DECISION FACTORS UNDERLYING TRANSPORT MODE CHOICE IN EUROPEAN FREIGHT TRANSPORT

Transcription

1 DECISION FACTORS UNDERLYING TRANSPORT MODE CHOICE IN EUROPEAN FREIGHT TRANSPORT Berit Grue and Johanna Ludvigsen Institute of Transport Economics, Oslo, Norway 1 INTRODUCTION This paper summarizes the outcomes of research performed under REORIENT project, which the European Commission awarded to a consortium of seven European and American research institutions under the 6 th Research Framework. The overall objective of the project was 1) to identify and develop new business concepts for trans-european rail freight transport that will make rail conveyance more competitive that truck, and 2) to assess the extent to which the EC rail liberalization legislation contributed to transfer of freight from roads to rail. This paper presents results from three research tasks which accomplished the following: 1. Identified the structure of supply and demand for rail and truck freight supply solutions in twelve European countries 2. Explored the contents of market demand from small, medium and large European shippers in manufacturing, merchandising and logistics provision industries, and 3. Assessed important predictors of shippers rail and road transport choices based on shippers satisfaction with important service qualities and the types of cargo shipped The interest of European Commission in freight transport solutions used by European shippers, and factors that may explain their mode selection decisions derives from two political concerns. The first is the need to stop continuously growing scope of socio-environmental negative externalities associated with wide-spread use of motor carriers for international freight movement. The other arises from the need for empirically verified knowledge of factors detrimental to use of environmentally friendlier freight carriage such as rail-based intermodal and single-modal transport for transfer of goods in European corridors. The latter would also provide the European Commission with meaningful feedback on important impediments hindering utilization of new business opportunities created by its rails liberalization policy. In terms of policy utility, this research sought hard facts on the structure of supply and demand for freight carriage that the European Commission

2 may use for making infrastructure decisions and/or design policy instruments that will enhance market attractiveness of rail freight dispatch. In analytical terms, the objective was to produce new knowledge on important predictors that affect shippers modal choices in countries along the REORIENT intermodal freight supply system. The REORIENT freight supply corridor connects the Nordic countries with Central and South eastern European states which became new members of European Union. The paper represents the first attempt to collect data on Central and Southeastern shippers who use, rail and road for freight conveyance and systematically analyze the structure of service supply and demand in these two European regions. 2 DEFINITION OF INTERMODALITY In order to asses the patterns of freight transit solutions that currently are supplied and used within and between the twelve European countries surveyed by WP 6.1 study, a precise definition of single-modal and intermodal freight carriage was needed. Whereas the concept of singlemodality is technically straightforward and pertains to door-to-door freight supply effectuated by one transportation mode and maximally two operators (e.g., truck and a Ro-Ro ferry), the notion of intermodality is more complex. For this reason a definition from the European Commission s Directive on Intermodality and Intermodal Freight Transport in the European Union [COM (97) 243/4] was used. There, intermodality was defined as a characteristic of a transport system that allows at least two different modes to be used in an integrated manner in door-to-door transport chain. This description actually allows that the transfer of unit load devices is interrupted by opening the load carrying units at intermediate points within an overall journey, for partial break-bulk at freight handling stations. 3 METHODOLOGY 3.1 Respondents selection Respondents in this survey included: 1. Large companies from manufacturing and/or merchandizing industries located in twelve European countries along the REORIENT corridor 2. Large logistics service providers in the same twelve countries who serve international freight flows of their clients For the need of first-hand-knowledge of quality delivered by single-modal and intermodal transport operators, a criterion was initially imposed that

3 respondents in this survey should concurrently and/or interchangeably use both forms of freight carriage. However, data collection revealed very early that intermodal freight services are not yet well developed in several European countries and that very few shippers could be identified which used both types of conveyance. Particularly, this prerequisite could not be fulfilled by shippers operating in Slovakia, Hungary, Poland, and Romania. 3.2 Measurement of demand construct Searches in multiple data bases disclosed the dearth of academic research on demand for freight transportation, modal choice decision criteria and the structure of freight supply solutions in Europe. In addition, no European studies were found which used the numerical and attitudinal data for recording revealed preferences for transport qualities harbored by the actual decision makers. Review of several engineering studies on transport mode choices confirmed that the discrepancy between the espoused and the revealed preferences was not only acknowledged, but analytically ignored. Therefore, the survey instrument was initially based on measurement and data collection technique devised by Evers, Harper and Needham (1996) in order to assess The Determinants of Shippers Perception of Modes, Transportation Journal Vol. 36, No.2, pp This instrument has later been adapted for use in the European context by J. Ludvigsen, whose study Freight Transport Supply and Demand Conditions in the Nordic Countries: Recent Evidence, Transportation Journal Vol.39. No.2, pp (1999) captured revealed preferences for transport quality of Nordic shippers. Subsequently, the instrument was amended in 2003 for survey trials in Norway, Hungary, Lithuania and Sweden under the Eureka PolCorridor project. 3.3 The sample Table 1: Firms in survey Country of Location Forwarders and/or Logistic Suppliers Manufacturers and/or Merchandisers Austria 3 27 Bulgaria 7 2 Czech Republic 7 16 Finland 4 22 Greece 7 3 Lithuania 27 0 Norway 9 8 Poland Romania 6 0 Hungary 5 5 Slovak Republic 6 23 Sweden 1 19 Total

4 The survey polled 246 business respondents; 140 were manufacturers and/or merchandisers performing import and export shipments while 106 were Logistics Service Providers (LSPs) who served international freight flows. Both groups of informants are referred to as shippers in this paper. Figure 1 shows the sample composition of large, medium and small firms. 21 % < 10 mill EUR mill EUR > 50 mill EUR 52 % 27 % Figure 1: Size of companies in the sample 3.4 Data collection Data were collected in 2005 by six research institutions and three business consultancies with good knowledge of transportation and logistics industries in twelve European countries. Data collectors identified the target companies and obtained consent from companies management for participation in survey. Subsequently, they queried the transport professionals (dispatchers) inside the companies surveyed who procured transport services and assigned freight shipments to migration solutions. These individuals provided the actual data. 3.5 Measuring instrument and variable definition The final questionnaire contained two thematic sections. The first gathered information on shippers business demographics. The second main part included decision scenarios where shippers assigned two regularly used consignments (exports and/or imports) to transport solution offered by operators of international corridor(s). By applying a decision simulation scenario, we sought hard facts on how the variation in service quality on freight routes would affect shippers choices of transport solutions. In contrast to stated preference method, which is often solely used to assess the transport users espoused quality preferences, this design revealed the actual preferences. The respondents specified duration of door-to-door transit, price for one-way freight carriage, frequency of shipments, and all types of modes used under consignment transfer. Next, informants evaluated twenty three quality dimensions of transport solutions used for shipments of a given cargo and then, the overall quality standards on each route. Service quality on routes selected was evaluated by two measures. The first was importance that

5 a given informant assigned to each of the twenty three quality-attributes before a consignment was expedited. The second appraised satisfaction that the informants attributed to the same twenty three quality dimensions after a consignment has been delivered. Our causal reasoning presupposed that quality requirements would affect the shippers choices of shipment routes, these routes modal configuration, and eventually, the use of road and rail transit. Majority of freight shipping routes in Europe are served in parallel by carriers who offer different levels of service quality and modal combinations. Besides, European freight migration lanes run through highly variable quality of national infrastructure stretches, which also affects the quality of services offered. So, we also hypothesized that priorities for different services qualities would also affect choices of geographical freight corridors. Since we assumed that the different quality requirements would affect the choices of modes and routes, these two types of decisions were defined as two of the three dependent variables. The second dependent variable was the geographical outlines of the freight migration lanes calculated by use of GIS. This information provided later a basis for estimates of freight movement speeds and probabilities of delays on different corridors. The third dependent variable was composed of dispatchers appraisals of the overall quality standards on the routes used measured by assessments of importance and satisfaction. By so doing, we implicitly assumed that service quality on European routes would constitute an independent variable in our analytical model. The aforementioned influenced the structure and the causal order of our analytical model, and the statistical methods for model testing. 3.6 Model specification and statistical methods Figure 2: Analytical model of relationship among variables affecting the shippers route and modal choice decisions

6 The model consisted of antecedent, independent and dependent variables. The dependent variables were divided into two main variables and one subsidiary. Shippers demographics, such as the company size together with the types freight/cargo served constituted the antecedent variables that affected the shippers service quality requirements. The independent variable, the quality construct was measured by twenty three performance dimensions, that together constituted an overall service quality standards dichotomized by shippers appraisals of quality importance and satisfaction. The overall importance and/or satisfaction with quality standard were also affected by antecedent variables such as the types of cargo handled, the types of load carrying units (LCUs) used, and the kinds of shipments forwarded. These two groups of variables affected the shippers choices of routes, and subsequently, selection of rail and/or road conveyance for freight transit. The two main dependent variables were: 1) the shippers assessment of an overall importance and satisfaction with service quality on the routes chosen, and 2) the shippers actual choices of road and rail lanes. The first main dependent variable was measured by the dispatchers evaluation of importance of and satisfaction with twenty three quality dimensions. Both importance and satisfaction were measured by five-category Likert scale. By assuming that service qualities desired before execution of a given shipment may differ from the qualities actually delivered, these measures identified two stages in dispatchers decision-making. Importance served as a tool for finding a suitable offering, while satisfaction provides foundation for inclusion/exclusion of a given transport carrier from a pool of operators eligible for future procurements. These two measures revealed the sequential pattern in dispatchers transport decision making where stated preferences for quality required preceded satisfaction with service delivered. Both, importance and satisfaction provided baseline for assessing the overall standards for required and delivered service qualities. The second main dependent variable, the shippers selection of shipment modes, was affected by the quality standards on the routes used. Shippers route choices thus preceded the choices of transport modes, and functioned as a subsidiary dependent variable, which was later fed into GIS software and presented graphically at European road and rail networks. Testing the above model involved a stepwise examination of relationships between the antecedent, the independent and the two main dependent variables calibrated at two levels of measurement. First, factor analysis was applied to extract the main underlying dimensions of the quality construct from the data material and reduce the variance in the data to few meaningful factors that explained a large

7 portion of total variance. Using the principal component method, several factors were extracted from data. The resulting factors represented a set of quality indicators which were highly correlated with the factor that represented them. The service indicators which loaded on a given factor provided a particular dimension of service quality. Next, the factors were regressed on the ratings of the overall importance of service quality that shippers assigned to the routes chosen. Regression analyses showed the importance of each factor to extend to which each factor was correlated with the overall rating of shippers importance. Subsequently, a logistic regression model was developed to assess the relative probability of distribution of shippers selection of rail and road routes. The model was used to test the sensitivity of rail-road probability distribution on improvements of service qualities which proved important in the preceding analyses. 4 FINDINGS 4.1 Supply of single-modal and intermodal transport in Europe Table 2: Intermodal and single-modal transport solutions on survey routes Single-modal Transfer Number of Shipments/Lanes Intermodal Transfer Number of Shipments/Lanes Truck 219 Truck + RoRo** 19 Truck + RoRo* 62 Truck + Rail 29 Rail 48 Truck + Ship 32 Rail + Rail Ferry 3 Rail + Ship 3 Truck+ RoRo+ Rail 4 Rail+Ship+Truck 6 Sum *Although on the face this form of freight dispatch involves two modes, it was defined as single-modal because it pertained to shipments involving Nordic countries and/or England. Geographical location in these countries requires that all trucks need to cross the sea en route to Continental Europe. Therefore, RoRo ferries are considered here as a part of road infrastructure for sea crossing. ** This freight dispatch category included un-accompanied load carrying units (LCUs) such as semi-trailers and/or swap bodies which although carried by truck to a Ro-Ro quay may optionally be carried by truck or rail after arrival at port of discharge. Table 2 shows the types of single-modal and intermodal freight supply that the study s respondents currently use for trans-european cargo transfer. The structures of transport chains show that single-modal solutions involve maximum two operators, while the intermodal ones two to three transport carriers between shipments origins and destinations. Common for both forms of freight transfer is that different operators need to function in highly synchronized manner in order to produce on-time and breakage-free freight delivery.

8 The collaboration between truck and rail operators involves transfer of containers, swap-bodies and semi-trailers on rail flatcars (TOFC) as well as bulk-breaking at intermodal terminals and/or freight service stations. Similarly, working relationships between railways and short-sea and/or deep-sea shipping lines consists in transfer of containers on flatcars (COFC), or supply of vessel-rail service combination for port-to-port freight transfer and/or between harbors and t inland destinations. Collaboration between rail-ship-truck operators involves sea-land bridging for terminalport and port-to-door rail freight supply. The amount of intermodal freight lanes captured by the survey confirms the dominance of single-modal freight carriage at European mainland. It also reveals high dependence of Nordic countries as well as Lithuania on intermodal solutions for inbound and outbound goods transfer. Figure 3 shows the types of load carrying units (LCUs) used by the survey respondents for dispatch of freight volumes. Load Carrying Units Used to Transfer Goods in Overland Transit (% of volumes shipped) 100 % 90 % 80 % 70 % 60 % 50 % 40 % 30 % 20 % 10 % 0 % 1 Truck 2 Rail Traditional Rail Wagon Tank Car/Wagon Standard Container Maritime Container Swapbody Semitrailer Reefer Other Figure 3: LCUs used by shippers surveyed The types of LCUs used reveal that a good deal of intermodal competition already exists between road and rail conveyance. Swap-bodies carry 78 p.c. of tonnage dispatched by both single-modal and intermodal truckbased freight supply. On the other hand, the amount of freight moved by traditional full rail wagonloads (FWL) amounted to just 39 p.c. of volumes carried by rail. This indicates that about 50 p.c. of freight carried by rail is

9 conveyed in intermodal transport units like 20 and 40-feet containers (1 and 2 TEU), swap-bodies and tank cars. A relatively high share of intermodal transport units carried by rail may also reflect the usage of trans-european freight supply where rail haulage is combined with freight consolidation and/or bulk-breaking operations performed by truck at both ends of logistics chains. Although market segment for this intermodal transit is small (only 4 p.c. of tonnage shipped by the shippers surveyed), we conjectured that it may quickly rise if the quality of rail service is improved. Inspection of LCUs used by motorized operators reveals that about 90 percent of tonnage that today is carried by truck could be forwarded in tanks, semi-trailers, swap-bodies and containers on rail flatcars. In the following we shall explore why the truck conveyance sill dominates trans- European transit of goods and what obstacles hinder rail from expansion into truck-dominated freight market. Table 3 displays variation in amounts of freight carried by different types of LCUs. Table 3: Freight volumes shipped by different types of LCU (Tons) Type of LCU Mean Std. Deviation Minimum Maximum Median Rail Wagon Tank Car Container Container Swap Body Semi-trailer Reefer The above display underlies a conclusion that the degree of utilization of different LCUs varies considerably. One reason could be carriage of voluminous goods with low unit weights in rail cars, semi-trailers and 40 feet containers which causes under-usage of load carrying capacity. Another could be inability to stow more freight in each freight box due to time pressure for shipments expedition. The account of commodities shipped by respondents and the prices paid for freight carried by rail, truck and truck-rail intermodal, are presented in this section.

10 Goods Categories on International Road and Rail Shipments (% of volumes shipped) 100 % 90 % 80 % 70 % 60 % 50 % 40 % 30 % 20 % 10 % 0 % 1 Truck 2 Rail 8 Manufactured goods 7 Machinery and transport equipment 5 Chemicals 2 Crude materials and fuels 0 Foodstuff and beverages Figure 4: Commodity categories shipped by in the survey Table 4: Commodity categories in shipments registered Commodity Categories in Figure no 4 Denomination in Terms of SITC Main Chapters Manufactured goods SITC6 + SITC8 Machinery and transport equipment SITC7 Chemicals SITC5 Crude materials and fuels SITC2 + SITC3 Foodstuff and beverages SITC0 + SITC1 Figure 4 reveals that rail and road compete in several markets for freight transfer. Competition interfaces, albeit of different size and intensity, could be identified in transfer of freight within four of the categories. Only machinery/transport equipment appears to be almost solely moved by truck. Manufactured goods dominate commodities forwarded. Since rail accounted for carriage of 80 p.c. of manufactured goods and truck for 48 p.c., this indicates that rail-road competitive interface exists in this market segment. This competition is dominated by rail. About nine percent of crude materials and fuels were carried by rail, but only for 3 p.c. by truck. Chemicals represented another commodity category where rail and road compete for the same type of shippers, with respectively 7 and 2 p.c. of volumes carried by each mode. Foodstuff and beverages represented another market segment with road-rail rivalry, which, however, was

11 dominated by truck carrying 17 p.c. of tonnage registered against 2 p.c. conveyed by rail. Machinery and transport equipment represented a large freight segment where about 23 p.c. of tonnage was moved by truck. Table 5 shows variation in Euro prices paid by shippers for unit ton of freight carried over one kilometer in rail cars and semi-trailers. Table 5: Mean prices for transfer of unit tons of commodity categories by truck and rail (Euro) Goods Categories (SITC1) Shipped by Rail and Truck Mean Price Euro for Transfer of Unit Ton by Truck Rail Food Stuffs Chemicals Semi-finished products Technical products Finished products Table 5 reveals that rail obtains much lower unit ton prices than truck for transfer of semi-finished, technical and finished products. This discrepancy may: 1. Reflect the service quality differentials between truck and rail in transfer markets for these three cargo categories, and 2. Be an effect of differences in rail and road positioning in specialty sub-markets created by these aggregate goods categories. The data also show that prices gained by rail for carriage of foodstuffs and chemicals match those moved by truck. Analyses of demographics of rail and road users shown in figure 5 reveal that it was the medium-size shippers who forwarded the largest volumes by rail and not the biggest ones. This finding has significant implications for rail and rail-intermodal operators: for access to stable and large goods repositories required for scale, scope and density economies please target medium-size manufacturing companies and LSPs because these business entities may represent prospective customers with positive experience from usage of freight rail. Therefore, they may harbour greater propensity for extended usage of rail, provided important service quality requirements are fulfilled.

12 70 % 60 % 59 % 50 % 42 % 40 % 30 % 30 % 29 % 37 % Truck Rail 20 % 10 % 0 % 4 % Small (< 10 mill EUR) Medium (10-50 mill EUR) Large (> 50 mill EUR) Figure.5: Road and rail shipments weighted by freight volumes shipped yearly 4.2 Quality requirements for transport by road and rail This chapter reports on how importance and satisfaction assigned by shippers to twenty three quality dimensions varied between the routes served by rail and truck operators. The shippers assessments are shown in the next figures. Importance Satisfaction Reliability of service Cost of door-to-door delivery Amount of loss and damage Service availability at origin point Service availability at destination point Quality of freight handling Duration of transit time from origin to destination Processing of loss and damage Frequency of service Flexibility for dealing with seasonal variations in no. of shipments Directness of shipment Equipment free on time for loading/unloading Information promptness on cargo under shipment and after arrival Suitability of Load Carrying Units for commodity carried Suitability of Load Carrying Units for shipment size Expediency of ordering/chartering service Availability of tracing/tracking services After delivery service Availability of Load Carrying Units (LCUs) Kindness of service staff Environmental friendliness Goods availability at destination, for return transport Efficiency at transloading stations Figure 6: Gaps between scores of importance and satisfaction assigned by shippers to with twenty three quality attributes of truck shipments.

13 Importance Satisfaction Service availability at origin point Cost of door-to-door delivery Amount of loss and damage Reliability of service Processing of loss and damage Service availability at destination point Frequency of service Suitability of Load Carrying Units for commodity carried Duration of transit time from origin to destination Availability of Load Carrying Units (LCUs) Information promptness on cargo under shipment and after arrival Expediency of ordering/chartering service Flexibility for dealing with seasonal variations in no. of shipments Equipment free on time for loading/unloading Kindness of service staff Suitability of Load Carrying Units for shipment size Directness of shipment Quality of freight handling Environmental friendliness Availability of tracing/tracking services After delivery service Efficiency at transloading stations Goods availability at destination, for return transport Figure 7: Gaps between scores of importance and satisfaction assigned by shippers to twenty three quality dimensions of rail shipments. The quality factors rank differently between truck and rail users, Shippers importance and satisfaction with quality requirements differed broader for rail than for truck. This shows that by and large participants in the study were less satisfied with services supplied by rail than by truck. Both negative and positive gaps exist between the service quality expected and delivered by truck and rail. The reliability of supply and the cost of service were the most significant quality requirements that truck users pointed out deviated from the quality expected, and thus displayed a negative quality gap. On the other hand, the shippers using rail were mostly dissatisfied with poor availability of rail service at shipments origins and (poor) value for money paid for freight transfer. Reliability of freight delivery, quality of processing of loss and damage, duration of goods transit and information promptness on cargo under shipment and after arrival, all scored low on shippers satisfaction with rail services supplied. Yet, rail operators scored better than road on environmental friendliness, and availability of LCUs suitable for shipment size and types of commodity carried. All in all, rail performance scored better on five qualities as compared to shippers expectations. For truck, the number was seven. Motorized operators excelled at kindness of service, service availability at shipments destinations for provision of back-hauls and at efficiency at trans-loading stations. The following maps exhibit the road and rail routes in Europe used by shippers surveyed for freight transfer and delivery.

14 Figure 8: Map of road routes used by shippers in the survey Figure 9 visualizes the currently used parts of rail corridors laid over the Trans-European Transport Network to which the European Commission, DG TREN assigns great socio-political value and which will receive considerable European and national investments over The picture shows that shippers surveyed use majority of TEN-T axes for railbased consignments. This indicates that shipments registered by our study represent typical freight consignments moved along the existing European transport pipelines. Figure 9: Map of rail routes overlaid the TEN-T intermodal priority axes

15 4.3 Determinants of shippers choices of rail based freight supply solutions Table 6: Service quality dimensions and loadings of importance assigned to qualities of rail-based freight supply Variable number Factor Description Loadings Factor 1: Dealing with Service Failures Imp_20 Amount of loss and damage.654 Imp_23 After delivery service.804 Imp-21 Kindness of service staff.709 Imp_10 Processing the loss and damage.632 Factor 2: Intermodal Expediency Imp_10 Suitability of Load Carrying Units for commodity carried.814 Imp-16 Duration of transit time.714 Imp_8 Efficiency at trans-loading stations.611 Factor 3: Efficiency of Cargo Intake& Discharge Imp_4 Expediency of ordering/chartering service.853 Imp_9 Equipment free time for loading/unloading.799 Four variables loaded on Factor 1 Dealing with Service Failures, which emphasized importance that shippers assigned to reduction of loss and damage, needs for staff performing kindly when rendering after-deliveryservice, and especially, when processing cargo loss and damage. Failure by rail operators to fulfil these service requirements enhances the scope of shippers financial loss when their cargo gets damaged and/or lost. The fact that this factor explained the largest amount of variance in data material related to rail shipment routes reveals that this service dimension is important, but not necessarily well fulfilled by rail carriers and logistics suppliers who deliver rail-based freight supply solutions. Three variables loaded on Factor 2, Intermodal Expediency, which shows that shippers emphasized importance of suitable load carrying devises for commodities carried, reliable transit time for cargo transfer and efficiently functioning trans-loading stations. Failure to meet these requirements leads to considerable transportation and non-transportation costs for consignors, consignees and LSP. LSP who take over cargo after rail haulage waist their time by waiting at trans-loading stations when real time of cargo arrival deviates from the schedules. Inefficient intermodal terminals cause higher inventory and capital costs for retailers and distributors who use more stocks to prevent stock-outs and loss of business. Besides, when cargo arrives later than scheduled by master productions plans and/or inventory replenishment timetables, manufacturers suffer from equipment down-time and production stops LCUs suitable for goods shipped hold transport costs down by maximally utilizing the load carrying capacity. Arrivals of LCUs unsuitable for a given

16 cargo category causes shipments of under-used boxes and/or generates needs for more transport units to expedite the same freight volumes. Finally, Factor 3 Efficiency of Cargo Intake and Discharge demonstrates importance that shippers assigned to expediency of ordering and chartering and equipment free time for loading/unloading. Poor synchronization between the times of freight availability and the time of transport equipment arrivals and departures disturbs the delivery schedule and increases costs of logistics. Late LCUs arrivals cause cargo to wait for shipment and thus, inflate the costs of storage. Similarly, LSP incur demurrage costs for retaining freight carriage equipment beyond the lease deadlines when cargo discharge is delayed. The results of regressions run on overall importance of service quality assigned by shippers to rail routes they used for European freight transfer is shown in table 7. In order to assess the impacts of shippers attitudinal appraisals and contextual factors which jointly affect importance assigned to rail service quality, the model included three dummies extracted from descriptive analyses of rail freight flows and rail users. These included FWL and tank-wagon consignments, shipments of foodstuffs and those executed by large and medium-size shippers whose revenues exceeded 10 million Euros. Table 7: Regression of factor scores on overall importance of rail service quality Independent Variables Parameter Estimate Standard Error t-stat. p-value Intercept Full Wagon Loads Tank-Wagons Shipments of foodstuffs Shippers >10-50 million Euros and > Factor 1: Service Failure Factor 2: Intermodal Expedience Factor 3: Cargo Intake & Discharge The model parameters explained 60 percent of variance in the overall importance that European shippers assigned to rail service quality on used shipment lanes. We can see that it was solely Factor 2, Intermodal Expedience, which contributed to the latter. Other significant variables included shipments of foodstuffs and shipments by companies with revenues exceeding 10 million Euros. Consignments of full wagon loads and tank wagons, although significant were not considered very important for the overall service quality on the routes chosen. This is understandable given the fact that wagon loads constitute traditions rail service which usually operate in a single-modal fashion. The positive and highly significant impact of Factor 2 signals that shippers attached more

17 importance to efficiency of intermodal operations which are much more challenging as regards service quality for high-value cargo such as foodstuffs. 4.4 Determinants of shippers choices of road based freight supply solutions Table 8: Service quality dimensions and loadings of importance assigned to qualities of road-based freight supply Variable number Factor Description Loadings Factor 1: Operational Efficiency & Sustainability Imp_19 Environmental friendliness.730 Imp_ 5 Kindness of service.690 Imp- 8 Efficiency at trans-loading stations.631 Imp_23 After delivery service.629 Imp_22 Information promptness on cargo under shipment and after arrival.557 Factor 2: Service Availability Imp_ 1 Service availability at origin point.786 Imp-12 Reliability of service.753 Imp_ 2 Service availability at destination point.750 Factor 3: Dealing with Service Failures Imp_21 Processing of loss and damage.855 Imp_20 Amount of loss and damage.789 Imp_13 Directness of shipment.678 Factor 4: Technical Efficiency Imp_10 Suitability of Load Carrying Units for commodity carried.845 Imp_11 Suitability of Load Carrying Units for shipment size.795 Factor 5: Value for Money Imp_17 Cost of door-to-door delivery.816 Imp_18 Flexibility of dealing with variations in no of shipments.560 Five variables loaded on Factor 1 Operational Efficiency and Sustainability, showing that shippers emphasized importance of environmental friendliness, kind service, and especially error-free invoicing and prompt information after delivery. Significance of access to information on cargo whereabouts under transfer and efficient transloading stations for load consolidation and deconsolidation was also articulated by the users. Factor 2 Service Availability involved loadings of three quality indicators that revealed that shippers ranked service availability at shipments origins and destinations as important for cost-effectiveness of forward and return trips with import/export shipments. Reliability of service was also in focus here as service availability may affect consistency of transit time, and delay cargo arrivals and pick-ups.

18 Factor 3 Dealing with Service Failures comprised loadings of three variables which emphasized importance that shippers assigned to good after-delivery-service for processing cargo loss and damage, reduction of loss and damage and negative consequences of lees direct shipments. Similarly to rail carriers, failure by truck and truck-intermodal operators to fulfil these service requirements enhances shippers financial loss when their cargo gets damaged, lost and/or delayed. The observation, however, that this factor ranked third in the amount of variance explained in data material related to truck-served routes reveals that this service dimension is important, but maybe better attended to by motorized hauliers than by rail-based operators. Two variables that loaded on Factor 4 Technical Efficiency indicate that shippers rated as significant suitability of LCUs for the commodity carried and shipments size. This finding confirms a tendency observed in European and the US transportation and logistics markets. According to interviews with leaders from major European and global LSP and manufacturing companies, cheap transport ceased to be an unstated component of logistics design on which so much world s economy has been based. Lean inventory and globally sourced supply chains depended on cheap transport. However, the consequent huge demand for transport has unsurprisingly resulted in price increases. Greater demand, combined with implications of political and military crisis in oil-rich Middle East Asian countries have pushed-up the price of fuel. In European countries with tight labor markets, this has brought about higher operations and personnel costs, particularly for truck drivers. Under these circumstances the more volumes one manages to squeeze out of LCU s carrying capacity, the more money is saved on freight shipment. The latter provides also causally plausible explanation for Factor 5 Value for Money involving loadings of two variables, the cost of door-to door delivery and flexibility of dealing with seasonal shipment variations. The combination of these two variables underscores the everlasting downward pressures on unit cost of freight movement. It also signifies that price paid by shippers for freight conveyance is an important decision-making factor even if its ranking was preceded by other quality attributes. Table 9: Regression of factor scores on overall importance of road service quality Independent Variables Parameter Estimate Standard Error t-stat. p-value Intercept Factor 2:Service Availability Factor 3 :Dealing with Service Failures Factor 4: Technical Efficiency Factor 5: Value for Money

19 The above regression model explained 35 percent of variance in the overall importance assigned by European shippers to service quality on motorized routes. We see that four of the five factors extracted from the data on road transit have significantly contributed to the above. The relatively low percentage of variance explained and the high level of error term indicate that variables other than those specified in the equation exert causal impacts on shippers choices of truck-served lanes. These variables may be related to cargo specifics shipped from/to the countries analyzed, and/or to national features of truck service markets in the twelve countries of our survey. Finally, one more reason for that may be that important determinants of truck service selection have not been covered by our survey instrument, and thus could not be tested by the above model. 4.5 Determinants of transit time by road and rail The regression models on the dependent variable transit time assessed how this factor was affected by distances travelled and the features of shipment corridors in different countries and regions. Transit time by road Table 10: Results of linear regression analysis on total transit time by road Independent Variables Parameter Estimate Standard Error t-stat. p-value Intercept Sea transit (km) Transfer <1,000 km, East Transfer >1,001-2,000 km, East Transfer >2,001-3,000 km, East , Transfer > 3,001 km, East , Transfer < 1,000 km, West , Transfer >1,001-2,000 km, West , Transfer >2,001-3,000 km, West , Transfer >3,000 km, West , Frequency of service (daily) , Goods shipped (chemicals) , Shipments origins (West Europe) , The model parameters in table 14 explained 89 percent of variance in time of door-to-door freight transfer by road. Route kilometres in transit by the different modes are, not unexpectedly, the strongest determinants of duration of transit time. However, the negative and significant coefficient values on three dummy variables: service with daily frequency, shipments of chemicals and consignments with origins located in west-europe reveal types of shipments that are moved significantly faster.

20 Transit time by rail Table 11: Results of linear regression analysis on transit time by rail Independent Variables Parameter Estimate Standard Error t-stat. p-value Intercept 22,585 9, Sea transit (km) , Transfer [1,000-2,000> km, West , Transfers > 2,001 km, West , Transfers < 1,000 km, East , Transfers >1,001 km, East , Shippers >50 million Euros Intermodal transfer by road (km) Intermodal transfer by rail (km) Eastern border crossings, rail (km) , Mode shifts The model explains 79 percent of variance in duration of freight transit by rail. The nine parameters do positively and negatively affect the duration of rail transit time. The impact of the tenth determinant was not significant due to co-linearity with other residual variables. Rail transit time increases with distances of sea crossings, overland haulage, feeder travel to terminals trans-loading cargo onto rail carriage, and the border crossings into and within east-europe. The model indicates also that large shippers manage to reduce transit time for their consignments and that usage of rail haulage under intermodal freight transfer shortens the shipments overall travel time. This may happen because large shippers possess efficient service procurement departments who ship large and regular consignments, and their shipments are prioritized by service providers who manage to produce operational advantage from inclusion of rail line-haul into intermodal supply chains. Comparison of determinants of travel times by rail and by road discloses that border crossings on routes through west and east-europe do not affect duration of truck transport. Border crossings on east-european corridors do however, influence duration of freight travel by rail. Yet, this did not apply to west European lanes. These findings provide empirical support for anecdotal evidence that long stops at borders in north-eastern and south-eastern European countries disrupt freight movement by rail and by so doing constitute significant barriers to seamless freight flows. 4.6 Logistic regression model for choices between rail and road routes In the logistic regression model, the dependent variable in this model was defined as a binary choice between road and rail transit. In this model, the

21 impacts of variables representing shippers demographics and types of shipments, the GIS-overlays of geographical routes as well evaluation of shippers satisfaction with service qualities on the routes used have been tested as explanatory variables. Model with a combination of variables from each of these categories showed useful for prediction of road and rail modal choices. The independent variables in the model involved shipment volumes, transit speed, commodity types, and satisfaction score of shipment reliability. The logistic regression coefficients from the model were used to estimate odds ratios for each of the independent variables through an iterative maximum likelihood method. In the final step, the fit between the model prediction and the data reached 86 percent. Rail was more difficult transport choice to predict than road. Model-data accuracy for rail choice prediction was 41 percent. The truck was correctly predicted by model for 97 percent of road choices in data set. The model s explanatory variables are shown in table below, along with the corresponding ratio-changes in the odds, Exp(B), of rail-choice as a consequence of one-unit increment in predictors values. Table 12: Explanatory variables in logistic regression model for choices between road and rail. Variable Unit Exp(B) - Ratio-change in odds for rail Goods volume yearly shipped by respondent on route 1,000 tons Transit speed for SITC6 (semi finished goods) 1 km/h Transit speed for SITC7 (machinery, technical equipment) 5 km/h Satisfaction score (1-5) on Reliability of Shipment 1 point The Exp(B)-factors of above 1.0 imply that the probability of choosing rail on shipment routes is expected to rise with increase/growth in the value levels of model parameters. An assumption of increase in volume of yearly freight shipped by 1,000 tons is probably not unlikely for some large shippers in the South-eastern Europe with fast-growing markets for international freight transport. The speed of freight travel variable gave the best prediction when combined with SITC main categories 6 and 7, semi-finished products and machinery/technical equipment. Since the speed of transit interacts positively with these two goods categories, the effect of improved speed of rail freight movement increases the probability of choosing rail for supply of these two goods categories, whilst other cargo categories are not significantly affected. In this example, a one-point rise in satisfaction with reliability of rail service achieved the highest odds of mode change in this example. A more

22 realistic scenario would rather be to level up shipper s satisfaction with truck by improving rail customers satisfaction by 0.5 point. This moderate option indicates that considerable mode shifts are possible in regions/corridors with potentials for higher service reliability. One hypothetical explanation for these prospects is that, by and large rail is still cheaper than truck as regards transfer of some processed commodities, provided of course, that other requirements related to e.g., volumes of shipments could be met Baseline Yearly volume tons Reliability +1 point Speed +5 kmh Food and live animals 5 Chemicals and related products 6 Manufactured goods classified chiefly by material 7 Machinery and transport equipment 8 Miscellaneous manufactured articles Total Figure 10: Probability of choosing rail by shippers of five goods categories in four scenarios The above figure shows the probabilities for shippers choices of rail for shipments of the five main goods categories obtained from runs of the three scenarios (in addition to baseline). The baseline included current values of explanatory factors. The runs of three scenarios show the changes in predicted probabilities when: 1) Shippers yearly volumes increase by 1,000 tons, 2) Customer satisfaction with rail service reliability rises by 1 point, or that 3) Speed of rail transit is improved by 5 km/h. The latter affects two commodity categories. These probabilities refer to the median score for each of the commodity segments. 5 SYNTHESIS OF FINDINGS The first important finding deduced from this study is that shippers surveyed use the TEN-T rail and road networks which are focus for

23 European and national investments for improvement of freight transfer between the different stretches of intermodal infrastructures. The second important result is that railways transfer today more intermodal freight transport units than rail wagons. Three types of LCUs dominate rail transit in countries analyzed, 20 and 40feet containers, and swap-bodies. This indicates that rail capitalizes on its inherent competitive advantage in door-to-door segments and intermodal chains. Rail competes with road in markets for manufactured goods, chemicals, crude materials and fuels and even foodstuffs and beverages. This marks considerable up-market shift from bulk haulage where rail used to compete with low-value high-volume sea-going cargo. However, rail commands about 50 percent lower prices for one ton of freight carried in markets for finished and semi-finished products, machinery and technical equipment. The most avid users of rail services are medium-size shipper companies with yearly turnover 10 and 50 million Euros. This important and counter-intuitive finding emphasizes that given that medium-size business are much more numerous in Europe than very large establishments, rail market potentials for rail may increase considerably with increased service qualities. However, shippers satisfaction with qualities delivered by rail was lower than with truck. It is not to say that shippers were uniformly satisfied with performance of motorized operators. Yet, the gaps between expected and delivered service qualities were broader for rail than for truck. Comparisons of regression models testing impacts of factor scores extracted from choices of road and rail routes shows that factor Intermodal Expedience was the most significant for shippers overall assessment of rail quality. For track users, three factors, Service Availability, Dealing with Service Failures, Technical Efficiency, and Value for Money were important. The above signals that efficiency of intermodal operations need to be improved by rail carriers. The structure of rail and road models reveals also that more parameters affect modal choices of rail lanes that road. This result was the third important outcome from this study. Finally, the fourth substantial finding was derived from the logistic regression model testing sensitivity of relative probabilities for shippers choices of rail versus road conveyance. Model s runs showed that three service improvements may significantly increase the likelihood of shippers future selection of rail: 1) Shippers increase their volumes shipped. 2) Rail improves speed of freight transit. This may facilitate more shipments of semi-finished products, machinery and technical equipment with relatively high unit ton price. 3) Rail improves shippers satisfaction with reliability of rail. This may considerably increase rail service opportunities for transfer time-sensitive high-margin goods, such as foodstuffs.