R E S E A R C H

includes research articles that focus on the analysis and resolution of managerial and academic issues based on analytical and empirical or case research

Modal Split between Rail and Road Modes of Transport in India

Prosenjit Dey Chaudhury

Executive Summary

KEY WORDS Modal Split User Cost Vehicle Operating Costs Value of Passenger Time Feasible Generalized Least Squares

In the 1950s, the rail mode occupied a dominant position in transport within India. Since then, however, the transport sector in the country has been characterized by a secular decline in the share of rail mode. Internalization of the external costs of transport may not be sufficient for the achievement of a socially optimal modal split unless account is taken of the factors behind the current modal split. This paper attempts an investigation of these issues on the basis of data relating to eight representative sections in the country where the two modes are in competition. India became a decidedly road-dominant economy in the beginning of the eighties with the railways losing out in respect of freight traffic in addition to its already declining share in passenger traffic. The dominance of road over rail has since continued unabated till the present and is likely to continue into the future. This paper reviews the trends in transport and modal split in India from the fifties onwards and looks at the factors likely to influence modal choice. In the literature, an individuals choice of mode is divided into two main categories: personal characteristics of the individual (income, tastes, auto ownership, competing family needs for the car) characteristics of transportation alternatives available (relative time, cost, and comfort). Based on time-series including user costs, per capita domestic product, and consumption expenditure, an econometric analysis of inter-modal competition in the eight sections selected for the current study reveals the following: In the case of passenger traffic, increases in the user cost difference and the user cost ratio between road and rail have an upward impact on the relative traffic volume of rail. Income (as represented by per capita gross state domestic product) seems to play a part in determining choice between travel by car on road and first-class/air-conditioned travel on rail. The relationship between modal split and user cost difference/cost ratio in the case of competition between bus on road and second-class/sleeper-class travel on rail appears to be a non-linear one. In the case of freight competition, the modal share of rail does not go up with increase in the user cost difference or cost ratio between road and rail. It is the income variable that appears to influence modal choice in freight transport in the expected manner with shippers patronizing the qualitatively superior road mode when per capita state domestic product goes up. To arrive at a socially optimal modal split, therefore, it is necessary to concentrate on improvements in the quality of service on rail while at the same time devising measures to internalize the external costs of transport.

VIKALPA VOLUME 30 NO 1 JANUARY - MARCH 2005

17
17

he rail and road modes are worldwide the dominant modes of transport. The origin and rapid growth of railways in the nineteenth century meant, in some cases, the displacement of the road mode, both for passenger and freight movement. This was especially the case for long-distance travel before and in the early days of the internal combustion engine. In the early years of the twentieth century, however, the era of motorization set in and travel by road became more popular. After the Second World War, rapid industrial development was accompanied by acceleration in the growth of motorized transport. Volumes of traffic on the rail and road modes grew significantly with the latter often showing a greater increase than the former as it could more readily meet the demand for transport among different sections of the population. In the 1950s, the rail mode occupied a dominant position in transport within India. Since then, however, the transport sector in the country has been characterized by a secular decline in the share of rail in the total traffic carried by both road and rail, although, in absolute terms, traffic on both modes has increased significantly. The decline in the rail share has been pronounced for both passenger and goods traffic. This phenomenon gives rise to a number of issues that must engage the attention of the policy-maker. While it is true that the road mode has inherent advantages of convenience, flexibility, and adaptability and may in many cases be qualitatively superior to the rail mode, nonetheless, its dominance may not imply a socially desirable modal split. A number of studies has found that the external costs of rail transport are lower than those for road transport (Button, 1993; Government of Australia, 1994, 1995, and 1996; Government of New Zealand, 1996; Rennings et al. , 1999; Savelli and Domergue, 1998; Wiederkehr, 1998). The market-determined split between rail and road may be corrected through the internalization of the external costs of transport such as resulting from pollution and accidents. Accordingly, the policy-or decision-maker must find ways to internalize these external costs in order to ensure desirable modal choice in transport. However, a proper inquiry into the subject should begin with an understanding of the factors that determine the current modal split in transport. Internalization of external costs may not be sufficient for the achievement of a socially optimal modal split unless an account is taken of these factors.

We shall, in this paper, attempt an investigation of the factors behind the choice between rail and road in India on the basis of data relating to eight representative sections in the country where the two modes are in competition. Since the data is aggregative in nature and does not cover all the variables that should ideally be included in such an exercise, the findings are meant to provide preliminary, general ideas about the factors behind modal choice in passenger travel and freight shipment. We first review the trends in transport and modal split in India concentrating on the two principal modes and the findings of the important committees such as the National Transport Policy Committee. Next, we look at the factors likely to influence modal choice in transport and, finally, describe our own exercise in understanding modal choice between rail and road in the country.

TRENDS IN TRANSPORT AND MODAL SHARES IN INDIA

The Committee on Transport Planning and Coordination (Planning Commission, 1966), set up in 1959, noted that the burden of the increase in internal traffic since the First Plan had fallen mainly on the railways and on the road transport. Over the period 1950-51 to 1964-65, rail freight traffic increased nearly two and a half times and the same category of traffic on road went up almost four times. During the same period, the share of rail in the total freight traffic carried by rail and road came down from 79 per cent to 73 per cent. The Committee noted that railways accounted for close to 77 per cent of the movement of bulk commodities such as coal, iron ore, limestone, cement, and petroleum products. Passenger traffic by rail increased by nearly 40 per cent over the period 1950-51 to 1964-65, while passenger traffic by road went up more than three times. During the period 1950-51 to 1963-64, freight traffic increased at a rate distinctly faster than either the rate of growth of national income or the expansion of output in the industrial, mineral, and agricultural sectors. While national income went up almost 60 per cent over the period, the total freight tonne-kilometres of rail and road showed a more than two-fold increase. Passenger traffic also tended to rise somewhat faster than growth in national income, showing an almost two-fold increase. The Committee attributed the faster growth of transport output to the emphasis given to the development of industries, especially heavy industries, since the Second
MODAL SPLIT BETWEEN RAIL AND ROAD MODES OF TRANSPORT IN INDIA

18
18

Plan. However, the higher rate of growth of the transport sector did not indicate that the supply of transport services had always been able to keep up with demand. During the period under review, it was seen that in the years when there was a slackening in economic growth, excess capacity was present in the transport sector, especially on the railways. However, in the other years, when the tempo of economic activity was picking up, the transport sector could not cope with the demand and there were severe bottlenecks. The emphasis of the Five Year Plans on heavy and basic industries tended to increase the proportion of traffic in bulk commodities carried by the railways. On the other hand, the growth of consumer goods industries and industries requiring special facilities had led to increased demand for road transport facilities. The report of the National Transport Policy Committee (NTPC) noted a marked decline in the rail share in total traffic carried during the period 1950-51 to 1977-78 (Planning Commission, 1980). However, in absolute terms, the volume of traffic carried by rail had undergone a sharp increase. Thus, while the freight traffic carried by rail in 1950-51 was 44 billion tonne kilometres (btkms), the number had gone up to 163 in 1977-78, an almost four-fold increase. Not only had the volume of originating traffic carried by rail undergone a sharp increase, but the average lead of freight shipment on rail had increased one and a half times over the same period standing at 686 kms in 1977-78. The bulk of freight traffic carried by rail comprised of goods like coal, iron and steel, cement, fertilisers, and petroleum products. The proportion of such goods increased from 55 per cent to around 80 per cent of the total rail freight shipment between 1950-51 and 1977-78. The traffic in general goods remained more or less stationary between 1960-61 and 1977-78, confined to a range of 45-50 million tonnes. In keeping with the trend observed by the Committee on Transport Planning and Coordination, an increasing proportion of traffic in manufactured or high-value products had gone over to road transport which had been carrying such freight over progressively longer distances. Freight traffic carried by both the rail and road modes increased almost five times between 1950-51 and 1977-78. While in the fifties and the sixties, the rate of growth of freight traffic was nearly twice as much as that of national income, in the seventies, freight traffic on road and rail had slowed down, growing at the same
VIKALPA VOLUME 30 NO 1 JANUARY - MARCH 2005

rate as national income. The rate of growth of passenger traffic had, however, been much higher than the growth rate of population and national income. Passengers travelling in second-class constituted over 95 per cent of the total passenger traffic of the Indian Railways. Although in respect of long-distance travel, rail was the cheapest and quickest mode of transport, especially for second-class passengers, nevertheless, roadways provided a significantly better service for short-distance travel. The NTPC Report observed that, in the seventies, the growth of transport capacity lagged considerably behind the requirements of the national economy. The railways came under considerable pressure to meet the burden of transport without commensurate investment in rolling stock or line haul capacity, resulting in bottlenecks. Unforeseen shifts in the pattern of traffic placed additional strain on the railway system which from time to time had to also cope with dislocations caused by floods and other natural calamities. At times, even road transport could not meet the increasing demand for freight shipments. Although there had been a steady growth in the number of commercial vehicles, there was, at times, an acute shortage of trucks. Since the primary objectives of the NTPC were to recommend an optimal inter-modal mix of different modes of transport and to suggest organizational, administrative, fiscal, and legal measures for giving effect to recommended national transport policy, it dealt extensively with traffic forecasts and the optimal allocation between modes. Estimates were worked out for a time horizon extending till the end of the twentieth century. According to the NTPCs projections, the railways were expected to carry 468 btkms in the year 2000 as against the 155 btkms carried in 1978-79. The major part of the projected traffic would be moving over long distances. Road transport was expected to carry 182 btkms of traffic at the end of the twentieth century. Of this, nearly 130 btkms would be intra-regional and 52 btkms inter-regional. Accordingly, the percentage shares of rail and road in freight traffic worked out to 72 per cent and 28 per cent respectively. The inter-modal allocation was based on calculations of resource costs and took into account the shadow price of scarce inputs like energy. The NTPC took into account an expected rise in the price of diesel and its consequential impact on break-even levels (i.e., distances of traffic where the costs of transport across different modes are equalized),

19
19

and assumed a shift to rail of at least 50 per cent of traffic moving by road beyond these break-even levels. However, the Committee stated that the increase in rail share would not materialize unless appropriate investment and pricing policies were pursued to ensure the suggested modal split. It was of the view that if minimum resource cost was to be the guiding principle for determining inter-modal mix, the railways should play a larger role in the nations transport system. The NTPCs expectations of future modal split in this regard have not been fulfilled: the modal shares of rail and road in the total freight traffic carried in 2000-01 were 26 per cent and 74 per cent respectively (Ministry of Railways, 2002; Ministry of Surface Transport, 2001). The last major committee to look into the development of the transport sector as a whole in India was the Steering Committee on Perspective Planning for Transport Development (Planning Commission, 1988). The report of the Committee contains projections of transport demand based on a study by RITES which makes projections of traffic volumes and average leads for the years 1994-95 and 1999-2000. Rail freight traffic was projected at 462 btkms in 1999-2000 with an average lead of 852 kms. The corresponding road traffic was projected at 157 btkms, the average lead being 397 kms. The break-up between rail and road is accordingly in the ratio 75:25. Assuming that recent trends of transport coefficients and average leads would continue for the future, the Committee projected freight traffic on rail in the year 2000 at 516 btkms. By the trend growth rate approach, rail freight traffic was projected at 374 btkms
Table 1: Rail and Road Traffic Volumes and Modal Shares
Year Rail Freight Traffic Rail Pass. Traffic Road Freight Traffic

and road freight traffic at 686 btkms in 1999-2000, giving a modal split of 35:65 between rail and road respectively. For passenger traffic in the same year, the rail volume was projected at 256 billion passenger kilometres (bpkms) and the road traffic at 2916 bpkms, giving a modal split of 8:92 between rail and road. While no transport committees with such broad objectives as the ones discussed here have since been set up, nevertheless, from time to time, the Government of India has constituted expert groups to look into the aspects of one or more transport modes in the country. Table 1 gives a summary of the historical trend in traffic volumes and modal shares of the rail and road modes of transport (Planning Commission, 1988, 2001; Ministry of Surface Transport, 1996, 1999, 2001; and the Annual Statistical Statements of the Indian Railways for various years). India became a decidedly road-dominant economy in the beginning of the eighties with the railways losing out in respect of freight traffic in addition to its already declining share in passenger traffic. The dominance of road over rail has since continued unabated till the present and is almost certain to continue into the future. The share of rail in the total freight traffic carried by both rail and road declined from 61 per cent in 197071 to 47 per cent in 1980-81, 30 per cent in 1990-91, and 26 per cent in 2000-01. The decline in the share of rail passenger traffic is almost equally dramatic: the rail mode had a much reduced share of 31 per cent in 197071 which declined to 24 per cent in 1980-81, 15 per cent in 1990-91, and seems to have risen slightly to 18 per cent in 2000-01.

Road Pass. Traffic

Total Freight Traffic

Total Pass. Traffic

(BTKM) 1950-51 1960-61 1963-64 1964-65 1970-71 1977-78 1980-81 1985-86 1990-91 1995-96 2000-01 44 88 107 107 127 163 159 206 243 274 312

(BPKM) 67 78 89 93 118 177 209 241 296 342 457

(BTKM) 12.09 32.53 41.05 45.46 82.36 114.97 178.36 307.03 566.66 762.00 899.26

(BPKM) 44.80 105.04 139.70 162.13 263.09 484.98 664.83 1038.56 1615.20 2238.00 2127.96

(BTKM) 56.09 120.53 148.05 152.46 209.36 277.97 337.36 513.03 809.66 1036.00 1211.26

(BPKM) 111.80 183.04 228.70 255.13 381.09 661.98 873.83 1279.56 1911.20 2580.00 2584.96

Rail Modal Share in Freight Transport (%) 78.45 73.01 72.27 70.18 60.66 58.64 47.13 40.15 30.01 26.45 25.76

Rail Modal Share in Pass. Transport (%) 59.93 42.61 38.92 36.45 30.96 26.74 23.92 18.83 15.49 13.26 17.68

Note: pass. passenger, BTKM billion tonne kilometres, BPKM billion passenger kilometres.

20
20

MODAL SPLIT BETWEEN RAIL AND ROAD MODES OF TRANSPORT IN INDIA

During the period 1950-51 to 2000-01, the elasticity of total freight transport carried by road and rail in India with respect to GDP was 1.48 while the same for total passenger transport was 1.70. Thus, the total transport output in India with respect to both freight and passenger service has grown faster than the national income. The total freight traffic increased from 56 btkms in 1950-51 to 337 btkms in 1980-81 and then to 1,211 btkms in 200001, a more than twenty-fold increase since 1950-51. The total passenger traffic grew from 112 bpkms in 1950-51 to 874 bpkms in 1980-81 and subsequently to 2,580 bpkms in 2000-01, a twenty three-fold increase since 1950-51. Between 1950-51 and 2000-01, the elasticity of rail passenger transport service in India with respect to GDP was 1.01. Passenger service on rail, therefore, appears to have kept pace with GDP. From a figure of 67 bpkms in 1950-51, rail carried 209 bpkms in 1980-81 and subsequently 457 bpkms in 2000-01, an almost sevenfold increase since 1950-51. Suburban traffic accounted for about 19.4 per cent of the total bpkms in 2000-01. The elasticity of rail freight transport with respect to GDP turns out to be 0.86. The number of net freight tonne kilometres (revenue-earning) went up from 44 billion in 1950-51 to 159 billion in 1980-81 and then to 312 billion in 2000-01, an increase of seven times since the beginning. All forms of road transport have shown spectacular increase in volume since independence. Road passenger traffic and road freight traffic grew at annual rates of 8.02 per cent and 9 per cent respectively during the period 1950-51 to 2000-01. The elasticity of road freight transport with respect to GDP was 2.15 while for road passenger transport, it was 2.00. These elasticities are substantially higher than those of rail transport. The passenger kilometres of road transport went up from 45 billion in 1950-51 to 665 billion in 1980-81 and then to 2,128 billion in 2000-01, an almost fifty-fold increase since the initial period. In terms of net tonne kilometres, freight movement by road transport rose from 12 billion in 1950-51 to 178 billion in 1980-81 and subsequently to 899 billion in 2000-01, an increase of seventy-five times since 1950-51.* Transport volumes have actually grown to levels greater than those predicted in the work of the committees described above. The pattern of economic development with increasing dispersion of industries and markets, the nature of modern production with requirements of

efficient delivery of factors and products, and, to a certain extent, the spurt in passenger movement on account of higher incomes all mean that the demand for transport has been growing at a faster rate than the growth in national product. These developments have also meant a lower share for the rail mode than predicted in earlier studies. Road transport has been more flexible than rail transport in adapting to the needs of the economy, specializing in the transport of high-value, non-bulk products. There has been spectacular development of motorized road transport both for passenger and freight movement. While the greater share of the road mode in transport demand may be explained by inherent advantages in terms of accessibility, convenience, and door-to-door delivery, factors such as underinvestment in rolling stock and line haul capacity on the rail mode, along with the lack of a customer-oriented approach, have led to an increasing shift in patronage towards the road mode.

FACTORS LIKELY TO INFLUENCE MODAL CHOICE

A number of variables might be included in a study of the factors behind a choice of transport mode. Intuitively, the relative cost of alternative transport modes should have an influence on the decision-maker: the cheaper mode ought to be the preferred mode. However, it is not immediately known which form of the cost variable is most relevant. Some studies have used the cost ratio between alternative modes while others have used the cost difference. Under the difference formulation, the consumer gives the same amount of consideration to choosing between a Rs 1.05 and Re 1.00 pair of alternative as between a Re 0.10 and a Re 0.05 pair. On the other hand, if the cost ratio between alternative modes is of importance, then the first pair of alternatives would have to be, say, Rs 2.00 and Re 1.00 if the consumer is to be indifferent between this pair and the Re 0.10/Re 0.05 pair. Lave (1969) is of the opinion that neither the difference nor the ratio formulation seems to be absolutely correct but that the truth would seem to lie close to the former. The variable of relative time of travel of alternative transport modes also presents the same problem of choice as that between the difference and the ratio specification. In support of the difference formulation for the relative time variable, Lave (1969) cites contemporary writings concerning the value of time (Becker, 1965; Moses and

* Calculated from data in the above sources. VIKALPA VOLUME 30 NO 1 JANUARY - MARCH 2005

21
21

Williamson, 1965). Once it is assumed that time has monetary value, then a time differential can be expressed in currency units, and the relative time and relative cost information for a given fair of alternatives may be combined into a single relative cost figure to describe the difference between the two modes. Among other things, this makes it possible to say that, other things being equal, the passenger or consignor will choose that mode which has the lowest combined cost. The measurement of the value of time is important because time savings amount to most of the potential benefit from improvements in transport, typically about 80 per cent. The next important influencing variable on modal choice is relative comfort and convenience. Lave (1969) argues that deficiency in both comfort and convenience of public transport in the US had been the most important factor in the post-war decline in the patronage of mass transit. However, hardly any study had made an attempt to quantify the comfort variable and use it in a modal split model. Ideally, the analyst should have a scale of subjective valuations of comfort and a corresponding list of objective characteristics such as seating dimensions, crowding, booking procedure for shipments, etc. It would then be possible to develop on objectively measurable set of indices of comfort. The above three variables relative costs, relative time, and relative comfort are instrumental variables in the sense that they could be useful for implementing some normative goal on the part of the decision-maker. The next important variable is personal income. It is a difficult variable to handle since its influence may be felt in other directions and it may interact in a complex manner with other variables influencing modal choice. On account of its collinearity with many other variables, a number of aggregate modal split models have derived the major part of their explanatory power from the income variable alone. If one is analysing the modal choice between, say, road and rail in respect of first-class passenger travel, then the income variable is important since rising income levels might help to explain why in certain situations car travel, which is more expensive, is being preferred by the consumer of transport services to travel on first-class rail. One may also look upon the income variable as operating a constraint on choice of mode. It is reasonable to assume that if incomes fall below a certain critical level, then the commuter will not be able to afford car travel. Other variables that might be included in a detailed

study of modal split are purpose of trip, family size and composition, sex and age of the commuter, and distance of travel.

GENERATION OF DATASETS
In India, there is no detailed or extensive database on modal split between transport modes on important routes along with the costs of transport operations and other important variables such as user perception of travel or shipment on alternative modes. Studies in the past have looked at the important trends in transport in the country and sometimes discussed specific modal costs. However, the information contained in these studies does not permit the construction of a sufficient database for econometric analysis. On the one hand, the data on modal splits is either at an aggregate level or confined to a few selected routes, and on the other hand, in many cases, the data on modal splits is not accompanied by corresponding data on user costs on the part of the passenger or shipper. In addition, there is no time series of modal split and accompanying factors such as user costs and perceptions of the quality of service. For this study, relevant data on modal split between rail and road, user costs, and per capita income could be obtained either directly or estimated for eight representative sections in the country: New Delhi-Mughal Sarai, Jalandhar-Jammu, Jabalpur-Allahabad, LucknowGorakhpur, Secunderabad-Wadi, Gudur-Renigunta, Bhopal-Ujjain, and Ratlam-Godhra. In all these sections, the rail and road modes are in competition with each other, both in passenger and freight traffic, the rail track (whether single-line or double-line) being contiguous with a national or state highway (mostly two-lane). The first four of the above sections have national highways while the remainder have state highways. Besides, the selected sections vary in respect of terrain and length. The lengths for railways were worked out by looking at the zonal working time-tables giving distances between successive stations for the concerned sections. The lengths of national highways were obtained from the Ministry of Road Transport and Highways (formerly Ministry of Surface Transport), while for state highways, they were calculated on the basis of state road maps. The longest ection, namely New Delhi-Mughal Sarai, has a rail route length of 780 kms and a road length of 825 kms, while the shortest section, Gudur-Renigunta, has a rail route length of 83 kms and a road length of 75 kms. Our objective is to relate the modal split between
MODAL SPLIT BETWEEN RAIL AND ROAD MODES OF TRANSPORT IN INDIA

22
22

rail and road to important explanatory variables such as relative user costs and per capita income, given that suitable data on other factors, such as user perception of quality of service on alternative modes, is not available. The main hypothesis of this study is that as the difference between road and rail user costs, or the ratio between the two goes up, then, other things being equal, the share of the rail mode in the total traffic should increase. In addition, we wish to bring out the role of measures of personal income in our analysis. Since the road mode has such advantages as greater convenience and accessibility, the hypothesis to be tested with regard to the income and consumption expenditure variables is that as these go up, the share of the rail mode should decline as consumers are enabled to choose the (qualitatively better but probably more expensive) road mode. These hypotheses are examined under three scenarios of inter-modal competition. It should be borne in mind that all traffic within our selected sections, which are a part of the larger network, does not necessarily begin and end at the terminal points. The origin and destination of traffic movements may lie outside or inside the sections. Thus, a certain sectional distance is not a criterion for competition between rail and road. However, these traffic volumes are related to explanatory variables such as user costs within the sections; it seems reasonable to assume that the latter variables, in so far as they have any influencing power, can explain more or less the pattern of traffic moving within the selected sections. The role of network effects, therefore, may not be fully captured in the user cost estimates derived for the current analysis, especially in so far as the component of operating cost reflected in the fares and charges paid to the transport operator is concerned. We shall now describe how time-series data on each of the variables is constructed.

the latter); (ii) between bus on road and second-class/ sleeper-class travel on rail; and (iii) between freight services on rail and road. On the basis of the data contained in the given sources, calculated growth rates and assumptions and estimates of the average daily numbers of cars, buses, and trucks in each of the selected road sections were first made for the period 1986-87 to 2000-01, and then the passenger kilometres and net tonne kilometres represented by these vehicle numbers were worked out. In the case of rail traffic volumes, the total daily passenger kilometres (pkms) for a particular section were calculated for the year 1998-99 by multiplying the occupancy of each train1 by the lead and frequency, and then summing up across all passenger trains. For the years before and after 1998-99, the rates of change of passenger kilometres for the previous and successive years for the regional railways covering the selected section of interest were used to estimate the daily pkms of rail traffic. The following classes of rail passenger travel were taken to be in competition with travel by car: general air-conditioned, general first-class, airconditioned chair, air-conditioned first class, airconditioned sleeper, air-conditioned three tier, first-class rail, and first-class ordinary classes. The classes of rail travel that are taken to be competitive with travel by bus on road are second-class mail, second-class ordinary, sleeper-class mail, and ordinary sleeper classes. Coming to the derivation of freight transport volumes on rail, we made use of the statements of line capacity utilization in 1998-99 and the tonnage of a four-wheeler wagon to arrive at daily net tonne kilometres (ntkms) for a particular section in 1998-99. Rates of change of ntkms were calculated on the basis of data in the Annual Statistical Statements in order to derive the ntkms of other years for the particular section.

User Costs
Our next objective is to estimate the user costs of travel or shipment on each of the rail and road modes for the period in order later to analyse the relationship, if any, between cost differentials or cost ratios between the modes and the share of traffic of one mode in total traffic. The total user cost for transport service consists of a number of components. There is, first, the financial payment to the supplier (which may depend on tax and subsidy elements). This payment is reflected in bus fares, train fares, and shipment rates. Apart from this basic

Traffic Volumes
The time series of traffic volumes relating to both passenger and freight transport for both the rail and road modes were first derived in each of the selected sections. Table 2 summarizes the steps in the generation of data on traffic volumes for each case of competition between rail and road. We have looked at three cases of intermodal competition: (i) between car on road and firstclass/air-conditioned (AC) travel on rail (such categories as AC chair and AC sleeper classes being included in
VIKALPA VOLUME 30 NO 1 JANUARY - MARCH 2005

23
23

Table 2: Generation of Data on Traffic Volumes

Nature of Intermodal Competition Car on road vs first-class/AC travel on rail Variable Road passenger traffic volume Rail passenger traffic volume Sources of Data Road surveys, discussions, motor transport statistics, statistics of the ASRTU H.Q. of zonal railways of interest for the year 1998-99, Indian Railways ASST for other years Same as for the road traffic volume in the previous scenario of competition Same as for the rail traffic volume in the previous scenario of competition Road surveys, discussions, motor transport statistics, road user cost study Assumptions Car occupancy factor = 4 Average daily occupancy of train for intercity travel = 80% of stated carrying capacity Bus occupancy factor = 40 Average daily occupancy of train as above Values Obtained Time-series from 1987-88 to 1999-2000 do

Bus on road vs second-class/sleeperclass travel on rail

Road passenger traffic volume Rail passenger traffic volume

Time-series from 1986-87 to 1999-2000 do

Freight shipment by road vs freight shipment by rail

Road freight traffic volume

Rail freight traffic volume

Line capacity utilization statements for 1998-99, Indian Railways ASST for other years

Proportion of LCVs = 15%, Time-series from 1986-87 proportion of MAVs = 10% in to 1999-2000 north India and 8% in south India, LCV payload = 5 tonnes, HCV payload = 9 tonnes, MAV Av. comp. payload = 18 tonnes, load factor for LCV, HCV and MAV = 100%, 100% and 90% respectively on NH, and 90%, 90% and 80% on SH, 20% empty trucks on NH and 30% empties on SH Payload of four-wheeler do wagon = 24 tonnes

Notes: ASRTU Association of State Road Transport Undertakings, H.Q. headquarters, ASST Annual Statistical Statements, LCV light commercial vehicle, MAV multi-axle vehicle, HCV heavy commercial vehicle, Av. Comp. average composite, NH national highways, SH state highways; the reference for the road user cost study is Ministry of Surface Transport/CES (1989).

payment, the user of transport service has to incur costs for services other than those provided by the transport supplier. This category of cost includes the following: (i) cost of porterage and local transport (in the case of passenger service), and (ii) cost of packing, handling, and local cartage (in the case of freight service). The cost of porterage and local transport is relevant in the cases of travel by bus on road and travel or shipment by rail. The cost of packing, handling, and local cartage is incurred when the shipper avails of freight service on either the rail or road mode. A special category of costs in the case of freight service comprises illegal rental payments or unofficial fees that are charged to the user by various parties directly or indirectly connected with the provision of transport service. Besides these categories of costs, the passenger or shipper incurs special costs that are not reflected in transactions with other parties. When a passenger is travelling, a value is attached to the time he or she spends in transit, depending upon the opportunity cost of travel. This opportunity cost is the loss in earnings which is reflected in the hourly wage rate. The value of passenger

time (VOPT) is included as a component of the total user cost of passenger travel. Similarly, when a shipper undertakes to avail of freight service by a particular mode, he or she incurs a special cost while the commodity is in transit. This is again an opportunity cost which is measured by the interest the shipper would earn on the total value of the shipped commodity. The opportunity cost of freight shipment is termed the cost of commodity in transit and is a component of the total user cost of freight shipment. The total user cost, therefore, includes the financial cost of transport service charged by the supplier, costs that are incurred apart from the payment to the supplier, and opportunity costs. The three cases of competition and the categories of user cost applicable to each case are presented in Table 3. We now discuss the derivation of the components of user cost for each of the three cases of inter-modal competition. Exhibit 1 illustrates the different steps in the generation of data on user costs for both modes. In the case of road transport, the manual of the Indian Roads Congress (IRC) provided the means of making
MODAL SPLIT BETWEEN RAIL AND ROAD MODES OF TRANSPORT IN INDIA

24
24

Table 3: Cases of Inter-modal Competition and Components of Rail and Road User Costs Considered in the Current Study
Nature of Inter-modal Competition Car on road vs first-class/AC travel on rail Bus on road vs second-class/ sleeper-class travel on rail Freight shipment by road vs freight shipment by rail Fare/Charge Paid to Supplier of Transport Service XX XX XX Cost of Cost of Packing, Unofficial Fees/ Porterage and Handling, and Illegal Rents Local Transport Local Cartage X* XX XX XX Value of Passenger Time XX XX XX Cost of Commodity in Transit

* This component of user cost is included in the case of travel by air. X User cost is included in the user cost of only one mode. XX Particular category of user cost is included in total user cost for both the modes.

estimates of road operating cost for all types of vehicles in different types of terrain, road surface, and traffic congestion (Indian Roads Congress, 1993). The data contained in the manual was based on a road user cost study of 1982 and further results were obtained in the Study for Updating Road User Cost Data (Kadiyali and Associates, 1992). Adjusting the data for congestion and making use of assumptions relating to the composition of traffic, we have derived the estimated costs of travel by car per passenger kilometre incorporating both the higher cost of travel by taxi and VOPT. 2 In respect of the rail mode, the Annual Statistical Statements were used to derive estimates of the average rail fares per passenger kilometre for the air-conditioned and first classes during the period of interest in all the selected eight rail sections. The estimates were then expressed in 1997-98 prices by use of the GDP deflator. The costs of porterage and local transport for passenger travel by rail, as given in Planning Commission/RITES (1987-88), were adjusted for inflation to express them in terms of prices of the year 1997-98. The same source was used to estimate VOPT on rail. The sum of rail fare, costs of porterage, local transport, and VOPT on rail gives us the total user cost per passenger kilometre for travel by first-class/air-conditioned rail classes in each of our selected horizons across the given time period. The average fares per passenger kilometre for travel by bus were estimated by making use of the performance statistics of public bus companies published by the Association of State Road Transport Undertakings (ASRTU). As we are concerned with intercity traffic, in making these estimates, we have excluded the data relating to operations in metropolitan areas. The cost of porterage and local transport for bus travel was taken from the same source as mentioned above for rail travel. The VOPT pertaining to travel by bus was next worked
VIKALPA VOLUME 30 NO 1 JANUARY - MARCH 2005

out by using the data in the IRC manual which allowed us to estimate different values of VOPT in different conditions of traffic density. To estimate the corresponding user costs of travel by rail, the Annual Statistical Statements of the Indian Railways were used for the period under consideration and the average rail fare per passenger kilometre was worked out. To these estimates of rail fares, the cost of porterage and local transport and the values of passenger time per passenger kilometre (as worked out earlier) were added to arrive at the total user costs of travel in second-class/sleeper-class on rail. To estimate the road freight bill, we first estimated the costs of operation of trucks per net tonne kilometre in the same way as was done for car in the first scenario of inter-modal competition. Applying a mark-up to these operating costs, estimates of freight bills paid by the shipper on road were derived. An average cost of packing, handling, and local cartage for road shipment was worked out by taking seven principal commodities for which data was available for both road and rail. Other expenses incurred en route on the part of the truck operator included check-post expenses, charges paid to transport officials and the police, loading and unloading charges and others. Using information in the report of the Steering Committee on Trucking Operations in India, we worked out the unofficial expenses of freight shipment on road per tonne kilometre and applied them to each of our selected sections in accordance with geographical proximity. An average cost of commodity in transit on road was worked out by making use of the VOC tables referred to earlier duly adjusted for congestion as reflected in the prevailing density of traffic. In the case of rail mode, rail freight rates were estimated by using data on earnings from revenueearning freight traffic in the Annual Statistical Statements.

25
25

An average cost of packing, handling, and local cartage was next worked out by taking into account the same important commodities as in the case of road mode. The next item of user cost for freight shipment by rail considered in our analysis was the cost of extra-official fees and charges relating to documentation, claims settlement, etc. Data on commissions and extra-official fees in the Indian Railways was difficult to obtain. In working out the value of illegal rental payments in India for 1964, Krueger reproduces data of a study done in 1966 in which some information is given on this subject (Krueger, 1974). For this study, the data was adjusted in line with inflation, increasing corruption, documentation, and other charges unique to the railways to arrive at a figure representing the sum of extra-official fees and other charges to be paid by the user of rail freight service. Finally, the cost of commodity in transit for rail shipment was derived by using data for the seven specific commodities mentioned earlier.

Measures of Income/Expenditure
The next important explanatory variable that we included in our datasets was per capita income. Unfortunately, the available data did not permit us to construct a time series of personal income for each of the scenarios of competition in the sections selected for this study. Accordingly, we had to devise some proxy measures of this variable. In the case of two of the scenarios of competitionbetween sleeper-class/second-class rail and bus and between rail and road freight service we used the series on gross state domestic product (GSDP) published by the Central Statistical Organization and population figures of each state to work out values of per capita GSDP for each of the selected sections in all the years considered. The per capita GSDP values thus derived served as proxies for the income variable in our analysis of the two scenarios of competition mentioned above. While considering the competition between airconditioned, first-class rail, and car, we need to have estimates of income of the upper-income bracket of the population who are most likely to use these transport services. For this particular case of competition, we estimated the levels of minimum consumption expenditure of the richest 10 per cent of households in each of the states covering our selected sections. For the period covered by this study, there are three quinquennial surveys of consumer expenditure published by the

National Sample Survey Organization (NSSO): 1987-88, 1993-94, and 1999-2000 (termed the 43rd, 50th and 55th rounds respectively). In each of these surveys, we have looked at the distribution of a thousand sampled households in the urban sector of each concerned state over classes of monthly per capita consumer expenditure (MPCE). A lognormal distribution was fitted to the given data in order to make estimates of consumption expenditure of the richest 10 per cent of the population. Given these estimates of upper-level consumption expenditure in the years 1987-88, 1993-94, and 19992000, the values for the intervening years were filled in by looking at the changes in the ratio of estimated consumption expenditure to the per capita GSDP. It was assumed that the ratio would change according to a geometric progression. We thus constructed three datasets in which were included relative traffic volumes of the rail mode (or the ratios of rail to road traffic volumes), user cost differences, user cost ratios, per capita GSDP, and upper income level consumption expenditures. Our next task is to analyse whether there are any statistically significant relationships between the share of rail and the explanatory variables.

THE MODEL
We have taken the relative traffic volume of rail (the ratio of the rail traffic volume to the road volume) as the dependent variable while the independent variables (depending on the particular case of inter-modal competition) include: (i) the difference between the user cost on road and the same on rail, (ii) the ratio of the user cost on road to the same on rail, (iii) per capita monthly consumption expenditure of the upper-income class, and (iv) per capita yearly gross state domestic product. The inclusion of more variables in our analysis could have resulted in a greater probability of finding significant relationships under the options of various econometric models. However, because of paucity of data, we had to limit ourselves to the above explanatory variables in explaining modal spilt between rail and road. The restricted nature of the choice of explanatory variables may influence the nature of the model in which statistically significant relationships are ultimately derived. The econometric analysis of the trends in traffic in all the three cases of competition was carried out with the help of Stata 6.0. The datasets were arranged in
MODAL SPLIT BETWEEN RAIL AND ROAD MODES OF TRANSPORT IN INDIA

26
26

panels, each of which is composed of the data of a particular transport section from 1986-87/1987-88 to 19992000. Heteroscedasticity was detected in the data in all the three cases. The datasets were, therefore, analysed under generalized least squares (GLS) regression involving both random effects and fixed effects. In all the datasets, the Hausman test did not indicate any systematic differences in coefficients between the two models. The coefficients of the explanatory variables turned out to be insignificant at the 10 per cent level in most cases. Subsequently, cross-sectional time-series feasible generalized least squares (FGLS) was carried out on the datasets. The models tested were: (i) generalized least squares with heteroscedastic panels, (ii) GLS with heteroscedastic panels and within-panel correlation in the form of a common AR(1) coefficient for all panels, (iii) GLS with heteroscedastic panels and panel-specific AR(1) correlation. The model that finally yielded statistically significant relationships was found to be the cross-sectional time-series FGLS model involving heteroscedastic panels with cross-sectional correlation and panel-specific AR(1) correlation. This model is described in the Appendix.

Table 4: Cross-sectional Time-series FGLS Regressions of Relative Traffic Volume of Rail on Cost and Other Variables
Form of Equation Explanatory Variables Estimated Coefficient

Passenger transport: Competition between car on road and firstclass/air-conditioned rail Log-linear Cost difference 0.022 Consumption expenditure -0.985 Log-linear Cost ratio 0.533 Consumption expenditure -1.157 Passenger transport: Competition between bus on road and second-class/sleeper-class rail Linear Cost difference 0.162 Linear Cost ratio 0.171 Cubic Cost difference 0.475 Cost difference squared -1.088 Cost difference cubed 1.009 Cubic Cost ratio 7.399 Cost ratio squared -5.501 Cost ratio cubed 1.386 Freight transport: Competition in freight traffic between road and rail Log-linear Cost difference -1.253 Per capita SDP -0.352 Log-linear Cost ratio -0.88 Per capita SDP -0.351 Quadratic Cost difference 9.438 Cost difference squared -3.697 Per capita SDP -0.0002 Note: All the coefficients are significant at 5 per cent level.

RESULTS AND INTERPRETATION

The econometric exercise was aimed at seeing whether there is change in the relative traffic volume of rail in relation to the user cost difference or ratio between road and rail and measures of personal income and expenditure. The share of rail was expected to rise with increase in the values of the cost variables as the least expensive mode is preferred. Conversely, it was expected to fall with increase in income or expenditure since customers and shippers were likely to favour the road mode with its attendant qualitative advantages. It should be mentioned that the effects of a change in the inter-modal user cost difference (user cost ratio) cannot be isolated from the accompanying change in the inter-modal cost ratio (cost difference). Furthermore, if one of the variables changes, the other need not change in a fixed manner. Hence, the statistical relationships that we have been able to establish are conditional on the structure of changes, in the given datasets, in one cost variable and accompanying movements in the other. The main results are presented in Table 4. While sectional and year dummy variables are included in the analysis, we concentrate on the impact of the main explanatory variables of cost and income on the relative traffic volume of rail.3
VIKALPA VOLUME 30 NO 1 JANUARY - MARCH 2005

Passenger Transport
In the case of inter-modal competition involving car on road and first-class/air-conditioned rail, we find that linear relationships hold between relative traffic volume on the one hand and user cost difference/ratio and consumption expenditure on the other. The elasticity of the relative traffic volume of rail with respect to user cost difference between road and rail is only 0.022. However, a 10 per cent rise in monthly per capita consumption expenditure leads to a 9.9 per cent decrease in the relative traffic volume, with the cost difference variable being held constant. Coming next to the effect of the cost ratio between road and rail, we find that the elasticity of relative traffic volume of rail with respect to user cost ratio is 0.53. If the cost ratio is unchanged, a 10 per cent increase in consumption expenditure leads to a 11.6 per cent fall in the relative traffic volume. Equiproportional increases in the cost difference and cost ratio variables, therefore, lead to a greater upward impact on the relative traffic volume in the case of the latter than for the former variable. The relationship

27
27

between consumption expenditure and relative traffic volume suggests that as incomes rise for the upperincome bracket of the population, there is a tendency to switch over to the more expensive car mode affording greater comfort and convenience. Next we turn to the case of competition between bus on road and second-class/sleeper-class travel on rail. The results of the cross-sectional time-series FGLS regression of relative traffic volume of rail on cost variables and per capita state domestic product (SDP) yielded coefficients of the SDP variable that suggested the lack of a statistically significant impact of this factor on relative traffic volume. Inter-modal competition with bus on road does not seem to be influenced by the income factor, which in this study is represented by per capita GSDP. Hence, only the cost factors were used in the subsequent analysis. For the vast majority of the travelling population, user cost differentials or ratios appear to be a more important factor in determining modal choice. Taking the cost difference between rail and road first, we find that linear regression yields a relationship in which a unit increase in this variable is related to an increase in the relative traffic volume of rail by an amount 0.162. If we come to the cost ratio, a unit rise in this variable leads to an increase in the traffic volume by an amount 0.171. Since the semi-log and log transformations of the regression yielded insignificant coefficients, we decided to examine the results of non-linear regression of relative traffic volume on either of the cost variables. Quadratic regression yielded statistically insignificant coefficients in both the cost difference and cost variable cases. However, regression using a cubic form yielded statistically significant coefficients in both cases. For a uniform change in cost difference or the cost ratio between road and rail, the relative traffic volume of rail increases in a fluctuating manner with declining percentage increases for an initial range of cost difference or cost ratio values and increasing rates of increase for subsequent values. Ignoring the coefficients of the sectional and year dummies, we have used, as an illustrative case, the equation of the cubic regression of relative traffic volume of rail on the cost difference between road and rail to examine the behaviour of the latter if the value of the cost variable is increased uniformly in the range given by the dataset of this scenario of inter-modal competition (Table 5). For a uniform change in the cost difference between

road and rail, the relative traffic volume of rail increases in a fluctuating manner with declining percentage increases between the cost difference values 0.01 and 0.37, and thereafter increasing rates of increase for the subsequent values. The same behaviour is exhibited in the case of the cost ratio variable. Given the structure of cost differences and ratios in the given dataset, when either the user cost difference or cost ratio between road and rail rises to certain critical levels, the modal share of rail (as indicated by the relative traffic volume), which has been rising at a diminishing rate, begins to increase in an exponential manner.

Freight Transport
We finally come to the analysis of competition in freight transport between the two modes. Linear regression under the cross-sectional time-series FGLS model was variables carried out separately for cost difference and cost ratio variables. The elasticity of relative traffic volume of rail with respect to user cost difference is -1.25, while the same with respect to user cost ratio is -0.88. The modal share of rail does not go up with increase in the user cost difference or cost ratio between road and rail unlike the earlier cases of competition in passenger traffic. The elasticity of the traffic volume with respect to per capita SDP is about -0.35 in both cost difference and cost ratio relationships. It is per capita SDP, therefore, that
Table 5: Inter-modal Cost Difference and Relative Traffic Volume (competition between bus and rail)
Cost Difference between Road and Rail (Re) 0.01 0.05 0.09 0.13 0.17 0.21 0.25 0.29 0.33 0.37 0.41 0.45 0.49 0.53 0.57 0.61 0.65 0.69 Relative Traffic Volume of Rail* 0.749 0.765 0.779 0.789 0.798 0.805 0.810 0.815 0.818 0.822 0.825 0.829 0.834 0.840 0.848 0.858 0.870 0.885 Percentage Increase of Volume with Respect to Previous Value 2.20 1.76 1.40 1.10 0.86 0.67 0.53 0.45 0.41 0.42 0.48 0.59 0.73 0.93 1.16 1.43 1.73

* The ratio of rail volume to road volume.

MODAL SPLIT BETWEEN RAIL AND ROAD MODES OF TRANSPORT IN INDIA

28
28

appears to influence modal split in the expected manner, suggesting that users of rail switch to the road mode as incomes rise, the latter offering a range of facilities which the rail mode cannot provide. A quadratic regression yields statistically significant coefficients for the cost difference variable and insignificant coefficients for the cost ratio variable. We used the quadratic relationship between traffic volume, cost difference, and per capita SDP to examine how relative traffic volume of rail varies with uniform increase in the values of cost difference in the range given by the dataset, while per capita yearly SDP is fixed at an average level (Rs 11,500) and the coefficients of the section and year dummies in the equation are ignored (Table 6). The table shows that the modal share of rail increases at a diminishing rate when the cost difference between road and rail is increased from 1 to 1.28. Beyond this point, however, the modal share declines continuously at an increasing rate. We may surmise that the cost factor is not very helpful in explaining modal split in the case of competition in freight service. Qualitative factors seem to play a more important part. It is probable that increases in the user cost for shipment by road are accompanied by a rise in the perceived quality of service prompting users to switch over to this mode. It may be noted that generally, the relationship, where it exists, between the cost variable and modal
Table 6: Inter-modal Cost Difference and Relative Traffic Volume (inter-modal competition in freight service)
Cost Difference between Road and Rail 1.00 1.04 1.08 1.12 1.16 1.20 1.24 1.28 1.32 1.36 1.40 1.44 1.48 1.52 1.56 1.60 1.64 1.68 1.72 1.76 1.80 Relative Traffic Volume of Rail 3.324 3.400 3.464 3.517 3.557 3.586 3.602 3.607 3.600 3.582 3.551 3.509 3.454 3.388 3.310 3.221 3.119 3.006 2.880 2.743 2.594 Percentage Change of Volume with Respect to Previous Value 2.28 1.88 1.51 1.15 0.80 0.47 0.14 -0.19 -0.52 -0.85 -1.19 -1.55 -1.91 -2.30 -2.71 -3.15 -3.64 -4.17 -4.76 -5.43

share is a weak one. Our datasets show that the user cost difference between road and rail (as well as the user cost ratio) for both passenger and freight competition has moved upward although with significant fluctuations. The results on the competition between car on road and first-class/AC travel on rail implies that if this trend continues into the future, there will be an increasing share of the rail mode that will, however, be counterbalanced by a movement towards road as personal incomes go up. The relevant dataset shows that the rail share has indeed been rising for this set of competition. For the competition between bus on road and second-class/ sleeper-class travel on rail, the concerned dataset shows a declining share of rail although with significant fluctuation from year to year. Our results indicate a weak relationship between these fluctuations and movements in the user cost variable, other factors influencing modal split not being taken into account. The national trend of increasingly lower modal share of rail in passenger transport is almost certain to continue on account of these other factors, chief among them probably being determinants of the quality of service such as availability and comfort. Finally, as far as competition in freight transport is concerned, we have seen that increasing differentials in user costs do not explain modal split while future rise in incomes can only mean a lower share of the rail mode. Some idea on future aggregate modal splits in India may be obtained from recent studies (Expert Group on Indian Railways, 2001; Ministry of Surface Transport, 2001). As far as passenger traffic is concerned, modal shares of rail and road are expected to be 14 per cent and 86 per cent respectively in 2005-06 under the assumption of unchanged structure of rail fares. These shares are expected to change slightly in favour of road if there is adjustment in upper- and second-class rail fares. The modal shares of rail and road in freight traffic are projected at 21 per cent and 79 per cent for 201516 under the assumption of uniform growth rate in traffic of all commodities on rail. With the assumption of commodity-specific growth rates, this rail share is expected to go down slightly.

CONCLUSION
In the light of these results, it may be stated that efforts at reduction of high tariffs for shipment by rail should be accompanied by improvements in the quality of

VIKALPA VOLUME 30 NO 1 JANUARY - MARCH 2005

29
29

service. Shippers perception of the quality of service is influenced by factors such as connectivity, availability, reliability, transit time, ease of payment, negotiability, adaptability, product suitability for mode, claimprocessing time, access to decision-makers, suitability of price, customer-friendly attitude, and customer information. For all these parameters, a survey among shippers in India found that the rail mode is ranked significantly lower than the road mode (A F Ferguson & Co., 1999). The issue of quality of service is crucial in the segment of freight transport more so because it is one area in which the railways seem to have a greater environmental and financial advantage over roadways than in passenger transport (Dey Chaudhury, 2003). It is, therefore, imperative that the task of redressing the current distortion in freight modal shares be addressed on the part of the policy-maker in the interest of bringing about a socially desirable modal split. While our analysis shows that user cost factors appear to play a negligible role, attention should nevertheless be given to ways of internalizing the external costs of freight transport so that each mode of transport is made to bear the social costs of transport. In particular, heavy goods vehicles on road that cause damage to the pavement should be made to defray the cost of road repair and maintenance. Much of the profit generated by the railways in freight movement goes towards the subsidization of passenger transport. Because of social commitment, there is little scope for increase in passenger fares. If the objective is to divert passenger traffic from road, then measures aimed at the internalization of the external costs of transport, which studies show are generally lower for rail, need to be considered. Road users should be made to pay for the cost of infrastructure provision

and maintenance through toll charges. The pricing of transport services should take into account the costs of such factors as pollution and congestion. The external costs of accident need to be covered through compulsory subscription to an insurance regime. An appropriate legal and supervisory framework should be put in place to facilitate the settlement of claims. Again, as in the case of freight transport, improvement in the quality of service on rail must not be neglected. Since supply has often lagged behind demand, enhancement of capacity in passenger services on rail is all the more important in any policy initiative aimed at attracting customers from the road mode. The issue of modal choice in transport should be given more prominence by policy-makers and analysts. The loss of rail dominance to road in India is in line with the experience of many countries, but has not occupied the central position in policy discussions on the transport sector (World Bank, 1995). While the present study yields some general, preliminary results, more detailed exercises need to be carried out in order to understand better the factors behind modal choice. Studies concentrating on modal split in important transport sections would bring out the section-specific role played by various factors and possibly help to establish critical fare and tariff levels. While rail and road are the principal modes of transport in the country and the competition between the two is of clear relevance to policy-makers, it should be remembered that waterways and pipelines have a significant share in freight movement. Hence, studies on modal choice should also take into account these other modes and examine their relation to the rail and road modes.

ENDNOTES
1. In the discussions with railway officials, it was suggested that the average daily occupancy of trains for intercity travel could be assumed to be 80 per cent of the stated carrying capacity. 2. The VOPT for travel on road varies with the level of congestion whereas for rail it is fixed. This may be justified on the ground that the speed of intercity transit by rail has not varied significantly over the period. 3. The coefficients of the sectional and year dummies indicate that spatial and temporal effects are in some cases significant across all the scenarios of competition studied.

30
30

MODAL SPLIT BETWEEN RAIL AND ROAD MODES OF TRANSPORT IN INDIA

Exhibit 1: Generation of Data on User Costs

Nature of Inter-modal Competition
Car on road vs first-class/ AC travel on rail

Variable

Component

Sources of Data

Assumptions

Values Obtained

Road user cost

Operating cost

Indian Roads Congress (1993), Kadiyali and Associates (1992)

Rail user cost

VOPT Passenger fare Cost of porterage and local transport

do Indian Railways ASST Planning Commission/ RITES (1987-88)

Ratio of old-tech to new-tech cars=1:3 for 1986-87 to 1991-92 and 1:4 thereafter, proportion of taxis=15%, cost of taxi travel 25% higher than cost of travel by personal car Case of travel between met. city and mof. town may be applied to all but two of the selected sections

Time-series from 1987-88 to 1999-2000

do do Rs 29.41 per passenger for travel between met. city and mof. town and Rs. 12.00 for travel between two mof. towns (in economic terms at 1997-98 prices for the 50 km distance slab) Re. 0.913 per passenger kilometre for travel between met. city and mof. town and Rs. 1.122 for travel between two mof. towns (at 1997-98 prices) Time-series from 198687 to 1999-2000

VOPT

do

do

Bus on road vs second-class/ sleeper-class travel on rail

Road user cost

Passenger fare

Statistics of the ASRTU

Cost of porterage and local transport

Planning Commission/ RITES (1987-88)

Same as for this component in rail user cost above

VOPT Rail user cost Passenger fare Cost of porterage and local transport

Indian Roads Congress (1993) Indian Railways ASST Planning Commission/ RITES (1987-88)

Same as for this component in rail user cost in above scenario of competition do Mark-up of 13% on operating cost in order to cover brokers commission and truckers profit Weighted average based on commodity shares applicable for the period under study Costs pertaining to important routes applicable to selected sections by geographical proximity Same as for this component in road user cost i) Figure given for each wagon loaded applicable to average rail shipper ii) Increase in corruption since 1966. Weighted average based on seven principal commodities applicable for the period under study

VOPT Freight shipment by road vs freight shipment by rail Road user cost Road freight bill

do Indian Roads Congress (1993), Ministry of Railways/CES (1993)

Rs. 7.92 per passenger for travel between met. city and mof. town and Rs. 13.18 for travel between two mof. towns (in economic terms at 1997-98 prices for the 50 km distance slab) Time-series from 198687 to 1999-2000 do Same as for this component in rail user cost in above scenario of competition do Time-series from 198687 to 1999-2000

Cost of packing, handling, and local cartage Unofficial expenses

Planning Commission/ RITES (1987-88), Ministry of Railways/ RITES (1996) Ministry of Surface Transport/ AITD (1999)

Rail user cost

Transit cost of commodity Freight rate Cost of packing, handling, and local cartage Unofficial expenses

Indian Roads Congress (1993) Indian Railways ASST Planning Commission/ RITES (1987-88), latest commodity shares Krueger (1974)

Rs. 100.54 per tonne moved as the average cost (in 1997-98 prices) Re. 0.139 per tonne km for Mumbai-Delhi, Re. 0.044 for Kolkata-Delhi, Re. 0.038 for KolkataChennai Time-series from 198687 to 1999-2000 do Rs. 117.41 per tonne moved as the average cost (at 1997-98 prices) Rs. 15.09 per tonne loaded (inclusive of documentation charges and in 1997-98 prices) Rs. 7.27 per tonne for a distance slab of 650 km (in 1997-98 prices)

Transit cost of commodity

Planning Commission/ RITES (1987-88)

Notes:

old-tech old-technology, new-tech new technology, met. city metropolitan city, mof. town mofussil town, ASST Annual Statistical Statements, ASRTU Association of State Road Transport Undertakings.

VIKALPA VOLUME 30 NO 1 JANUARY - MARCH 2005

31
31

REFERENCES
A F Ferguson & Co. (1999). Final Report on All-India Shipper Survey , New Delhi. Becker, G S (1965). A Theory of the Allocation of Time, Economic Journal , 75(299), 493-517. Expert Group on Indian Railways (2001). Policy Imperatives for Reinvention and Growth: The Indian Railways Report, New Delhi. Government of Australia (1994). Victorian Transport Externalities Study: Volume I, Environment Protection Authority, Bureau of Transport and Communications Economics. Government of Australia (1995). Greenhouse Gas Emissions from Australian Transport: Long-term Projections, Report 88 , Bureau of Transport and Communications Economics. Government of Australia (1996). Transport and Greenhouse: Cost and Options for Reducing Emissions, Report 94, Bureau of Transport and Communications Economics. Government of New Zealand (1996). Land Transport Pricing Study: Safety Externalities, Discussion Paper, Ministry of Transport. Greene, William H (2000). Econometric Analysis, New Jersey: Prentice Hall International, Inc. Indian Roads Congress (1993). Manual on Economic Evaluation of Highway Projects in India, First Revision, New Delhi. Kadiyali, L R and Associates (1992). Study for Updating Road User Cost Data , Ministry of Surface Transport/Asian Development Bank, New Delhi: Government of India. Kmenta, J (1986). Elements of Econometrics , New York: Macmillan Publishing Company. Krueger, A (1974). Political Economy of the Rent-Seeking Society, American Economic Review, 64(3), 291-303. Lave, Charles A (1969). A Behavioral Approach to Modal Split Forecasting, in Mohring, H (ed.) (1994). The Economics of Transport, Volume 1, Aldershot, UK: Edward Elgar. Ministry of Railways/CES (1993). Integrated Rail Road Transport System for Movement of Long Distance Freight: Final Report, New Delhi: Government of India. Ministry of Railways/RITES (1996). Study on Decline in Railways Share in Total Land Traffic Volume: Draft Report, New Delhi: Government of India. Ministry of Railways (2002). Indian Railways Annual Statistical Statements , New Delhi: Government of India. Ministry of Surface Transport/CES (1989). Road User Charges Study, New Delhi: Government of India. Button, K J (1993). Transport, the Environment and Economic Policy, Aldershot, UK: Edward Elgar. Dey Chaudhury, P (2003). Environmental Sustainability of Transport: Issues in the Case of Rail and Road, unpublished Ph.D. thesis, Centre for Economic Studies and Planning, New Delhi: Jawaharlal Nehru University. Ministry of Surface Transport (1996). Report of the Working Group on Road Transport for the Ninth Five Year Plan , New Delhi: Government of India. Ministry of Surface Transport (1999). Comprehensive Study of Road Traffic Flows in the Country: Final Report, New Delhi: Government of India. Ministry of Surface Transport/AITD (1999). Trucking Operations in India: Report of the Steering Committee, New Delhi: Government of India. Ministry of Surface Transport (2001). Report of the SubGroup on Traffic Forecasts and Fleet Requirement in the Tenth Plan, New Delhi: Government of India. Moses, L N and Williamson, H F (1965). Choice of Mode in Urban Transportation, Unpublished Report, Northwestern University, USA. Planning Commission, (1966). Final Report of the Committee on Transport Policy and Coordination , New Delhi: Government of India. Planning Commission (1980). Report of the National Transport Policy Committee, New Delhi: Government of India. Planning Commission (1988). Perspective Planning for Transport Development, Report of Steering Committee, New Delhi: Government of India. Planning Commission/RITES (1987-88). Total Transport System Study, New Delhi: Government of India. Planning Commission (2001). Report of the Working Group on Road Transport for the Tenth Five Year Plan, New Delhi: Government of India. Rennings, K et al . (1999). Valuation of Transport Externalities, Institute for Transport Studies, University of Leeds: Project No PL 97-Z064 commissioned by the European Commission. Savelli, G and Domergue, P (1998). Rail Transport and Greenhouse Effect, paper presented at the UIC-MAPS seminar, New Delhi. Wiederkehr, P (1998). Environmentally Sustainable Transport (EST): International Perspectives, paper presented at the UIC-MAPS seminar, New Delhi. World Bank (1995). India Transport Sector: Long-Term Issues, Infrastructure Operations Division, South Asia Regional Office.

Prosenjit Dey Chaudhury is currently an Economist with Consulting Engineering Services (India) Private Limited, New Delhi. He has been involved in studies on the environmental and social sustainability of transport modes in India, overloading of commercial vehicles, proposed expressways in the National Capital Region, etc. In 2004, he received his doctoral degree
VIKALPA VOLUME 30 NO 1 JANUARY - MARCH 2005

for his thesis titled Environmental Sustainability of Transport: Issues in the Case of Rail and Road, submitted to the Jawaharlal Nehru University, New Delhi. His areas of interest include economic development, environment, infrastructure, and energy. e-mail: prosenjit_dc@yahoo.com

33
33