Other Commodities All of the coefficients except earnings are significant. This makes some sense as the commodity group contains a number of different commodities, some of which may not be directly related to manufacturing earnings. DISCUSSION OF RESULTS The public terminal variable is significant and has the right sign for all of the models. This is consistent with the hypothesis that railroads are engaging in pricing practices to divert intermodal freight onto the rail network. The relatively low R2 on all the models shows that other explanatory variables or better data should be considered for use in future research. Previous research suggests that agricultural chemicals are the commodity for which differential pricing has the greatest impact. (USACE, Rock Island District) This is consistent with the results of this research, which show the agricultural chemical model to be the strongest. The impact of rail rates on coal and grain shipments have also previously been found to be substantial. The coefficients for the grain and coal models are significant, although the R2 on these models are not particularly strong. The significance of the public terminal variable is notable in the context of the findings of a number of other studies. In "The National and Regional Economic Benefits of Commercial Navigation on the Snake River" Daeger and Burton argue that a major benefit of investment in inland waterways is to compel lower rates from the railroads. The benefits of water-compelled rates are often used to justify investment in inland waterway facilities. Freight traffic diversion at public barge terminals may not be a sign of a malfunction of the competitive market place, but rather, an outcome that occurs by design. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 11 Other researchers have argued that barge transportation has social benefits. Rail pricing policies, which use market power to divert freight from the inland waterway system, may be imposing external environmental costs on society. A study by the Tennessee Valley Authority found that barge transportation is in general environmentally cleaner per ton-kilometer, having a marginal pollution abatement benefit to the nation of approximately .003 cents per ton-kilometer move. (Bray, Burton, Dager, Henry, Koroa; 1998) The research findings of this study provide one measurement of how substantial the economic and environmental impact of differential pricing may be. CONCLUSIONS The measurement of the impact of rail pricing policies on barge transportation is important for a number of reasons. Differential pricing may impose external costs on society that need to be factored into the current debate concerning open access and competition in the rail industry. A number of studies have examined the impact of the inland waterways on rail rates, but little research has been conducted to examine this issue from the other side, calculating the economic impacts of differential pricing on inland waterway traffic. The results of this research suggest that differential pricing has had a measurable impact on barge freight commodity movements. FUTURE RESEARCH There are a number of ways that future research might improve on the research design employed in this study. Regional rail pricing data from the Carload Waybill Sample could be employed to measure rail prices directly. Rail tonnage data from the waybill sample could be used to calculate a modal split for freight traffic by BEA region. Modeling mode share as the dependent variable instead of the volume of tonnage would most likely improve the model. A measure of the concentration of the rail industry in the study region might also be employed. Since differential pricing is likely to exert more force in the presence of industry concentration, it might be instructive to model this using an industry concentration index. The dates of major mergers might also be used as dummy variables. Alternatively, the impact of specific railroads could be modeled on freight traffic. The specific railroad serving each terminal is available from the Navigation Data Center CD-ROM. An index of service quality might also be a useful addition to the model. Shippers choose transportation based on both the price and quality of service. Unfortunately, there currently is no widely accepted index of rail service quality. (GAO, April 1999). Similarly, barge rates might also be an important addition to the model. Transportation decisions are not driven by the cost of a particular mode, but by the relative cost of the available modes. Unfortunately, it is difficult to obtain barge rate data. Approximately 90% of barge freight moves under private, long-term contracts. The rates set competitively on the barge spot market are highly variable and may not be reflective of the rates paid by the industry as a whole. In summary, alternative models could be constructed. The results of such an analysis would be relevant to the ongoing debate about the proper role of competition in the railroad industry. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 12 ACKNOWLEDGEMENTS The author would like to thank the Army Corps of Engineers for funding a number of projects related to forecasting barge commodity movements at Jack Faucett Associates. Additionally, the author would like to thank his colleagues at Jack Faucett Associates and ICF Consulting for their encouragement and support of this research. TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 13 REFERENCES 1. Bray, Larry; Burton, Mark; Dager, Chris; Henry, Ron; Koroa, M. The undervalued Benefits of Water Transportation. Tennessee Valley Authority, September 1998 2. Cunningham, Paul; Jenkins, Robert. "Railing at Open Access". Regulation. Spring, 1997 3. Dager, Chris; Burton, Mark. "The National and Regional Economic Benefits of Commercial Navigation on the Snake River". Institute for Water Resources, June 30, 1998 4. Erickson, Thomas. "The over simplification Pendulum." Traffic World. May 10, 1999 5. Faucett, Jack. "Open Access Could Work." Traffic World. April 12, 1999 6. Haulk, Jake. Inland Waterways as a Vital National Infrastructure: Refuting Corporate Welfare Attacks. Allegheny Institute for Public Policy, 1998 7. Kahn, Fritz. "Myths About Competitive Access". Traffic World, March 15, 1999 8. Kleit, Andrew. "The Unclogged Bottleneck: Why Competitive Access Should Not Be an Antitrust Concern". Transportation Review. Vol. 26, No. 3 9. Rail Rates and the Availability of Water Transportation: The Upper Mississippi Basin, United States Army Corps of Engineers, Rock Island District, June 1996 10. Rail Road Regulation: Changes in Railroad Rates and Service Quality since 1990. General Accounting Office, GAO/RCED-99-93, April 1999 11. Rail Road Regulation: Current Issues Associated with the Rate Relief Process. General Accounting Office, GAO/RCED-99-46, February 1999 12. U.S. Code Congressional and Administrative News, 96th Congress, Second Session 13. S.621 -- Heavy Regulation and Bad Economics. Association of American Railroads Web Page TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 14 LIST OF TABLES TABLE 1: Regression Results TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal. O’Rourke, Laurence 15 TABLE 1: Regression Results Coefficients Robust Standard Error t P Value 0.01 115,579 -199,705 124,232 -132,015 0.002 90,244 31,973 35,738 161,496 5.63 1.28 -6.25 3.48 -0.82 0.000 0.202 0.000 0.001 0.415 0.05 -8278656 -6703023 828357 21800000 0.120 3,559,150 1,795,373 116,725 637,933 0.41 -2.33 -3.73 0.71 3.42 0.679 0.021 0.000 0.479 0.001 -0.01 314,917 -737,432 292,392 -87,706 0.007 253,837 116,335 162,513 392,642 -1.89 1.24 -6.34 1.80 -0.22 0.060 0.216 0.000 0.074 0.824 R = .14 Earnings (Agricultural Services) Average Revenue Per Ton Kilometer Public Terminal Percentage Rail Access Percentage Constant -0.94 -403,702 -208,875 313,599 798,882 0.529 368,036 66,030 86,013 635,430 -1.78 -1.10 -3.16 3.65 1.26 0.079 0.275 0.002 0.000 0.212 R2 = .17 Earnings (Manufacturing) Average Revenue Per Ton Kilometer Public Terminal Percentage Rail Access Percentage Constant -0.01 314,998 -737,432 292,392 -87,706 0.007 253,838 116,336 162,514 392,642 -1.89 1.24 -6.34 1.80 -0.22 0.060 0.216 0.000 0.074 0.824 -0.03 -491,437 -542,498 -39,728 1,810,598 0.015 442,853 770,573 185,636 753,235 -1.73 -1.11 -0.70 -0.21 2.40 0.086 0.268 0.482 0.831 0.017 0.01 -2,696,391 -2,643,059 1,683,064 5,988,693 0.022 773,424 541,342 451,960 1,429,606 0.36 -3.49 -4.88 3.72 4.19 0.721 0.001 0.000 0.000 0.000 Commodity Agricultural Chemicals (N=198) 2 R = .2147 Earnings (Agricultural Services) Average Revenue Per Ton Kilometer Public Terminal Percentage Rail Access Percentage Constant Coal (N=184) 2 R = .0892 Earnings (Manufacturing) Average Revenue Per Ton Kilometer Public Terminal Percentage Rail Access Percentage Constant Coke (N=177) 2 R = .1703 Earnings (Manufacturing) Average Revenue Per Ton Kilometer Public Terminal Percentage Rail Access Percentage Constant Grain (N=99) 2 Industrial Chemicals (N=177) Petroleum Products (N=213) 2 R = .03 Earnings (Manufacturing) Average Revenue Per Ton Kilometer Public Terminal Percentage Rail Access Percentage Constant All Other Commodities (N=228) 2 R = .1213 Earnings (Manufacturing) Average Revenue Per Ton Kilometer Public Terminal Percentage Rail Access Percentage Constant TRB 2003 Annual Meeting CD-ROM Paper revised from original submittal.