1 | uuu i Z TA245.7 B373 8-1583 / NO.1583 l;_____i ,- l LIBRARY lUN26l988 I Iexas ‘AQMBjXEQRY -> \- Effect of Railroad Deregulation on w Export - Grain V Transportation / Rate Structures Neville P. The Texas A&M University System College Station, Texas Texas Agricultural Experiment Station Clarke, Director CONTENTS Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Marketing of Grain in Study Regions . . . . . . . . . . . . . . . . . . . . . . . 3 Conceptual and Statistical Model . . . . . . . . . . . . . . . . . . . . . . . . . 4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Additional Plains Study Region Results . . . . . . . . . . . . . . . . . . . ._: 11 Plausibility and Implications of Results . . . . . . . . . . . . . . . A . . . . . . 14 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 11109101151 pu2 11112d8 f101A2_1_ p112 ‘sn>121,\1 ‘1s11n3 ' fAp2d8 pu2 1sp11s21ps11c1) 1101;21n3s1sp 3111psss1d p011sd fwq; 31111np ss1n;on1;s s;2.1 111213 101 s1s2q sq; s2m 3111s11d jOIAJQS -_10-sn12A ;2q; 1101;011 sq; 1101;ssnb 0; 111sss s1sq;0 pu2 3111pu11 s111 _1_ 's01;2.1 s3121 Aq psz1.1s;o2.12qs s1sm su121d sq; 111 ss;21 ;2sqm s11qm s01;2.1 ;s00 s1q2112A-0;-snus/\ -s1 MOI A1sA1;21s1 ps;1q1qxs ss;21 ;1sg1 11.10[)-;s110d0110111 3u1;2u111111ss1p 2 Aq ss21d 111 ;nd sq 0; psLusss 3111101115 s;2.1 111213 1121 s11; 10 11011111 ‘p21 111 "s1n;1no1132 l a10A21 Asqod 331 ;2q; mspx s11; ps;n1s1 sQL61 A112s sq; u1110s1sN pu2 ‘110s1sp11V ‘110s11s108 Aq Apn;s V "s;s>1121u u01;2;10ds1121; s;0np01d psn12A-q81q s11; 111 s1s1112s 10;0111 111011 u01;1;sd1110s 3111s2s1su1 10 ssn2ssq s111sqss 3111s11d ss1A1ss-10-sn12A sq; ps110p112q2 Q31 sq; s1111;1sA0 ;2q; 1101; -011 sA1;2u1s;12 112 s1 s1sq_1_ -(>1Q1AAZ p112 ‘110s211s;8 ‘>10sc1 ‘1sAs1A1 f1sus0C1) (ss;21 113111 A1sA1;21s1 s10q spoofi ps1n; -o21nu2111 psn12A-1sq31q s11qm s1sAs1m01;2 ;ss s1sm u1213 s>111 ss1;1p01111110s A>11nq psn12A-m01 101 ss;21 ‘ s1) ‘A10s11; 311101111 ss1A1ss-10-sn12A sq; M01101 0; pspus; Q31 sq; ;211; ;q3n0q; s2m ;1 ‘A112s110;s111 '21s s1s832;8-s1d sq; 31111np 101A2qsq s‘10;21n8s1 sq; 3u1p128s1 su01;0u 311801111100 Aq ps;2o11d1110o s1 u01;21n3s1sp 10 s1110s;n0 sq; fiu1;s1ps1d 'su013s1 su121(1 sq; pu2 ;1sg u10Q sq; 111 ;us1s111p s;1nb sq A2111 u01;21n3s1sp10 ;ss_11s sq; ‘s101s1sq1 "s/xow A2111 u1213 qs1qm 1sA0 s10p1110s 1101; -2;10dsu21; 8u1;sdu10s pu2 s;s1;n0 ;s>112111 Au2111 s12 s1sq; pu2 ‘u01;1;sdu10s s312q sA1;ss11s 0; ss01s s1 u013s1 sq; 10 qsnw ';s>112u1 u01;2;10dsu21; sA1;1;sd1110s 9101112 n1 sq 0; suxsss ‘su121d sq; 0; ;s21;u0s u1 ‘;1sg u10Q sq 1 's1s1112o 121sAss Aq u018s1 A1ddns sq; 0; ps>1u11 s1sm ;2q; s;s>112111 fl\()91l21'8d9S A1121;2ds 110 pspusdsp 1101Ss1 sq; 11 ;s1xs p1n0m 101;1;sd1110s u01;2;10dsu21; sA1;ss11s s101,\1 ';s>112u1 s1q; ‘o; u01Ss1 sq; s>11111 ;2q; (10p11100 ;10d 11n3-su121d) 10p11 -100 u01;2;10dsu21; 10_f2u1 suo sq; pu2 ‘;s>112111 ;10dxs sq; ‘;s>112u1 s13u1s 2 u0 ssuspusdsp AA2sq 11sq; 10 ssn2ssq s1q21su111A sq A2111 su121d 121;11s[) pu2 q;n08 sq; 111 s1sd -d1qs ‘1sq;1n_1 '(00)1 ‘u1n1_:1) 10;1;sdu10s 2 s2 sssusA1;ss1 -1s s;1 s;1u111 s1n;sn1;s ;s0s SflIOILIQA 10;0u1 sq; ‘sssu2;s1p pspus;xs ;2 s12 s;s>112111 u1213 ‘su121d sq; 10 Au2u1 sou1s pu2 ‘s1sA11 s1q231A2u ma] 10 ssn20sq u01;1;sd1110s s312q sA1;ss11su1 s1 s1sq; ‘s111111 1121 8u1;sd1u0010 1squ1nu 112u1s 2 Aq psA1ss s1 u013s1 s1q; fs1du12xs u2 s2 ps;1o us;10 s1 su121(1 sq _1_ "sspom 3u1;sdu10o 21A d1qs 0; ssA1;2u1s;12 ps;1 -u111 sA2q ‘sss2s s1110s u1 ‘pu2 s1s1112s 1121 ms} Aq psA1ss s12 su013s1 8u1snp01d u121S-sn1d1ns smos ;2q; sn312 Asql ';sV s1s332;S sq; 10 s1101s1A01d A12u01;n2os1d sq; q;1m usAs ‘;us111u011Aus u01;2;10dsu21; sA1;1;sdu10o sq; 10 A;111q2>110m sq; u01;ssnb s;ss1s;u1 121n;1ns1132 su108 'psonps1 A112s1;211121p ussq s2q sn2s1nq s;2110 s101 sq; ‘snql '1n2q;2q;101 ss;21 ;ss A1;u10_f A2111 s1n2q s11111s;u1 u0 s01A1ss 3u1p1A01d u1 psA10Au1 s1s1112s A1u0 ‘;sV s1s332;S sq; 10 s32ss2d q;1M ‘ss;21 uo s;210q2110s o; s1s1112s ps; -;1u11sd A112o110;s1q ;2q; su01;21o0ss2 psu01;ou2s-;usu1u1s r ‘sspuouosg 10 ;usu1;12ds(1 ‘A;1s1sA1u11 s;2;8 2u11012Q q;101\1 ;2 A1;us.11ns ‘A;1s.1sA11111 W281; sexsl ‘10sss101d s;21s0ss2 A11su1101 pu2 f1101;2;8 ;us11111sdxf51 12.111;1ns113V s2xs_1_ ‘so1u10u -0s§1 121n;1n:)113V 10 ;11s111;.12ds(1 ‘10sss101d fVqs f1 ‘ss1A1s8 qs12sss11 01111011003 ‘;s11110u0ss f1101;2;8 ;usu111sdx§1 121n;1ns 113V s2xs1 ‘ss11110u0sg 12111;1ns1.13V10 ;11su1;.12ds(1 ‘10sss101(1* -A08 s12 sn2s1nq s;2\c1 'sn2s1nq s;21 q3n01q; ss;21 uo s;21 -0q2110s 0; s1s1112s10 1110pss11 sq; ssonps1 ;o2 MQU sq; 10 s1n;2s1 ;u2;10d1u1 A112nbs uV 'u01;2u111111os1p ss11d 12110s -1sd ‘3u1p123s1 ss;n;2;s 120110;s1q sq; 3111;2101A 10 1291 ;n0 -q;1m 1sdd1qs 121ns1;12d 2 0; ss;21 psonps1 1s110 0; p2011121 sq; s;11111sd ‘3u10p 0s u1 ‘p112 Asus1o111s p2011121 ssu2qus Apusnbssqns ;2q; ss1npsq0s pu2 sss1;s21d ;d0p2 0; ssA1; -ussu1s1sdd1qs 1s110 0; sp2011121 sm0112 3u1;s21;u0Q ‘ss;21 11112; M01sq ;2 so1A1ss ssp1A01d p2011121 sq; s11qm sum —[OA p112 sz1s ;us111d1qs 111n1u1u1111 sulos 0; 1sdd1qs u1213 s11; ;11111110:) A1121sus3 s;s21;u0[) 's;s21;u0s s;2A11d 0;u1 1s;11s 0; s1sdd1qs pu2 s1s1112s s;11111sd ;sV s1s3S2;8 sq; 10 80g 1101;ss8 ‘1sq;1n_q 'sssus;211d01dd2u1 s;1101001d 1s110 0; s;21 pss0d01d s‘p2011121 2 0; fiupssfqo suo 311111nbs1 Aq pu2 sA1;ss11s smuossq 0; ;us111;sn_fp2 s;21 pss0d01d s‘p201 -1121 121ns1;12d 2 101 ps11nbs1 s1u1; sq; ssnps1 ;2q; su01s -1A01d Aq ps321n0s11s ussq s2q S111o11d p2011121 ;uspusdsp -u1 ‘1101;1pp2 111 '(1s1ss)1) ;us111;ssAu1 110 u1n;s1 “s;2nbsp2” u2 3111u12s s1 p2011121 sq; 11 A1110 A1dd2 ;nq s01;21 ;s0s s1q2112A 0; snusAs1 ps111osds 113n01q; ps111;usp1 s12 ss;21 111n1111x2w 's;s00 s1q2112A m01sq 8u11121 111011 ss;21 ;111111 0; pu2 s1sdd1qs sA1;d2s10 ss1;1p01u11100 101 s1sAs1 A10d0u0u1 0; 3u1s11 111011 ss;21 ;usAs1d 0; psu31ssp s1 u01;21n3s1 ;us1 -1n1) "ss;21 Eu1u11111s;sp u1A;111q1xs11 s1q21sp1su0s s1s1112s sM0112 1101;21s13s1 9111 'ss;21 0; s;21s1 ss1n;2s1 121s1sA01; -u0s pu2 ssSu2q0 A10;21n3s1 sA1sus;xs ;s0111 s;1 10 s1110s ‘su01s1A01d Au2111 sspn1su1 ;sV s1s332;8 sq; qfinoqqV 'u1s;sAs 1121 s1q21A A1121su2u11 ‘;us1s111s s10111 2 0; ss11 sA13 111m so21d;s>112111 sq; uo so11211s1 s10111 pu2 u01;21nfis1 sss1 ;2q; ss111nss2 ;02 sql 'sp2011121 Aq sssnq2 121;us;0d Au2 u121;su00 0; s;2nbsp2 A1121s11s3 s1 u01;1;sdu10s ;2q; p1sq s1 ;1 '(1s1ss)1) sn1; 1sfiuo1 0u s1 s1q; ‘1101;21n3s1 Q91 Eu111nbs1 ‘A10d0u0u1 2 ps;n;1;su0s ssuo A1;snpu1 p2011121 sq; s11qm ;2q; ss1111s1d sq; u0 pss2q s1 ;sV s1s382;8 sq 1 ‘($31) u01ss11111110f) s01su1u10f) s;2;s1s;u1 sq; 8u1;2s10 101 s1q1su0dss1;s2 sq; ‘ L881 10 ;0V ss1su1u10Q s;2;s1s;u1 sq; ssu1s sp2011121 p12m0; A0110d 121sps1 u1 ss3u2q0 o1;2u121p ;s0u1 sq; 10 suo s;usss1ds1 (1851 10 ;sV112\c1 s1s382;8 sq 1 'ss1;1p0u111100 121n;1ns1132 >11nq 101 ;usu1u011Aus u01;2;10d -su21; sq; 10 sssusA1;1;sdu100 sq; uo ;0s11s s;1 pu2 (;sV11211 s1s332;8) A1;snp111 p2011121 sq; 10 u01;21n3s1sp ;usss1 sq; s1 ;ss1s;u1 ;2s131() 'A;1unu1u10s 121n;1ns1132 sq; u1 Au2u1 Aq u1sou0s q;1m psmspx ussq s2q 10;sss u01;2;10ds1121; s(1101;2u sq; 10 u01;21n3s1sp 121sps1 p12m0; pus1; sq_1_ 1101;snp01;u1 *;u211s§§1q0M ‘w pu2 ‘1s1sssg1 (1 ‘p12u0q021/q '[ ‘1s11ni1 '8 ss1n;0n1;8 s;21] u01;2;10ds1121_1_ 11121f)-;.10dx1q 110 u01;21n3s1s(1 p20111211 10 ;os_1_1g K If ICC policy did subsidize agriculture through 10w rail rates, then the Staggers Act may cause rail rates t0 rise sharply in those regions where intermodal competition is limited and deregulation does not facilitate interrailroad competition. However, if the ICC allowed railroads to act as a cartel, the pre-Staggers rail rates may be near the maximum allowed by intermodal competition. And, if the Staggers Act failed to generate interrailroad competition, hen deregulation may not substantially alter rates. Con- versely, if the Staggers Act succeeds in promoting inter- rail competition, rates may decline. Economists have contributed to an extensive body of literature that identifies the considerable inefficiencies and costs generated by ICC regulation, but limited efforts have been made to extrapolate the likely consequences of deregulation. Most noteworthy among efforts that at- tempt to evaluate the likely effect of deregulation are those by F riedlaender and Levin. Friedlaender notes that the effect of deregulation with regard to bulk agricultural commodities is difficult to predict but speculates that ef- fective rate competition among railroads is remote, even in the absence of rate bureaus. Furthermore, Friedlaen- der observes that under pre-Staggers regulation, rail- roads were able to exploit a considerable amount of their monopoly power on bulk agricultural commodities. In view of the ineffective rate competition expected by F riedlaender in the deregulated environment, and be- cause of the railroads’ ability to generate maximum rates on noncompetitive traffic before deregulation, Friedlaen- der concludes that railroads have little untapped monopoly power on noncompetitive agricultural traffic. Thus, rate deregulation most likely would not lead to sub- stantial or general rate increases in noncompetitive trans- portation environments. Levin simulated the deregulation outcome under vari- ous assumptions concerning elasticity of rail demand, de- gree of interrailroad competition, magnitude of rail cost reduction attainable with enhanced commercial freedom, and various levels of truck competition. Levin shows the social desirability of the deregulation outcome is closely related to the degree of interrailroad competition. Rail- roads’ rate of return and welfare losses vary only moder- ately with reductions in rail and truck costs, but a modest degree of interrailroad competition has a profound effect. Levin indicates grain transportation markets are probably most accurately represented by an intermediate degree of railroad competition; therefore, grain rates are unlikely to increase much in a deregulated environment. The theoretical expectations regarding the effect of de- regulation are uncertain, but empirical evidence is in- creasing. Unfortunately, there are several problems as- sociated with measuring the effect of deregulation on rate M. vels. Because much of the rail-transported grain moves under confidential contracts, the published tariffs provide incomplete information, (i.e., accurate rate information is not accessible). Further, coincident events make it dif- ficult to segregate the effect of deregulation from other ‘happenings. As an example, grain exports have declined nearly 5O percent from the peak years in 1980-1981, and presumably this has led to declines in the demand for transportation and a downward pressure on rates. Clearly, a method that controls for these effects is necessary to ac- curately measure the impact of deregulation. Despite problems associated with the empirical analysis of deregulation, useful evidence has been col- lected. Several Kansas studies attempted to measure and contrast rate trends during the pre- and post-Staggers era (Babcock et al.; Klindworth et al; and Sorenson). They re- port that published rail rates between 14 Kansas elevator sites and Gulf ports increased an average of 38.9 cents per bushel in the 4 years preceding the Staggers Act and de- clined 37 cents between 1981 and 1984. Price spreads were found to closely follow rail rates during the 8-year study period. Adam and Anderson investigated corn and soybean price spreads for a sample of Nebraska elevators from September 1978 through August 1984. They con- clude price spreads declined in the post-Staggers period by large and statistically significant amounts. The Kansas and Nebraska studies imply similar impacts attributable to deregulation. This study attempts to measure the impact of rail dere- gulation on export rates that link Plains and Corn Belt re- gions with their respective port areas in order to learn more about the effect of the deregulated environment on railroad price structures. An effort is made to evaluate the effect of deregulation on rates by analyzing the price spread between port and selected hinterland regions from 1976 through 1985. The decision to focus on geographic price spreads was prompted by knowledge that many export-grain rail rates in the post-Staggers era are deter- mined through private negotiations and are not public in- formation. In this case, direct measurement of rates to identify the effect of deregulation would be difficult or i1n- possible. By controlling for changes in export demand, local supply, and costs of transportation, storage, and marketing services, an attempt is made to isolate the ef- fect of deregulation on rates. It is assumed that the grain- handling industry is sufficiently competitive for geo- graphic price spreads to reflect changes in transportation rates that may result from railroad deregulation. This report includes five additional sections. First, at- tention is given to describing the selected study regions and their historic grain transportation patterns. This is fol- lowed by a section outlining the conceptual model that di- rects the analysis and a statistical model designed to iso- late the effect of deregulation on rates. The results section focuses on that portion of the statistical model that analyzes the effect of deregulation and elaborates on sec- ondary issues associated with deregulation. The study findings are subsequently discussed in view of the pre- Staggers notions about deregulation and the ICC’s appar- ent regulatory philosophy. Finally, conclusions and re- commendations are offered. Marketing of Grain in Study Regions The study area was made up of subregions comprising the entire states of Kansas, Iowa, and Indiana as well as the Texas Panhandle and portions of Illinois. Kansas and the Texas Panhandle subregions constitute one study area. These Plains subregions are surplus producers of hard red winter wheat, a wheat class that has historically comprised about half of the United States’ annual wheat production. Principal hard red winter wheat producers are Kansas, Oklahoma, Texas, Colorado, Nebraska and several northern Plains states. USDA data show about 6O percent of the United States’ hard red winter wheat production is typically exported. A 1977 study by Leath, Hill, and Fuller showed the Texas regions exported from 6O percent to 81 percent of their wheat shipments while Kansas regions exported slightly more than half of their shipments. Gulf ports receive more than 99 percent of the Kansas and Texas regions’ ex- port shipments with the Houston-Galveston-Beaumont complex being the principal export location. All study re- gions rely heavily on rail transportation—about 90 per- cent of the western Kansas wheat outflow was carried by rail, whereas, northeast Kansas and the Texas Panhandle shipped about 75 percent by rail, the lowest share shipped by any region. l The corn study region includes areas of Iowa, Illinois, and Indiana. These states have historically produced nearly 5O percent of the United States’ annual corn pro- duction with Illinois and Iowa each typically producing 1.0 billion to 1.7 billion bu., while Indiana’s annual pro- duction ranges between 0.4 and 0.7 billion bu. About 30 percent of the United States’ annual corn production typ- ically moves to export markets. A 1977 study shows the Corn Belt study regions ship to alternative export locations (Hill, Leath, and Fuller). All study region states shipped export grain to at least three of the four export coastal areas (Great Lakes, Atlantic, Gulf, Pacific); however, for Illinois and Iowa, the majority of export shipments (88 percent-90 percent) were to Gulf ports, principally Mississippi River ports. In contrast, Indiana shipped about 75 percent of its corn exports to Atlantic ports. In 1977, Iowa and Illinois shipped up to 90 percent of their Gulf shipments by barge while all of Indiana’s corn movement to Atlantic ports were transported by rail. Conceptual and Statistical Model The following conceptual model is used to direct the analyses. Price relationships in major transportation cor- ridors are assumed to be determined under the derived demand framework depicted in Figure 1. In this analyti- cal framework, the grain price spread between a U. S. port and some hinterland producing region is determined by the interaction of the export grain demand curve at the port (Dg), the far1n—level supply function (Si), the demand for export marketing services derived from these two schedules (Diii), and the supply of marketing services on the corridor that links the hinterland and the port (Siii). The price spread (m) is the price difference, Pg-Pi, where Pg is the grain price at a port (e. g., a Gulf port, and Pi is the price at an interior location, such as an Illinois produc- ing region. Other things being equal, an increase in ex- port demand (D ), and farm-level supply (Si) or a decline in the supply of marketing services (Diii) will tend to widen the price spread. Two regression models are estimated to measure the ef- fect of deregulation; one model attempts to capture the ef- fect of deregulation on price spreads (m) and rates in the Plains region (surplus wheat-producing regions) while the second model focuses on the Corn Belt. The Plains or Price Quantity Figure 1. Model of grain transportation corri- dors price spread. wheat model includes, as the dependent variable, the monthly time series data on price spreads (1976-1985) be- tween 12 hinterland regions in Texas and Kansas and Texas Gulf ports, while the corn model includes monthly price spreads between 18 hinterland regions in Iowa, Il- linois, and Indiana and Gulf and/or Atlantic coast ports. In general, the adopted procedure involves estimating regression equations with price spreads as the dependent variable. The independent variables include: (1) controls for shifts in the above-noted demand and supply schedule§w (Figure 1), (2) region and time dummies, and (3) a dummy and an interaction term to isolate changes in price spreads that may have resulted from the 1980 deregulation. Monthly price spread observations are generated for each hinterland region over a multiyear period, thus pool- ing both cross-section and time-series data. It seemed un- reasonable to assume that the ordinary least-squares esti- mates of the intercept and slope would be constant for all hinterland regions across all time periods. Thus, dummy variables that allow the intercept term to vary over time and over regions were introduced. The region dummies attempt to control for cross-region price spread deter1ni- nants not formally incorporated in the model. The study focuses on isolating the effect of deregulation through analysis of price spreads. To accomplish this, a de- regulation dummy was introduced. This binary variable estimates the adjustment in the regression equation’s in- tercept that is associated with deregulation. The deregula- tion dummy is also used interactively with the time trend variable to generate an additional variable whose estimated coefficient relates changes in the slope of the time trend variable that may have occurred because of deregulation?” The general model estimated to determine the effect of deregulation on geographic price spreads is as follows: (1) Pii = a + BiDii + BQSii + BfiMSii + region dummies + time dummies + deregulation dummy + deregula-. a tion dummy x trend time + Uii where, Pii is the monthly average price spread between hinterland regions and ports in dollars/bushel; Dii repre- sents export demand shifters; S“ represents farm-level l supply shifters; MS“ represents marketing service supply hifters; and U“ are the unexplained residuals. Subscripts i and t refer t0 regions and time periods, respectively. The dependent variable is the monthly price spread (dollars/bushels) between each hinterland study region and its associated port area for the 120-month period (1976- 1985). Figure 2 shows this time-series profile for the est-central Kansas study region. In general, all of the wheat regions’ price spreads tend to widen through 1980 and then begin to narrow. Several types of variables were identified as potential shifters of export demand, farm-level grain supply, and marketing services supply. National export levels and in- ternational grain prices were considered as potential mea- sures of export demand. However, because of the lag be- tween the grain sale and its subsequent outflow, the quan- tity exported per month was predetermined with respect to the current months price. Thus, quantity, rather than price, was assumed to be the outside factor. Furthermore, including the international grain price (e.g., Rotterdam price) as a demand shifter may have created a spurious re- 1.5 - $/bushel Q 1. llllllgjAlllln -l44_;;l4‘n- 11111.41... - 0.5 llxnl4ll4 0.8 -1;‘l-- 0.2 - 0.1 — 11 gression problem (i.e., the error term may not have been independent of the explanatory variable). For these reasons, monthly national exports was selected as the ap- propriate proxy for export demand (Figure 3) (USDA, l976-1985a). It was assumed that port area demand was closely related to the national export level. Several variables were tentatively selected as a proxy for the farm-level supply function. An effort was made to col- lect annual grain supply for each hinterland study region, but for several states, the production and price data were not available for similar geographic units. It was assumed that the quantity of produced grain was predetermined, so, as an alternative, state-level and national grain supply data were the selected variables (USDA, l976-1985b; USDA, 1976-19850). During the study period, there were substantial changes in factor prices, and these changes may have affected spreads between hinterland farm-level prices and port area export prices. For purposes of this analysis, real costs of holding grain are represented by the difference be- tween the nominal interest rate (Figure 4) and the percent change in the wholesale price index (International Mone- n‘ n F 11111111111 1111111111 222222222233333333334¢4414HRHQ55S5S555SS6666BB6S567777777777BB8BB88BBB99999QQQQQDUOOOUUOUU1gégfigégggg ‘ 11 111111 o123115618901231156109012auss7a9n12a11ss1e9o12s11ss7aso12311551119012auss1a9o12a11ss1a9o1 23115511190123115510901 23115611390 JRN‘75 - DEC'B5 (months) Figure 2. Monthly west-central Kansas wheat price spread, 1976-1985} age Houston export wheat price. H lPrice spread is estimated by subtracting the monthly average west-central Kansas farm-level wheat price from the monthly aver- 100000: l i 900001 i 800001 vnonoé snounf *1 1 50000: 4 4 uonnof q ‘l 4 (1000 bushels) souuof '4 3 annou- 100005 I4 O LL11... . 11 IIlllllllllllllIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIllllIIIIIIIIIIIIIIIIIIIII 1 1111111122222222223333333333111111111-111111111115555555555868586586677777777778888888868999999999900000000001111111111222222 0123115878901231156789012311587890123115878901231158789012311587890123115878901231158789012311587890123115578901231155789012311587890123115 IIIIIIllllllllIIllllIIIIIIIIIIIIllllllllllll'llllll 1111111111111111111111111 JRN‘76 - 0EC'8S Figure 3. Monthly U.S. hard red winter wheat exports, 1976-1985. tary Fund). It was thought that a simultaneity problem would arise if the percent change in grain value were sub- stituted for the wholesale price index. Because changes in real wage rates may alter the price spread, an index of wages was included (Figure 5) (Inter- national Monetary Fund). In addition, an index of fuel prices was included as a potential shifter since these prices changed dramatically during the study period (Fig- ure 6) (Association of American Railroads, 1979-1986). Any real increase in wages or fuel prices would have shifted the farm-level and the marketing services supply function. In the case of the farm-level supply function, an increase in input values will tend to narrow the price spread; in contrast, increasing input prices will tend to shift the marketing services supply function to the left and increase the price spread. In this case, increasing input prices may create off-setting effects on the price spread and make it impossible to predict signs. However, for pur- poses of this study, this outcome does not represent a problem since interest is only in controlling for such ef- fects. To measure the potential effect of changing rail costs on rates and the subsequent impact on price spreads, two rail cost indices were selected. Both indices were gener- ated by the Association of American Railroads (Associa- tion of American Railroads, 1979-1986). One index mea- sures materials prices and wages, whereas the other in- cludes fuel, material prices, and wages (Figure 7). A rail utilization index was constructed from the AAR’s Rail Grain Traffic by Quarter and Rail Grain Fleet Data (Association of American Railroads, 1987). The numerator of the index includes quantity of grain rail-transported per unit of time, and the denominator reflects the railroads car supply or its grain-carrying capacity (Figure 8). Dur- ing a 10-year study period, the index exhibited an intra- year cycle except in 1981-82. The low and relatively con- stant rail utilization index in 1981-82 was the result of an increase in fleet capacity that peaked in January 1982 and declining rail grain traffic. Based on the AAR data, the one-time rail carrying capacity peaked at 845 million bu. in January 1982, while the 1981 and 1982 rail-carried graiifl volume declined 12 percent and 16 percent relative to 1980. A dramatic reduction in 40-ft boxcars during 1982, coupled with modest increases in grain traffic in 1983, reestablished the historical movement in the rail car utili- zation index. ~ To reflect a widening price spread that may be attribut able to shortages in hinterland storage capacity, a ratio variable that measures unused storage capacity relative to 1B — ll lLnlxin- 13 ‘l. 12 — 11-: U I Interest Rate % BAALIILIII 1 IllIIllIllIIIIIIIIIIIIlllllllllllllllllllllll|lllllllllllllllllllllll|lllllllllllIllllllllllllllllllllllllllllllllllllllllllll 1 llllllllllllllllllllll 11 1 I22222222223333333333444K4414405555555555656666666B77777777776888888BBBQQQQQQQQQBUUOUOUDDDD1111111111222222 11mm oizaussiasoxzaussvasoizaussveeoizaussvasoizaussvaeoizaussvaeuizsussvesoizsussvasuizaussvasoizaussvssoiaaussiasoizaussvasoizaus JRN'76 - caves (months) Figure 4. Monthly 3-month T-bill rates, 1976-1985. total storage capacity was constructed (USDA, 1976- 1985d). Because limited data was available, it was neces- sary t0 estimate the ratio on a statewide basis. The ratio, calculated using grain stock and storage capacity data, exhibits the expected intra-year pattern (Figure 9). Because the corn and wheat surplus study regions have different marketing and transportation characteristics, several of the proposed explanatory variables were unique to each model. Some study regions in the Corn Belt ship grain to both Atlantic and Gulf ports, in which case it was necessary to include price spreads based on both port areas. To account for potential differences in price spreads that were due to coastal port area destina- tion, a dummy variable was added to capture any “cross- coast” variation in price spread. In addition, export barge “ates on the Illinois and Mississippi River system were in- '\ cluded in the Corn Belt model (Illinois Department of Ag- riculture). The barge mode is the principal carrier of corn from Iowa and Illinois to Gulf ports and is a major com- petitor of the railroad industry (Figure 10). See Table 1 for definition of variables included in the analysis (Appendix A). Results Two regression models are estimated to measure the ef- fect of deregulation; one model attempts to capture the ef- fect of deregulation on price spreads and rates in the Plains region, while the second focuses on the Corn Belt. The Plains model includes, as the dependent variable, the monthly time series data on wheat price spreads (1976- 1985) between 12 hinterland regions in Texas and Kansas and Texas Gulf ports, while the Corn Belt model includes monthly corn price spreads between 18 hinterland re- gions in Iowa, Illinois, and Indiana and Gulf and/or Atlan- tic Coast ports. Parameter estimates for the Plains and Corn Belt covariance models are presented in Table 2.1 The signs lThe Durbin-Watson statistics showed the residuals of the Plains and Corn Belt models to be serially correlated; there- fore, both models were reestimated and corrected for au- tocorrelation. The first-order serial correlation coefficient for the Plains model had an estimated value of 0.466 with a standard error of 0.023. The Corn Belt models serial correla- tion coefficient had a value of 0.356 with a standard error of 0.019. The adjusted R-squared for the Plains and Corn Belt models were .60 and .66, respectively. Table 1. Definition of Variables Included in Regression Table 2. Estimated Linear Regression Coefficients for Equations the Plains and Corn Belt Covariance Modell .750 ,1. PSPD: Monthly hard red winter wheat or corn Plains model Corn Belt model w‘ price spread between port location and V _ b] C ff _ t t t_ C ff _ t t t_ hinterland Study region bu, = arla € 0e ICICH -ra 10 0C iclen ‘Ya l0 INTERCEPT "Q1182 "0829 "0.0007 "0009 EXPD: Monthly U.S. hard red winter wheat or EXPD 65037-07 6.067 16624-7 3.730; Com exports (1,099 bu) FSUP 0.00016 5.220 —3=;6s67-6 —0.46v& _ _ WAGE 0.0053 4.662 0.0026 3.622 FSUPZ Annual FBgIOH hard {Ed WIIItGF wheat OI‘ TLBILL 00001 0100 _0_0010 _1_048 corn production (1 million bu) EXPECT 0.0036 2.504 —0.001l —0.922 . __ RUI "0.0254 "L457 0.0219 1.462 WAGE: Monthly real wage rate index (1975 — STOR _0_0030 _0_550 __0.0740 _0‘040 100) TIME 0.0015 5,452 —0.0014 —6.7s0 a T-BILL: Monthly 3-month T-bill rate DINTER 02487 8-207 00065 0-271 _ DSLOPE "0.0056 "12535 "0.00013 "0361 EXPECT: Monthly percent change in wholesale BRAT .. _ 05096 16,333 price index COAST — —- "Q0149 "2732 " . . . . . . AN "0.0036 "0749 "0.0005 "0089 RUI: Monthly rail equipment utilization index {FEB _0_0154 _2‘505 00198 3740 STOR: Monthly ratio of unused storage capacity MAR “00007 ‘1012 0-0020 0-332 to total storage capacity in study region ‘fill; :g'gé?£ I222: :8'gi‘00 I322? TIME; Monthl time trend variable JUN —0.0361 —4.716 —0.0253 --4.023 y 0 0265 3 130 0 0276 3 807 DINTER: Variable t-o measure 0hanges in intercept _0:0004 _0_'015 :0:0200 _0:005 due t0 Pall deregulatlon (Dre-staggers = SEP —0.01s1 -2.725 —0.0i72 —2.013 0, post-Staggers = 1) 3C3, —8.g(2)1é —0.67g 0.8396 3.406 DSLOPE: Variable to measure change in TIME vari- TXOTRI _0'01§ _ 1238 ' ‘r514 6 able’s slope due to rail deregulation TXNOC 00387 3575 _ _ (DINTER >< TIME) TXSOL 0.0363 3.450 ~ - . . . KSNW 0.1172 11.506 —- — BRAT: Monthly ‘real barge rates lin,l<1ng mid- KSWC 000 45 0000 _ _ 0 Illinois River with Gulf ports locations KSSW 00682 6093 _ _ \ ($/bu, 1975 = 100) KSNC 0.0237 2.329 - — ‘W; COAST: Variable in Corn Belt model to measure _g'$gg _g'ggg : : effect of port location on price spread KSNE -008“; _7_918 _ _ (Gulf ports = 0, Atlantic ports = 1) KSEC -0.0264 -2.606 - - . IDNW — — 0.0051 0.519 IAN-NOV: Month dummies IDNC _ _ 0,0047 0480 TXTRI: Region dummy for Texas Triangle region IDNE — — "0-0627 -6.415 . . . IDWC — — 0.0250 2.560 TXNOC: Region dummy for Texas Canadian River IDC __ _ 00073 0745 YBgIOII IDEC - - —0.005s -0.591 . ' _ - _ IDSW — —— 0.0420 4.305 TXSOL. Region dummy for Texas Muleshoe Plain IDSC _ _ 00417 4.272 VIEW 169°" IDSE — - -0.0040 -0.405 KSNW-KSSE: Region dummies for Kansas regions. Last F1132’ " " 803g . . . . A _ _ _ _ Ugo letters identify geographic location IANE _ _ 0.0870 100015 ° r6510“ m State- IASW _ ~ 0.1101 12.664 IDNW-IDSE: Region dummies for Indiana regions. Egg " " g-gggg 1039(1) - - - _ IA - - . 4.11 Ifast ‘two letters identify geographic loca ILW _ _ 00206 3.000 0°“ m State- ILNC ~ - -0.0023 —0.353 IANW-IASE: Region dummies for Iowa regions. Last Adjusted 00017 00557 two letters identify geographic location R'5q"a"ed in State‘ N 1440 2520 Region dumlnies for Illinois regions. Last lEstimated equations corrected for serial correlation. two letters identify geographic location in state. ..““l..u.‘u-l nan -j nu 11.44....“ 160 111111: 150 f 110 € Wage Index 130 uni-ii“.- 120 é 110 € I 4 BU 1 B0 5 11 IIIIIIIIIIVIIIIIIIIIIIIlll'll'l'llllllllllIIIIIIIIIIIIIIIIIIII|I'|IIIIIIWIIIIIIIIIIIIIIIIPIIIIIIIIIIIIIll'llll'lllllllllll‘lllllllllq 1Illlllllllllllllllllllll I I12222222222333333333SHH444“HQEQSSSSSSSSSSSSSSBSBSBB7777777777688B888BBBQQQQQQQQQQDOOOODDOOOI111111111222222 mm oizaussnssoizalissvasoizaussvesoiasussvasoiaaussvasoizaussvasoizaussvssuizsussvaauizaussvesniasussvasoizaqssvasoizaussvasuizaus Jmms - nzcwss (IIDDthS) Figure 5. Monthly U.S. wage index, 1976-1985. and magnitudes of the estimated parameters appear plausible given the outlined theoretical framework? See Appendix A for the mean, minimum, and maximum values of variables that appear in regressions and the sim- ple correlation among variables. The effect of rail deregulation on rates is measured with the deregulation dummy (DINTER) and the interaction term (DSLOPE) that was created by interacting the time trend variable with DINTER. Both variables are highly -» significant in the Plains model but not significant at usual levels in the Corn Belt model. The outcome suggests that deregulation generated rate declines in the Plains but not in the Corn Belt. The Plains model rate trend during the "5. 2Various models that included non-linear forms and lagged variables were estimated. In general, the linear form proved best and lagged variables were not often significant. The ex- ception was the wheat model where one-month lag in de- mand (EXPD), rail utilization index (HUI), and storage (STOR) were significant. However, it was judged that the modest improvement did not warrant inclusion in the final model. pre-Staggers months (l-60) is calculated with the INTE R- CEPT coefficient ($—.ll82) and the TIME coefficient ($0015), which is multipled by month. The post-Staggers trend (months 61- 120) is estimated by aggregating the DINTER coefficient ($2488) with the IN TE RCE PT coef- ficient and the DSLOPE coefficient ($—.OO56) with the TIME coefficient. The estimated coefficients on the TIME, DINTER, and DSLOPE variables in the Plains model were generally unaffected by adding variables to the model or deleting them, suggesting the robust nature of the estimated coefficients. Figure 11 identifies the estimated rail rate trend in the Plains study region during the pre- and post-Staggers era. Real rates increased during the five-year period preced- ing deregulation ($ .0015) and then commenced a dramatic decline. Deregulation appears to have reduced the rail rate trend an average of $.OO41 per bushel ($.—OO56 + .0015) for each additional month into the deregulated period. During the analyzed post-Staggers period (five years), the rate trend declined an average of about 33 cents per bushel. This result closely parallels Sorenson’s observations regarding the impact of deregulation at selected Kansas locations. d Fuel Cost Index sol ll 012345678901 Figure 6. Monthly fuel cost index, 1976-1985. The Corn Belt model shows a contrasting trend pattern and, in addition, indicates barge transportation rates (BRAT) have an important impact on Corn Belt price spreads?’ The statistical insignificance (at usual levels) of the DINTER and DSLOPE variables show deregulation had little sustained effect on price spreads and trend in the Corn Belt. And in contrast to the Plains region find- ings, barge rates have an extremely important effect on price spreads; in particular, a dollar decline in barge rates reduces the corn price spread about 5O cents ($5096). Thus, it seems that any decline in Corn Belt price spreads during the post-Staggers era must have been, in large part, due to the decline in barge rates, not to a reduction in rail rates. Further analysis was carried out by including lagged rail and barge rates in the Corn Belt model to iden- tify whether barge rates tend to affect rail rates or vice versa. Statistical results show barge rates affect rail rates in the Corn Belt but not vice versa. This is expected since 3It was thought that simultaneity may exist between barge rates (BRAT) and export corn demand (EXPD). To test this notion, empirical tests for causality were carried out by in- troducing lags. The analysis showed exports to be a function of lagged barge rates, but barge rates did not seem to be re- lated to lagged exports. Due to the unidirectional nature of causality, a simultaneously determined model did not seem necessary. 1O llllllllllllllllllllIllllIIIFIIIIIIIIIIIIIIIllllllllllll'lllIllllllqlqllllllllllllllllllllllllIIIIllllllllllllllllllllll 111111111111111111 1 111111]122222222223333333333444“44444HSSSSSS555555658668567777777777BB8B8BBBBBQ9Q999999QUUDOUDDDUUI11111111 234567890123Q5B7B9U123QS57B9D12345578901234567890123QS57B90123Q557B9U123kS67BQO1234557890123Q5B7B9U12345S7B JRN‘76 - DEC'B5 (rronths) it is generally acknowledged that water transportation has a substantial cost advantage for movement of low-valued bulky commodities, and empirical evidence verifies that barge carriage is dominant in those grain-surplus regions with access to a navigable river. The export demand variable in both models had a highly significant (1 percent level) positive impact on price spreads. When estimated at the means, the esti- mated export demand elasticity for the Plains and Corn Belt model was .059 and .095, respectively. It follows that the magnitude of these elasticities is too small to account for the decline in price spreads that occurred during the 1981-85 period. Local commodity production or supply had the expected positive impact in the Plains model but was not significant (at usual levels) in the Corn Belt model. Other statistically significant continuous variables included in the models were the labor cost index (WAGE) and the grain storage variable (STOR). Both regions’ price spread appeared very sensitive to changing real labor__ costs (i.e. , the wage elasticity approximates unity in bot models). The rail cost index was excluded from the equa- tion because of its high correlation (.76) with the wage index. Price spreads in the Plains and Corn Belt transpor- tation corridors were relatively insensitive to the availabil- ity of storage. For both models, the estimated elasticity l‘ varies between - .11 and - .15. The interest rate (T-BILL) and rail utilization index (HUI) variables were not significant in either model, but . t5 ‘l i § 4.“... u‘, n». ,‘ .<-‘r‘ ~ "8 Rail Cost Index lllllllIIIIPIIIlllll'1'lllIllllllllllllllllllqlllllllllllllllllllllllllllllllllllllIllIllllIIIIIFPIIIIPIIIIIIIIIlllllllll Illllllllllilllllll 111111111122222222223333333333U4l4444Q4RSS5SSSSSSSBSBBSBBBBS7777777777688888888BQQQQQQQQQQUOOUUOUOUU111111111 O123l56789D1234SS7B90123Q587BQO123QSB7B9U1234557B9U123K5S?B§U123lSB7B9D1234567890123QS57BQD123K5S759012345S7B9C123QSB78 JfiN'75 - DEE‘B5 Figure 7. Monthly rail cost index, 1976-1985. the proxy for the expected change in grain price (EX- PECT) was significant in the Plains model. The measure for energy prices was highly correlated (.80) with T-BILL and accordingly was not included in the final model. Sea- sonality appeared to be a factor in explaining price spreads in the Plains and Corn Belt transportation corridors as did the region dummies. The region dummies account for cross-region variability that is unaccounted for by in- cluded variables. The notion of no difference between Gulf and East Coast corn price spreads is rejected because of the signficance of the COAST variable — the estimated coefficient (—$.O149) shows corn price spreads to be slightly less when based on East Coast prices. Additional Plains Study Region Results There was interest in learning whether the Staggers Rail Act had a similar effect across all Plains study regions and whether the iieffects of deregulation were uniform over time} The deregulation dummy (DINTER) and the 4The estimated models were not motivated by a desire to develop structural equations; accordingly, the estimated parameters should be viewed as concomitants rather than structural (Pratt and Schlaefer). The included variables are designed to purge the data of their effects so that the impact of deregulation could be measured more appropriately. (months) 1=PRICES UF HRTEHIHLS QND HHGES 2=HRTERlRLS PRICES, HRGES RND FUEL interaction term (DSLOPE) measure the average effect of deregulation on the study region’s price spreads. Because there is interest in knowing whether the effect of deregu- lation was similar across all hinterland regions, two addi- tional types of interaction terms were added. These terms measure the differential effect of deregulation on each re- gion by allowing each regions slope and intercept values to shift relative to the average for all regions. The inter- cept shifters are generated by multiplying the region and the deregulation dummy variables, while the slope shif- ters are created by multiplying the region dummy with the deregulation dummy and the time trend variable. If these estimated coefficients are significant, the notion that deregulation had a differential impact on the various regions will be supported. The analysis showed deregula- tion to have an unequal effect among regions. In particu- lar, the price spread in northwest Kansas (KS NW) and the three Texas regions (TXTRI, TXNOC, TXSOL) showed an additional narrowing as a result of deregulation. This im- plies deregulation led to above-average rate declines in these regions. The notion that deregulation may have had an unequal effect over time was tested by augmenting the model (Table 2) with several variables. The DSLOPE and DINTER variables measuring the average effect of dere- gulation over the 1981-85 period, but by adding year slope 11 u Trips Per Month 0.0 IlllllIIIIIIIIIIIIlllllllllllIIIIl‘llllIIIllIllIIllIllllllIIIIIIIIIIIIIIIIIIIIIIUIllllIllll1|lIIIIIIlIIIIIIIIIIIIIIIIIIIIIIIII llllllllllllllllllllllll ll I111111111Z22222222233333333SSQKQQQKKEQQSSSSSSSSSSBSBBSBBBS677777777778BBB888BBB9999QQQQQQUUDOODOUOO1111111111222222 D1234567890123QSB7B90l23456789D1234557890123E5B7B9U123KS67B90l23HS57890l23\587B90]23Q5S7B90l23E567B90123QS67B901234S67B9D123Q5 JHN‘7S - DEE‘85 (Inonths ) Figure 8. Monthly trips per Grain Rail Car, 1976-1985. and intercept shifters, the hypotheses that deregulation had a differential effect over time may be tested. Introduc- tion of year dummy variables (1981-85) facilitates measur- ing of yearly changes in intercept values while introduced interaction terms (DSLUPE x year dummy variables) measure changes in slope. The results show deregulation did not have a uniform effect over time. In particular, in 1982, and to a lesser extent in 1983, there were increased narrowings in the geographic price spread. Supporting evidence regarding the temporal effects of deregulation was revealed by removing the DSLOPE and DINTER variables from the augmented model but allowing the year intercept and slope variables to remain. This model measures the effect of deregulation in each year during the 1981-85 period. The analysis shows most year vari- ables to be significant at the l percent level with the great- est narrowing in price spread to have been in 1982 and 1983, with lesser impacts in 1984, 1981, and 1985. Conceptually, some variables may have had a differing effect on rail rates and price spreads during the pre- and post-Staggers era, and through further analysis of these variables additional insight regarding the effects of dereg- ulation may be gained. As an example, railroads may have an opportunity to adjust labor input because of deregula- tion, in which case labor may have affected the price 12 spread differently during the pre- and post-Staggers period. In addition, there may be interest in knowing whether railroad rates have become increasingly sensitive to demand as a result of deregulation. To test notions re- garding the differential impact of selected variables on pre- and post-deregulation price spreads, interaction terms were created and added to the model. The deregulated time period (1981-present) has been characterized by weak export demand, and it has been ar- gued that a resurgence in this demand will lead to sub- stantial increases in rail rates. To test the notion that rail- road rates have become increasingly sensitive to demand during the post-Staggers era, attention was focused on the monthly export level variable (EXPD) and an interaction term (EXPD X DINTER). Both variables were included in a reestimated model and found to be statistically signif- icant. Based on these variables’ estimated coefflcientsfi» " elasticities that measure the responsiveness of price spread to monthly export levels were calculated. Results indicate an equally insensitive relationship between price spreads and export levels during the pre- and post-Staggers periods. The estimated pre-Staggers elasticity coefficient was .07 while the post-Staggers coefficient was estimated I f" A to be .06. This outcome offers no support to the notion that increasing export levels or demand would lead to dra- XIWOZ- D F ‘IIllllIIIIIIlllllllllllllllllllll 11 IIIIIIIIIlllllllllllllllllllllll I I I I llIIllIIIIIIIIIllIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIYI 11111111111111111111111111 11Z2222222223333333333HQK\QK44445S55S5S555666B65566677777777778888BBBBBBQQBQQQQQQ900000000001111111111222222 111111 01234567890123M567B9D123H557B90123\SB78901235557650123lS67890123MS67B90123156789012345576901234567890123Q557B90123HSS789012345 JflN'76 - DEC‘85 (months) Figure 9. Monthly index of unutilized grain storage capacity (Kansas). matic and substantial rail rate increases. Analysis was attempted with the labor cost, fuel cost, rail cost, and rail car utilization variables t0 determine whether these variables had a differing effect 0n the price spread during the pre- and post-Staggers era. This was ac- complished by introducing interaction terms. However, only the rail cost index and its associated interaction term were found statistically significant. At the means, the pre- and post-Staggers rail cost elasticities were estimated to be 1.31 and 0.2, respectively. This outcome shows a dra- matic decline in the sensitivity of rail rates to changing rail costs in the post-Staggers period. The pre-Staggers study period was characterized by rising costs. Historically, the ICC authorized universal rate increases that were cost- based, thus the extreme sensitivity of rates to costs before deregulation. The insensitivity of rates to costs during the “deregulated period was probably due to several factors. K In part, because of increased productivity during the de- regulated period, rate structures may have been less re- sponsive to changing costs, and increased interrailroad competition may have made railroads reluctant to pass on increased costs to shippers. Others argue that railroads have increasingly tended to base rates on demand-related factors rather than costs during the post-Staggers period, thus the insensitivity between rates and costs. This argu- ment, however, would seem to hold little credence in view of the earlier finding regarding the insensitivity be- tween rates and export levels or demand. There is evidence that railroad productivity may have increased during the post-Staggers era and subsequently helped foster rate reductions and reduced the respon- sivenss of rates to changing input costs. A stratified ran- dom sample of rail waybills drawn from the ICC Waybill Statistics show unit train use spread rapidly on the Plains region export routes in the early 1980s, especially be- tween 1982 and 1984. In particular, in the 1981-82 period, about 44 percent of Kansas’ export grain shipments were carried by unit trains; by 1984, the unit train share had increased to 85 percent (MacDonald). Klindworth et al. indicate the single-car rate system in Kansas was replaced in the early 1980s by a system of differential single-car, multiple-car, and unit-train rates. This has important im- plications since large shipment size lowers railroads’ per bushel rail costs. Typically, unit-train rates are 2O percent to 3O percent below corrresponding single-car rates, and confidential contract rates may be still lower. Therefore, the spread of multiple car and unit train movements may have placed downward pressure on rail rates in the Plains region. 13 $/bushel 1 l LlJjlllll J 0.1 E 0.0 ""\ II||lll|llllllll|llllll|Y||||||||||||||q|||||||'|||I||"||||l"|||||'||||||l|||||l||||||||||||||||||li|||||'|||||||Il|I||||||||| x 11lllllllllllllllllllllll W ‘gm I 111111111122222222223333333333l4§444!Q445SSSSSSSSSBBBSBSSBBS77777777776888888BBBQQQQQQQQQQOOOOUUUODU11111l111l222222 O123Q587B9D1234S57B9012345578901234557890123Q587BQD123QSE78901234S87890123HSS7B90123K5E7890123R567B9Di23Q567B901234557890]2315 JHN‘7B - DEC'B5 (months) Figure 10. Monthly Illinois River barge rates t0 Gulf, 1976-1985. The results provide strong evidence that the Staggers Rail Act of 1980 led t0 a restructuring of export grain rates. There appears to have been a substantial decline in export rail rates linking Central and South Plains wheat-produc- ing regions with Gulf ports whereas deregulation seems to have had little or no effect on the Corn Belt’s export rates. Furthermore, the declining rates do not seem in large part attributable to the diminished foreign demand for U.S. grain or shifts in the farm- level or marketing ser- vices supply functions. Plausibility and Implications of Results To some extent, these findings are at variance with ear- lier predictions. It was generally held that the transporta- tion environment in the Corn Belt would promote com- petition and lower rates, whereas most Plains transporta- tion corridors are dominated by a few rail carriers, hence an oligopolistic market structure that would dampen price (rate) competition. Although the results of this study were somewhat unex- pected, there is supporting evidence. The findings of Sorenson, et al. in Kansas and Adam, et al. in Nebraska support the Plains model results that deregulation led to a substantial decline in rates. Hauser collected published l4 rail rates on various export-corn transportation corridors over the 1978-83 period and found rates to modestly in- crease ($.02) and decrease ($02-$08) in relatively small increments after deregulation. Furthermore, collected tariff rates on both Plains and Corn Belt transportation cor- ridors generally support the models’ findings (Appendix B). Several provisions of the Staggers Act were designed to create a competitive railroad pricing behavior. First, the Staggers Act ended rate bureaus’ anti-trust immmunity and subsequently removed the railroads’ ability to jointly set rates. Some argue that rate bureaus permitted rail- roads to act as a cartel when setting rates. If rate bureaus did serve‘ as cartels in the pre-Staggers era and the ICC did not adhere to the value-of-service pricing theory, rates in those grain surplus regions with little intermodal competition (Plains) would tend to be relatively high, r whereas areas with strong intermodal competition (Corn Belt) would have comparatively low rail rates. Second, the Staggers Act attempted to further enhance the com- petitive transportation environment through provisions that facilitated the widespread use of contracting. Be- cause of the confidential nature of contracts, rate infor- mation does not become public. As a result, competing railroads are denied essential information for tacit price- setting, a form of price-making that may evolve in lieu of .10- Year 00 19119 19171 19119 19119 19190 19191 1992 19199 19194 1919s E U) ‘.10- 3 .0 A -.20- -.30- -40- Figure 11. Estimated rail rate trend during the Pre- and Post-Staggers Period. formal arrangements historically offered through rate bureaus. In view of the regulatory modifications offered by Stag- gers and notions regarding the ICC's pre-Staggers be- havior, the explanation for the observed results come into perspective. Apparently, rate bureaus were allowed to function as a cartel, and the ICC permitted railroads to offer rates whose upper bound was determined by com- petitive forces (i.e., the value-of—service pricing theory had no role in the pre-Staggers export-grain rate struc- ture). In this case, the lack of intermodal competition in the Plains led to high rates, whereas the strong inter- modal competition in the Corn Belt created relatively low rates. With the passage of Staggers, interrailroad compe- tition was apparently facilitated. And in the Plains region, which had little intermodal competition and the highest pre-Staggers rates, there was a substantial opportunity for rates to decline because of competitive pricing behavior. Accordingly, rates declined precipitously in the Plains compared with the Corn Belt, where rates declined mod- estly or not at all. Strong water competition in the Corn Belt had kept pre-Staggers rail rates low, limiting the op- portunity for rate reductions. Thus, deregulation had an uneven effect on the Plains and Corn Belt regions’ export- grain rate structures. Several implications of the results seem important. First, there is strong evidence that ICC rail regulation was not aimed at protecting agricultural shippers. Rather, evidence supports the notion that rail regulation served to allow cartel pricing (Hilton). Rates have fallen in the Plains region where a cartel would have been effective, but in the Corn Belt, where intermodal competition would have made for a less effective cartel, rates have de- clined little or not at all. Second, this study’s findings seem to support"Levin’s earlier work. Levin showed the social desirability of deregulation to be closely related to the existence of interrailroad competition. It seems that removing the immunity of rate bureaus and contracting have generated interrailroad and possibly geographic competition in those regions of the United States where railroads have historically enjoyed monopolistic power and relatively high rates. Summary and Conclusions This study evaluates the effect of deregulation on ex- port grain rates. Because of deregulation, much of the ex- port grain moves under contract rather than published tariff rates, making direct rate measurement impossible. Therefore, to evaluate the effect of deregulation, price spreads between port and associated hinterland regions were analyzed. This involved estimating covariance models that include price spreads as the dependent variable, inde- pendent variables that control for shifts in demand and sup- ply functions, region and time dummies, and a dummy and interaction term to isolate the effect of deregulation. Study results show deregulation led to a substantial de- cline in wheat export rates but had no or little impact on corn export rates. During the pre-Staggers era, rate bureaus facilitated the operation of rate-fixing cartels and, as a result, Plains region rates were generally high be- cause of limited intermodal competition. Apparently, pas- sage of the Staggers Rail Act removed rate bureaus’ cartel- like features, and, in spite of the few operating rail com- panies on most wheat transportation corridors, interrail- road competition developed. As a result of the historically high rates in the Plains region, and with the advent of in- terrailroad competition, rail rates declined. In contrast, there has been little decline in corn export rates because the regions railroads have historically experienced strong intermodal competition from low-cost water transporta- tion. Accordingly, there has been little opportunity for railroads to reduce Corn Belt rates, even though intra- modal competition has been made to exist. Because of the rather short post-Staggers’ period of ob- servation, it is difficult to know whether railroads will be- have in the long run as they have in the short run. Regard- less, it is essential that agriculture carefully evaluate pro- posed modifications to Staggers, especially those that deal with contracting and rate bureaus. Also of great impor- tance is the policy that the ICC adopts toward rail and rail- barge mergers. The large number of merger applications will undoubtedly produce a major restructuring of the U.S. railroad network. It is important that Congress and the ICC play a constructive role in preserving the ob- served interrailroad competition. Acknowledgments The authors are indebted to Hector Viscencio-Brambila and Ronald Rice for their substantial contribution to this research effort. This work was partially supported by USDA-ERS, Agreement No. 58-3523-4-00308. Review comments by Wesley Peterson, Don Farris, and Mickey Paggi are appreciated. Fuller and Bessler are professors in the Department of Agricultural Economics at Texas A&M University. Wohlgenant, former associate professor at Texas A&M University, is an associate professor in the Department of Economics at North Carolina State University. Mac- Donald is an economist with USDA-ERS. 15 10. 11. 12. 13. 14. 15. 16. 17. References Adam, Brian C., and Dale C. Anderson, “Implica- tions of the Staggers Rail Act of 1980 on Level and Variability of Country Elevator Price Bids.” Trans- portation Research Forum 26(1985): 357-363. Association of American Railroads. The Grain Book, 1986. Washington, D.C.: Information and Public Af- fairs Department, 1987. Association of American Railroads. Railroads Cost Recovery Index, Washington, D.C.: Information and Public Affairs Department, various issues 1979-1986. . Babcock, Michael W, L. Orlo Sorenson, Ming H. Chow, and Keith Klindworth, “Impact of the Stag- gers Rail Act on Agriculture: A Kansas Study,” Trans- portation Research Forum 26(1985): 364-372. F riedlaender, Ann F. The Dilemma of Freight Trans- port Regulation. Washington, D.C.: The Brookings Institution, 1969. F riedlaender, Ann F. , and Richard H. Spady. Freight Transport Regulation: Equity, Efficiency, and Com- petition in the Rail and Trucking Industries. Cam- bridge, MA: The MIT Press, 1981. . Fruin, Jerry E. “Impacts on Agriculture of Deregulat- ing the Transportation System.” Amer. ]. Agr. Econ. 63(1981): 923-925. . Fuller, Stephen W., Larry Makus, and Merritt Taylor. “Effect of Railroad Deregulation on Export- Grain Rates.” N. Cent. ]. Agr. Econ. 5(1983): 51-63. . Hauser, Robert J. “Regional Measures of Competi- tion in the U. S. Inland Transport Industry for Export Grain, 1978-1983.” Unpublished final report to U.S. Department of Transportation, University of Illinois, 1986. Hill, Lowell D. , Mack N. Leath, and Stephen W. Ful- ler. Corn Movements in the United States. Urbana: Il- linois Agricultural Experiment Station, January 1977. Hilton, George W. “The Basic Behavior of Regulatory Commissions.” American Economic Review Papers and Proceedings 62(1972): 47-54. Illinois Department of Agriculture, Division of Mark- eting. “Transportation Situation.” Springfield, IL, various issues 1976-1985. International Monetary Fund. International Finan- cial Statistics. Washington, D.C .: Monthly Summary Statistics, various issues 1976-1985. Keeler, Theodore E. Railroads, Freight, and Public Policy. Washington, D.C . : The Brookings Institution, 1983. Klindworth, Keith A., L. Orlo Sorenson, Michael W. Babcock, and Ming H. Chow, Impacts of Rail Dere- gulation on the Marketing of Kansas Wheat. Washing- ton, D.C.: U.S. Department of Agriculture, Ofiice of Transportation, 1985. Koo, Won W. “Impacts on Agriculture of Deregulat- ing the Transportation System: Comment.” Amer. ]. Agr. Econ. 65(19s3)= 187- 189. Leath, Mack D., Lowell D. Hill, and Stephen W. Fuller. Wheat Movements in the United States. Ur- 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. bana: Illinois Agricultural Experiment Station, Bulle- tin 767, January 1981. Levin, Richard C. “Railroad Rates, Profitability, and Welfare Under Deregulation,” Bell]. Econ. 12(1981): 1-26. MacDonald, James. “Multicar Export Grain Ship- ments, 1982-1984.” Unpublished paper, USDA, 1986. Meyer, John R., Merton J. Peck, John Stenason, and Charles Zwick, The Economics of Competition in the Transportation Industries. Cambridge, MA: Harvard University Press, 1959. Posner, Richard A. “Taxation by Regulation.” Bell]. Econ. 2(1971)= 22-50. Pratt, J.W., and Robert Schlaefer, “On the Nature and Discovery of Structure.” ]. Amer. Stat. Assn. 790984). 9-28. Sorenson, L. Orlo. “Some Impacts of Rail Regulatory Changes on Grain Industries.” Amer. ]. of Agri. Econ. 66(l984): 645-650. Sorenson, L. Orlo, Dale C. Anderson, and David C. Nelson. Railroad Rate Discrimination: Applications to Great Plains Agriculture. Manhattan: Kansas Agri- cultural Experiment Station Publication No. 165, I973. Spann, Robert M., and Edward W. Erickson. “The Economics of Railroading: The Beginning of Carteli- zation and Regulation.” Bell]. Econ. 1(Autumn 1970): 227-244. U.S. Department of Agriculture, Agricultural Mark- eting Service. “Grain and Feed Market News. Washington, D.C., various issues 1976-1985a. U. S. Department of Agriculture, Economic Research Service. “Wheat Situation and Outlook.” Washington, D.C., various issues 1976-1985b. U.S. Department of Agriculture, Statistical Report- ing Service. “Agricultural Statistics." Washington, D.C., various issues 1976- 1985c. U.S. Department of Agriculture, Statistical Report- ing Service, Crop Reporting Board. “Grain Stocks." Washington, D.C., various issues 1976-1985d. \-\ I w. APPENDIX A Table Al. Mean, Minimum, and Maximum Values of Variables Included in the Plains and Corn Belt Regres- sion Equations Variable Mean Minimum Maximum -------------- -- Plains Model --------------- PSPD .51 .19 .95 FSUP 1101.96 829.90 1248.60 EXPD 48180.84 19129.00 118266.00 T-BILL 8.91 4.85 16.8 EXPECT .44 —8.28 6.02 WACEI 108.89 96.55 118.06 RUI 1.42 1.10 1.74 STOR .57 .82 .88 --------------- -- Corn Belt ----------------- PSPDl .28 — .06 .65 EXPD 162984.00 78776.00 256844.00 FSUP 1007.91 840.91 1781.25 STOR .56 .28 .88 BRAT‘ .15 .07 .40 lValues deflated with wholesale price index (1975 = 100). Table A2. Simple Correlation Coefficients for Variables Included in Plains Model Values in parenthesis represent probability that simple correlation coeflicient is zero. PSPD EXPD FSUP WAGE T-BILL EXPECT RUI sron PSPD 1.00 .091 -.422 -.440 .099 .264 .865 -.158 (.0005) (.0001) (.0001) (.0002) (.0001) (.0001) (.0001) EXPD 1.00 .212 .059 .492 -.088 .198 .025 (.0001) (.0246) (.0001) (.1480) (.0001) (.8526) FSUP 1.00 .166 .806 -.289 -.287 -.048 (.0001) (.0001) (.0001) (.0001) (.8018) 1 1.00 -.885 -.808 -.247 -.078 WAGE (.0001) (.0001) (.0142) (.0001) * 1.00 .092 — .012 - .040 TBILL (.0206) (.0001) (.1279) 1.00 .258 — .088 EXPECT (.0001) (.2726) 1.00 .202 HUI (.1900) sron 1.00 Table A3 . Simple Correlation Coefficients for Variables Included in Corn Belt Model PSPD EXPD FSUP WAGE T-BILL EXPECT RUI STOR BRAT PSPD 1.00 .223 .267 —.242 —.154 .250 .373 —.257 .531 (.0001) (.0001) (.0001) (.0001) ~ (.0001) (.0001) (.0001) (.0001) EXPD 1.00 .071 —.241 .443 .163 .320 <07]; .439 (.0004) (.0001) (.0001) (.0001) (.0001) (.0003) (.0001) FSUP 1.00 —.063 .051 .045 -.077 -.235 —.031 (.0014) (.0105) (.0244) (.0001) (.0001) (.0724) 1.00 —.335 —.303 — .243 .196 - .172 WAGE (.0001) (.0001) (.0001) (.0001) (.0001) 1.00 .092 —.012 —.023 .127 TBILL (.0001) (.5523) (.0001) (.0001) 1.00 .253 -.153 .376 EXPECT (.0001) (.0014) (.0001) 1.00 .125 .539 HUI (.0001) (.0001) 1.00 .003 STOR (D001) BRAT 1.00 Values in parenthesis represent probability that simple correlation coefficient is zero. 18 I-RIIDCC\U . . U. 8— L.4l......... APPENDIX B Table B1. Export Rate Linking Topeka, KS, with Houston, TX, 1976-1983 . . . . , l2 Zll 35 E8 8O 72 Bil 96 10B 2() I"'I'III.ICQ\U Table Bl2. Export Rate Linking Hutchinson, KS, with Houston, TX, 1976- 1983 1. u- 1 1. 2i 1. n1 1 4 I “ ‘- l“ n.0- 1 J 1 o.s-' 1 4 J I 0.\l 1 u. 2i 1 o. o1 I fi I I 1 I I ' I v I I ' I I o 12 2n as u so 12 an as 10a mum “i _ // FIWIIICC\D 10“ Table B3. Export Rate Linking Salina, KS, with Beaumont, TX, 1976- 1983 MONTH 108 21 22 FIWZD§C\Q Q o O LA)‘ . . All Table Bn4. Export Rate Linking Wichita, KS, with Galveston, TX, 1976- 1983 on‘. 1 0.2-1 l 0.0-‘ I I ‘fi I I I * I I I I ' I I 0 12 as u a0 n an as 10a HUNTH Table B5. Export Rate Linking Toledo, OH, with Baltimore, MD, 1976- 1983 , Y i . i E8 - Q a 1| <1 n < n < n <4-1 2 I I <4 < 1 < < <<-¢4<<<4 c D l‘; D I C .l.‘.;.,;‘4 O.$ O I N ..-‘-.;l D I D IAAAAlALAAI Table B7. Export Rate Linking Esterville, IA, with Houston, TX, 1976-1983 HUNTH I 108 25 26 FIHIIDCC\M Li} LO 0.5 Table B8. Export Rate Linking Indianapolis, IN, with Baltimore, MD, 1976-1983 I 108 FINICDCC\U n: n N lap.- no n Q LLJALlALLLLJLLA Table B9. Export Rate Linking Champaign, IL, with Philadelphia, PA, 1976- 1983 MONTH nan: r~ha LO $7 ti “U: $1 Table B10. Export Rates Linking Columbus, OH, with Philadelphia,PA, 1976- 1983 MNW n IM [Blank Page in Original Bulletin] ? i-vfb! Q, Mention of a trademark or a proprietary product does not constitute a guarantee or a warranty of the product by The Texas Agricultural Experiment Station and does not imply its approval to the exclusion of other products that also may be suitable. All programs and information of The Texas Agricultural Experiment Station are available to everyone without regard to race, color, religion, sex, age, handicap, or national origin. 2M—6-88