AN ANALYSIS OF PUBLIC BUS TRANSIT PERFORMANCE IN INDIAN CITIES

AN ANALYSIS OF PUBLIC BUS TRANSIT PERFORMANCE IN INDIAN CITIES by Madhav G. Badami* School of Urban Planning and McGill School of Environment McGill University, 815 Sherbrooke Street West, Montreal, QC, H3A 2K6, Canada Phone : (514) 398-3183; Fax : (514) 398-8376 e-mail: madhav.badami@mcgill.ca and Murtaza Haider School of Urban Planning and Department of Civil Engineering and Applied Mechanics McGill University, 815 Sherbrooke Street West, Montreal, QC, H3A 2K6, Canada Phone : (514) 398-4079; Fax : (514) 398-8376 e-mail: murtaza@regionomics.com Paper submitted for publication to Transportation Research Part A (Policy and Practice) July 1, 2006 * Corresponding author

Badami and Haider 1 ABSTRACT Maintaining and enhancing public transit service in Indian cities is important, to meet rapidly growing mass mobility needs, and curb personal motor vehicle activity and its impacts at low cost. Indian cities rely predominantly on buses for public transport, and are likely to continue to do so for years. However, the public bus transit service is inadequate, and unaffordable for the urban poor. The paper explores the factors that contribute to and affect efforts to improve this situation, based on an analysis of the financial and operational performance of the public bus transit service in the four metropolitan centres and four secondary cities during the 1990s. Overall, there were persistent losses, owing to increasing input costs and declining productivity. The losses occurred despite rapidly increasing fares, and ridership declined. The situation, and the ability to address it, is worse in the secondary cities than the metropolitan centres. We suggest a disaggregated approach based on the needs and motivations of different groups in relation to public transit, along with improved operating conditions and policies to internalize costs of personal motor vehicle use, to address the challenge of providing financially viable and affordable public bus transit service. Keywords Urban transport; Public transit; Low-income countries; India; Bus transit; Transit performance

Badami and Haider 2 AN ANALYSIS OF PUBLIC BUS TRANSIT PERFORMANCE IN INDIAN CITIES Madhav G. Badami and Murtaza Haider McGill University 1. Introduction -- Rapid Motor Vehicle Growth and Impacts Motor vehicle ownership and activity are growing rapidly in the low-income countries of Asia, owing to rapid urbanization, growing urban incomes and vehicle production increasingly moving there as markets become saturated in the OECD (Organization for Economic Co-operation and Development) countries. In India, which accounts for roughly 87% of South Asia s motor vehicles, the motor vehicle fleet has been doubling every four years over the last three decades (MORTH, 2004). Motor vehicle numbers have grown even more rapidly in the cities than nationally. Indeed, urban motor vehicle growth rates have far exceeded urban population growth rates, which themselves have been dramatic. Delhi s population grew 18-fold in the approximately 60 years since 1947, whereas its motor vehicle population grew more than twenty-fold in half the time. Delhi s motor vehicle fleet presently stands at 4.2 millions (NCTD, 2004), representing a doubling in a little over ten years. While Delhi is by far the most motorized city in India, motor vehicle growth has been more rapid in other cities, such as Ahmedabad, Bangalore and Chennai (CIRT, 1997; SIAM, 2002).

Badami and Haider 3 In order to better understand motorization in India, we consider the growth in various passenger motor vehicles in Delhi (Table 1). From the 1970s, the growth in motorized twowheeled (M2W) vehicles motorcycles, scooters and mopeds which offer affordable mobility to millions of people, has been more rapid than that for all motor vehicles. M2W vehicles have in fact been growing at a much higher rate, and far in excess of that for all motor vehicles, in India compared to Delhi. These vehicles now account for as much as 70% of motor vehicles in India. In Delhi, around 2.7 of its 4.2 million vehicles are M2W vehicles (NCTD, 2004; MORTH, 2004). The other feature of motorization, with important implications for urban transport outcomes in cities like Delhi, is that cars grew at a rate much higher than that for M2W vehicles, through the 1990s. These trends need to be seen in the context of the low growth rates for for-hire motorized three-wheeled (M3W) vehicles and taxis, and in particular for buses. As for buses, growth rates in the recent past have declined considerably from those in the 1970s and 1980s. The rapid growth in motor vehicle activity in Indian cities has brought in its wake a range of adverse impacts. Traffic congestion is increasing rapidly, causing significant time and productivity losses. Because of the concentration of motor vehicular and other energyintensive activities in metropolitan centres like Delhi, and the fact that until recently, these activities have been characterized by high pollution intensities, air quality has been poor in these centres since the late 1980s. Surveys have shown daily average suspended particulate levels, which are strongly correlated with respiratory and cardiovascular diseases, exceeding World Health Organization (WHO) guideline limits almost daily in Delhi (CPCB, 1996 and 2004). India s road accident record, already among the world s worst, is

Badami and Haider 4 deteriorating steadily, with the death toll rising from 56,400 to around 80,000 mortalities between 1991 and currently (MORTH, 2004). INSERT TABLE 1 HERE Rapid urbanization in India is accompanied by mass urban poverty, even as urban incomes rise. In Delhi, one of the wealthiest Indian cities, the average annual per capita income is only around US$ 1040, and about half the population reportedly lives in slums (NCTD, 2002; Tiwari, 2002). Although motor vehicle ownership and activity are growing rapidly, motor vehicle ownership rates, even in Delhi, the most motorized city in South Asia, is only around 275 per 1000 persons (calculations based on NCTD, 2004 and UN Population Division, 2002), as against 450-750 in European and North American cities in the early 1990s (Kenworthy and Laube, 1999). The vast majority who are poor cannot afford even the least expensive motorized modes, and therefore benefit little from motor vehicle activity. At the same time, they suffer the highest exposures and impacts due to, for example, transport emissions and road accidents, since many of them live and work roadside. While car and taxi occupants accounted for only 2% of Delhi s road accident fatalities in 1994, pedestrians, cyclists and M2W vehicle users accounted for 42, 14 and 27% respectively (Tiwari, 2002).

Badami and Haider 5 1.1 Urban Transport Prospects Motor vehicle activity in India is already causing much higher adverse local impacts than in the OECD countries, despite much lower motorization levels, and is likely to grow rapidly over the coming decades in Indian cities, along with growing urbanization and urban incomes. In 2015, a mere ten years from now, Mumbai and Delhi are expected to have populations exceeding 20 million, Kolkata about 17 million, and Bangalore, Chennai and Hyderabad between seven and nine million each. Additionally, India will likely have more than 42 cities with populations exceeding one million (UN Population Division, 2002). Technological and transport infrastructure measures to accommodate growing motor vehicle activity are certainly important, but worldwide experience has shown that such measures are quickly neutralized, and are therefore futile even in contexts without resource constraints. In India, financial, technological, institutional and administrative resources are far from adequate to accommodate even present levels of motor vehicle activity. But even if there were no resource constraints, transport infrastructure projects such as limited access expressways could result in the displacement of the urban poor, given the high density and poverty levels, and access and mobility could be further compromised for the millions who cannot afford motor vehicles. Given the foregoing, the urban transport challenge in India can be stated as being one of catering for rapidly growing mass mobility needs with a minimum of adverse impacts, particularly for the poor, and at low cost. An important aspect of this challenge is that the bulk of the urban growth is now occurring in medium-

Badami and Haider 6 sized cities, in which the urban transport situation and the resources required to address it are more stressed than in the metropolitan centres. 2. The Importance of Bus Transit An important consequence of low affordability of motor vehicles is that, despite growing motor vehicle ownership and use, public transit (and non-motorized modes) continue to account for significant shares of total trips. The share of public transit (both buses and trains) is as high as 88% in Mumbai; indeed, it is estimated that Mumbai s suburban trains carry 6.4 million passengers daily (MMRDA, 2005). Per capita motor vehicle levels in Mumbai are significantly lower compared to those for Delhi, precisely because of a high degree of public transit (specifically, urban rail transit) provision, despite a comparable per capita income. But even in Delhi, public transit accounts for about 43% of all trips among residential households; further, while M2W vehicle mode shares increased from 1% to 18% of trips by mechanical modes between 1957 and 1994, bus mode shares increased from 22 to 62% over this period (RITES/ORG, 1994; Tiwari, 2002). Notwithstanding the foregoing, of course, motor vehicle ownership and use are growing rapidly, and not merely due to rising incomes and factors such as easy credit for motor vehicle purchase. Many people have been forced to live in the urban periphery, in areas poorly served by transit, because they have been priced out of the land market in the inner cities. Consequently, and also because of increased traffic congestion and the lack of convenient transit service, many people have been forced to purchase and use personal

Badami and Haider 7 motor vehicles, if they can afford them. The motorized mode that people can most easily afford is M2W vehicles. These vehicles cost a fraction of what cars do, offer excellent mobility and easy access to employment and other essential services, and have thus become the preferred mode in a context in which there are few other attractive options. The preponderance of M2W vehicles is therefore hardly surprising. If public transit supply should deteriorate, personal motor vehicle ownership and activity are bound to grow, particularly in terms of M2W vehicles. This prospect is of concern, because the bulk of inuse M2W vehicles are even now powered by two-stroke engines, which tend to have high emission levels. Additionally, M2W vehicles are unstable, and are rendered more so by being used to carry passengers and goods. Note the high share of M2W vehicle users among road fatalities, as discussed. It is important to maintain and enhance public transit service that is accessible, reliable, convenient and affordable in Indian cities, not only to curb personal motor vehicle use, which is imperative given growing urban transport impacts, but because public transit is likely to remain the mainstay of low-income commuters. The poor have no choice but to use public transit if they can afford it, and however inconvenient and time-consuming it may be. They depend on it for their economic survival. Low-cost mobility is particularly important for this group, because while transport expenditure accounted for about 12% of total household income on average, low income households already spend larger shares of their income on transport, thus affecting their expenditure on health, shelter and food (RITES/ORG, 1994; Tiwari, 2002). This situation is only likely to be exacerbated further, given the recent massive fuel price increases. Finally, public transit accounts for

Badami and Haider 8 significantly lower health-critical emissions, fatalities and road space on a passengerkilometre basis than personal motorized modes. These characteristics are extremely important in a context in which transport impacts already overwhelm scarce resources, and expanding road infrastructure is difficult on account of high population densities. Urban rail projects are either being built or are being considered for various Indian cities including Delhi. The first phase of Delhi s Mass Rapid Transit System, measuring 65 kilometres, is expected to be completed soon, at a cost of around US$2.4 billion, 64% of which was financed by a Japanese government loan (Delhi Metro Rail Corporation, 2006). It remains to be seen how effective and financially viable the urban rail projects will be in India, given their high capital costs, the necessity of low fares on account of low per capita incomes, and the tendency of urban rail to attract bus users but not private motor vehicle users. This is especially so in the absence of controls on such vehicles, and the inability, because of constrained resources, to build systems that are both extensive and fine-grained, which would be required to make a significant dent in motor vehicle use in urban forms such as Delhi s that are growing rapidly in all directions. Even if urban rail systems are effective, it will be several years before they become fully operational, and motor vehicle activity will likely grow in the meanwhile. Finally, it is unlikely that urban rail would be economically viable in the medium-sized cities, in which per capita incomes are low and population growth and mobility needs are increasing rapidly. There is therefore a critical need for low cost transit solutions that are appropriate to the realities of the context. Busbased systems have the potential to be such a solution (Mohan et al, 1997). Besides, even if urban rail systems are implemented, efficient feeder buses will likely be needed, in order

Badami and Haider 9 for these systems to be truly effective. Lastly, Indian cities rely predominantly on buses for their public transport needs, with the exception of Mumbai, where rail plays a major role, and are likely to continue to do so over the coming years. For all these reasons, there is a great stake in the vitality of bus transit systems for the purposes of meeting mass mobility needs at low cost. 2.1 The Present State of Bus Transit While urban bus transit is important for meeting rapidly growing mass mobility needs, and curbing personal motor vehicle activity and its impacts at low cost, the state of the India s urban bus transport systems unfortunately leaves much to be desired, in terms of outcomes for users. In a survey of M2W vehicle users conducted by the first author in Delhi in 1997, around 45% of the respondents rated bus service in Delhi to be poor. Further, twice as many reported substantially higher usage five years earlier than presently as those that reported the reverse. The most common bus service problems reported were overcrowding, particularly during peak hours, and the lack of service reliability. Other problems included poor service during off-peak hours, lack of coverage, particularly on the outskirts, unsafe driving by and rude behaviour of personnel, long wait and journey times, and the poor condition of buses. Many respondents noted that while private bus operators (of which more later), because of the profit motive, stopped to pick up passengers at unscheduled stops, causing over-loading and inordinate journey times, and often bribed the police to avoid being fined for overcrowding, publicly owned buses often did not stop even at designated stops, making it extremely difficult to travel with ladies, children and the

Badami and Haider 10 elderly. Lastly, almost every respondent remarked on the sexual harassment female passengers had to deal with on a daily basis. The state of public bus transit appears to have changed very little since (Tiwari, 2002; Pucher et al, 2004). The importance of low-cost mobility has been highlighted, given the socio-economic realities, but bus travel has been simply unaffordable for a significant proportion of urban residents, and despite bus fares having been heavily subsidized until recently. On the outskirts of Delhi, where bus service is poor, daily round-trip bus fares for just one worker in a household with a monthly income of Indian Rupees (INR) 2000 (such households form the bulk of the city s population) would consume 25% or more of this income. For households with lower incomes, the bus fares would be prohibitively expensive (Tiwari, 2002). So, the situation is that, while the mode shares for buses, which form the bulk of public transit are substantial, the quality of urban bus transit is poor. If bus transit shares are high despite the poor service, it is simply because of the lack of viable alternatives for the bulk of users. At the same time, it appears to be out of reach of the urban poor, who are the most affected by rapid motorization. Apart from the inability to meet rapidly growing mass mobility needs, it is likely that the poor urban bus transit situation will cause accelerated growth in private motor vehicle (particularly M2W vehicle) ownership and activity, as noted. M2W vehicles have many advantages with respect to public transit, including their low marginal costs of use, their low door-to-door journey times relative to buses and even cars (Badami, 2005), and their ability to be parked easily (and the difficulty of controlling

Badami and Haider 11 their parking). In view of this prospect, there is an urgent need to understand the causes of the poor condition of urban bus transit in India. 3. Rationale, Scope and Methodological Approach To better understand the state of urban bus transit in India, the paper first reports on an indepth analysis of its financial and operational performance. It then discusses the challenge faced by urban bus transit in meeting mass mobility needs in India, specifically in terms of financial viability, which is essential for effective service delivery over the long term, and affordable service provision, given the high levels of urban poverty. The paper discusses the dilemmas inherent in reconciling financial viability on the one hand and providing affordable services to the masses on the other, and suggests some broad strategies for meeting this difficult challenge. 3.1 Data and Related Issues For our in-depth analysis of the financial and operational performance of urban bus transit in India, we use the data generated by the CIRT (CIRT), and published for the Association of Road Transport Undertakings (ASRTU), for the publicly owned and operated bus transit corporations in Indian cities, over the period 1990-91 to 2000-01. We use three points during this period, 1990-91, 1995-96, and 2000-01, to track the performance of these urban bus transit operations.

Badami and Haider 12 The CIRT data relates to publicly owned bus transit operations ( state transport undertakings, or STUs) across India, which are classified as urban, rural and hilly (CIRT, 1992, 1997 and 2002). As far as the urban classification is concerned, the bus operations in only some Indian cities are characterized as such. Many cities with populations and bus fleets comparable to the ones characterized as urban are not so classified. More importantly for our analysis, data is not provided separately for such cities; rather, the data for these cities is included in that for the transport corporations of the states of which the city operations happen to be a part. So, for Bangalore, which has the fifth largest urban population in India, it is only for 2000-01 that data is available separately for this city; in the previous years, the data for Bangalore are included in that for the Karnataka State Road Transport Corporation (KSRTC). While data for urban operations within the state of Karnataka are available in the previous years, this is not helpful, because this data includes that for other cities in the state, such as Mysore, and it is not possible to extract the data for Bangalore for 1990-91 and 1995-96. The situation is similar for Hyderabad, the capital of the state of Andhra Pradesh, and India s sixth largest city. Data is presented separately for urban and rural bus operations in the state of Andhra Pradesh, at least for some parameters, but it is not possible to extract data specifically related to Hyderabad. The problem is a little different, but the result is the same, for major cities in the state of Tamil Nadu, such as Coimbatore and Madurai. There, the bus transport operations have been constantly re-organized and re-named, making it very difficult to be confident about the data for each of the cities, let alone to make year to year comparisons for them.

Badami and Haider 13 A second important issue related to the CIRT data is that, even the cities that are characterized as urban and for which data is provided separately, vary at each of the points during the 1990-2001 period that we have chosen for our analysis. In 1990-91, the bus operations in the following cities were characterized as urban : Delhi, Mumbai, Chennai, Kolkata, Ahmedabad, Pune, Chandigarh, Pimpri-Chinchwad, Solapur, Kolhapur, Thane, Ludhiana, Amritsar, and Bhavnagar. In 1995-96, bus operations in all of the above cities were characterized as urban, except the last three. And in 2000-01, bus operations in the same cities as in 1995-96 were characterized as urban, with Navi Mumbai and Bangalore added. The upshot of all of this is that a consistent comparison of all urban bus transit operations in India is not possible. But beyond this, even for the cities for which data is available for 1990-91, 1995-96, and 2000-01, data is not available for all parameters of interest. So, for example, in the case of Solapur, no financial performance or capital structure information is available for 1995-96 or 2000-01, and for Pimpri-Chinchwad and Chandigarh, no capital structure information is available for the same two years. Our analysis therefore includes only the following eight cities: Chennai, Delhi, Kolkata, Mumbai, Ahmedabad, Kolhapur, Pune, and Thane. For Pune, the capital structure data used for 1990-91 is actually from 1989-90, and this data has been interpolated for 1995-96, since it is not available for this year. Fortunately, however, these eight cities include India s four largest metropolitan centres (Chennai, Delhi, Kolkata and Mumbai), allowing a consistent comparison of bus transit operations in these cities over the 1990s.

Badami and Haider 14 Table 2 shows data related to buses held, passengers carried and passenger-kilometres from 1990/91 to 2000/01 on twelve of the thirteen publicly owned bus transit operations designated as urban for 2000/01. Only one bus transit operation, namely the Navi Mumbai Municipal Transport, in Mumbai s satellite city of Navi Mumbai, has been omitted; this is a very small operation, with only 126 buses in 2000, and in any case, no data is available for previous years. It should also be noted that we have restricted ourselves to those urban bus transit operations that are publicly owned and operated. In 2000-01, Delhi (and Bangalore) had buses that were under the management of their respective public bus transport corporations, but were privately operated. These operations are excluded from Table 2, and indeed from our discussions in our paper until we address the issue of privatization. INSERT TABLE 2 HERE An analysis of the data in Table 2 follows later in the paper, but for now, suffice it to say that the eight sample cities on which we have chosen to base our analysis (in the shaded portion of Table 2) accounted for nearly 80% or more of buses held and passengers carried by all the bus transit operations characterized as urban in 1990-91, 1995-96 and 2000-01. Also, the four metropolitan centres themselves accounted for more than two-thirds of buses held as well as passengers carried by the urban bus operations in these years. While this shows their pre-dominant position, note that their shares in terms of these parameters have been declining over the years, showing the growing importance of bus operations in the other cities.

Badami and Haider 15 3.2 Analytical Framework Our analysis of the bus transit operations in the eight sample cities is conducted in the following sequence: to explain the user service outcomes, we study quality of service and coverage, which is then sought to be explained based on an examination of fleet strength, and fleet and capacity utilization. This is followed by an examination of the financial performance, in terms of traffic revenues and operating costs, profit and loss, and financial ratios. The financial performance is then further analyzed in terms of bus and labour productivity, and fuel economy. Finally, as part of our concluding section, the ability of the urban bus transit operations to provide quality service over the long term is analyzed based on an examination of their capital structure, and sources of finance including government contributions. The evolution of these operational and financial performance parameters is analyzed over the period from 1990-91 to 2000-01. We have considered the bus transit operations in the four metropolitan centres separately from those in the secondary cities, namely, Ahmedabad, Kolhapur, Pune and Thane, to study the situation in the metropolitan centres, and also to see if there are discernible differences in the performance of bus operations in these centres and the secondary cities. Financial performance parameters are compared over time in terms of constant Indian rupees (INR). This is accomplished by deflating the rupee figures over time by applying the consumer price index for industrial workers (CPI (IW)) for the years 1995-96 and 2000-01 relative to 1990-91, obtained from the Labour Bureau of the Government of India (Ministry

Badami and Haider 16 of Labour and Employment, 2006). As the Economic Survey of India notes, the CPI (IW) is the most commonly used price index for tracking inflation (Ministry of Finance, 1997). 3.3 Choice of Measures We have used the measures, and data in terms of these measures, provided by the CIRT, but in several cases, we have calculated figures for measures that we feel better capture important bus transit objectives. For example, rather than use the CIRT measure of buses held per 100,000 population for assessing accessibility to bus transit, we use capacitykilometres per million population, because this takes into account both the seating plus standing capacity as well as the effective kilometres (that is, kilometres traveled in actually carrying passengers). Of course, even this measure does not address the issue of how these kilometres are distributed, either in space or by, for example, income group. In the case of capacity utilization, whereas data for passengers/bus on road/day is provided by CIRT, and this may be used as a measure of bus loading, % occupation ratio (the ratio of passenger-kilometres to seat-kilometres) and % load factor (the ratio of passengerkilometres to carrying capacity-kilometres), both of which are also provided by CIRT, are better measures for this purpose, since with the first measure, there is no accounting of the number of trips over which the loading is spread out. We use % load factor; this includes both seating and standing capacity, and is therefore a less stringent, but more realistic, measure of bus loading. In the case of financial performance, the CIRT provides data for total operating cost/bus on road/day. While this may be an adequate measure from the

Badami and Haider 17 perspective of the bus operator, a better measure from the point of view of society at large is the total cost per unit of social benefit in terms of passenger-kilometres, and this is the measure which we have calculated. In trying to explain bus transit operational and financial performance, we use various productivity and efficiency measures. As far as the buses themselves are concerned, the CIRT provides data for the percentage of buses held that are on road (and actually carrying passengers) and the total effective kilometres per bus on road as measures of fleet and bus utilization respectively. A much better measure of the actual utilization of buses (and therefore of the extent to which buses being off-road is minimized) would be the total effective kilometres per buses held. But even better from the perspective of societal benefit derived from available resources would be passengerkilometres per bus held, which is the measure which we have calculated. Similarly, and for the same reason, whereas the CIRT provides fleet fuel efficiency data in terms of kilometres per litre, we have calculated fuel efficiency in terms of passenger-kilometres per litre. Lastly, in terms of labour productivity, the CIRT provides data for bus-staff ratio, which is the total number of staff per bus on road, and total effective kilometres per manday paid for. While the second measure is better than the first in terms of labour productivity from the perspective of the bus operator, passenger-kilometres per employee per day is the best measure in our judgment from the point of view of society, and this is the measure we have calculated in our paper.

Badami and Haider 18 4. Analysis of Data 4.1 Bus Transit Provision Table 2 shows that the number of buses held by the publicly owned and operated bus transit providers increased fairly steadily in the four secondary cities over the 1990s. On the other hand, the number of buses held in the four metropolitan centres remained flat, and in fact dropped between 1990-91 and 1995-96. Of particular note is the steady reduction in buses held by Delhi Transport Corporation (DTC), the publicly owned operator in the Indian capital (and the most motorized Indian city). Indeed, Delhi s public bus fleet was fully 33% lower in 2000-01 compared to ten years earlier. However, it should be pointed out that there were in fact considerably more buses held by private operators (but under DTC management) in 2000-01. When these are included, Delhi in fact had the largest fleet (of 6028 buses) of any Indian city in this year, and 37% more buses than it did in 1990-91. Further, Delhi has had a large number of buses owned and operated by private operators (Table 1). Unfortunately, no data relating to operational or financial performance is available for these operations. As already pointed out, a much better measure of bus transit supply than buses held is carrying capacity-kilometres per capita. This measure is calculated for the various cities, taking into account only the publicly owned and operated bus transit providers, in Table 3, and graphically represented, for the four metropolitan centres and four secondary cities, in Figure 1. Interestingly, bus transit supply in terms of this measure was consistently higher

Badami and Haider 19 on average in the former than in the latter cities. This is despite the much higher population levels in the metropolitan centres, showing the significantly higher bus transit provision in these cities. At the same time, though, capacity-kilometres per capita steadily dropped over the decade in the metropolitan centres (showing the potential for over-loading), while marginally increasing in the four secondary cities. In these cities, carrying capacity-kilometres actually increased significantly, but population growth, at close to 5% per annum, was nearly as rapid. The decline in capacity provision on a per capita basis is particularly striking in the case of Delhi (Table 3). If the private bus operations under DTC management were included for 2000-01, the situation would look more respectable, but still inferior to that in 1990-91, and the averages for the four metropolitan and all eight sample cities would improve only marginally. It is also worth noting that, while Chennai consistently had the highest capacity-kilometres per capita, Kolkata stands out among the metropolitan centres for its significantly low capacity provision, which was on par of that in the smaller secondary cities. In Mumbai s favour, it should be pointed out that, while capacity provision per capita was much lower than in Chennai (and flat over the decade), rail, and not buses, account for the bulk of public transit travel. 4.2 Service Utilization We have so far focused on the provision of bus capacity-kilometres in our sample cities. We now discuss the extent to which this capacity is utilized by the public, and the quality of

Badami and Haider 20 service. Remarkably, the number of passengers carried by public bus transit actually dropped in the four metropolitan centres over the 1990s (Table 2). The only exception was Chennai, which showed a steady increase over the decade. As for Delhi, the number of passengers carried by DTC dropped precipitously in 1995-96 to about 36% of that in 1990-91; although there was an improvement in the second half of the decade, DTC carried only 55% of the passengers in 2000-01 that it did in 1990-91. Even when the private bus operations under DTC management are included for 2000-01, the number of passengers carried improved only marginally over 1990-91. While the passengers carried in the secondary cities increased over the first half of the decade, they remained flat over the decade, with significant declines between 1995-96 and 2000-01 in Pune and Kolhapur. The trend in passenger-kilometres is similar to that for passengers carried in the metropolitan centres, average bus transit journey lengths having remained fairly constant (about 9 km) over the decade. In the secondary cities, average journey lengths increased (from 6.4 to 8.4 km) over the decade; this, along with the trend in passengers carried caused passengerkilometres to increase over the decade, but decline in its second half (Table 3). This of course means that, although capacity-kilometres per capita declined significantly (suggesting potential for over-loading), and contrary to the common belief that buses are becoming more crowded, the opposite was the case (Figure 3). The percentage load factor in fact reduced significantly over the decade. In the secondary cities, recall that passengerkilometres went up somewhat, but obviously this increase was more than made up by the increased capacity. In the metropolitan centres, the load factor was close to 90% in 1990-91, showing fairly heavy bus loadings, but the situation improved, despite declining

Badami and Haider 21 capacity, because of reduced passenger travel. Overall, capacity provision per person was much higher in the metropolitan centres, and this capacity was much better used, and buses were loaded more, than in the secondary cities. INSERT FIGURES 1 AND 2 HERE INSERT TABLE 3 HERE INSERT FIGURE 3 HERE The decline, or at best, no increase, in passengers carried (Figure 2) is certainly very interesting, given the rapid population growth, and the fact that the number of buses, and capacity-kilometres, held fairly steady overall, and actually increased dramatically in the secondary cities. This is an issue we explore in our conclusions. 4.3 Financial Performance Bus operations in the metropolitan centres and secondary cities had losses even before tax right through the 1990s, except for Thane, which showed a small profit, and Kolhapur, which did so in 1990-91 and 2000-01 (Table 4). Delhi s publicly owned bus corporation alone accounted for 60-70% of the overall losses in 1990-91 and 1995-96, with the situation improving considerably at the end of the decade, very likely due to the private operation of a significant portion of the bus fleet, to be discussed later. In 2000-01,

Badami and Haider 22 Mumbai s was the worst performing bus operation, with losses accounting for a third of the total losses in all eight cities. Overall, the pre-tax financial performance in the metropolitan centres improved over the decade after deteriorating in the first half, largely due to the situation in Delhi, whereas that in the secondary cities declined significantly, largely due to the operation in Ahmedabad. The consistent losses suffered by the operations in both the metropolitan centres and secondary cities obviously show that the total costs even before taxes consistently exceeded total revenues (mainly from passenger fares but also other sources). Interestingly, while total revenues in the four metropolitan centres increased steadily in real terms, despite a steady reduction in passenger-kilometres (discussed in Section 4.2 above), suggesting a considerable real increase in passenger fares, pre-tax costs increased less rapidly than revenues over the decade, leading to the improved financial performance, albeit a loss, at the end of the decade. In the secondary cities, total revenues increased at the same rate as in the metropolitan centres, but pre-tax costs increased much more rapidly than revenues, very likely due to the significant increase in bus capacity, leading to the considerably increased losses at the end of the decade as discussed. INSERT TABLE 4 HERE

Badami and Haider 23 4.3.1 Traffic Revenue and Operating Cost per Passenger-kilometre Since the passenger-kilometre levels varied over the period of analysis, a proper comparison of financial performance across bus operations and over time would require that the revenues and costs be calculated per passenger-kilometre and with inflation adjusted, as already discussed. Also, to measure the financial performance of bus operations strictly in relation to carrying passengers, which after all is their primary function, it is desirable to compare traffic (as opposed to total) revenues with operating costs, that is, total costs less taxes as well as interest payments. These calculations, based on the data in Tables 2 and 4 are presented in Figure 4. Traffic revenue per passenger-kilometre (which is in fact the average fare per passenger per kilometre) increased steadily, by as much as 68% in real terms, over the decade in the metropolitan centres. The fares were the lowest overall in Chennai and Kolkata, and the highest (about 2.5 times higher) in Mumbai; they increased least rapidly in Chennai, and by far the most rapidly (by 139%) in Delhi, which, it should be pointed out, started the decade with the lowest fare of all the metropolitan centres. Operating costs per passenger-kilometre were of course consistently higher than traffic revenues in all of the metropolitan centres, but increased at a lower rate (52% in real terms) than traffic revenues, over the decade. As with fares, operating costs per passenger-kilometre were by far the lowest in Chennai, and the highest in Mumbai (about 2.5 times higher on average). Operating costs increased most rapidly (by 91%) in Delhi among the metropolitan centres.

Badami and Haider 24 In the secondary cities, the average fares in terms of traffic revenue per passenger-kilometre were consistently higher than in the metropolitan centres, although the fares in Mumbai were in fact the highest of all eight sample cities. Average fares in the secondary cities actually declined in the first half of the decade, and then increased at about the same rate as in the metropolitan centres in the second half of the decade, as shown in Figure 4. But fares in Thane, the smallest of the eight bus operations, fell continuously over the decade. Operating costs per passenger-kilometre also fell in the first half of the decade, and then increased more rapidly than traffic revenues, in the second half. This differential was particularly pronounced for Kolhapur, for which the former increased by as much as 162% over the decade, as against only a 75% increase in traffic revenues per passenger-kilometre. The trends in operating costs per passenger-kilometre in the secondary cities need to be seen in the context of the significantly increased bus capacity, and the fact that passengerkilometres increased significantly in the first half, and then fell nearly as significantly in the second half of the decade. 4.3.2 Operating Cost Recovery through Fares A particularly useful measure of the financial effectiveness with which bus operations carry passengers is the extent to which operating costs are recovered through passenger fares. This measure, expressed as the percentage of operating costs accounted for by traffic revenue, is plotted in Figure 5. The first point to note is that a much higher share of operating costs was recovered through fares in the secondary cities than in the metropolitan centres. At the same time, though, while the traffic revenue share of operating cost

Badami and Haider 25 improved steadily in the metropolitan centres, it dropped markedly in the second half of the decade in the secondary cities; this trend is related to the fact that operating costs increased much more rapidly than traffic revenues in these cities, as discussed. Chennai had the highest rate of cost recovery through fares of all metropolitan centre operations (around 82% on average), followed closely by Mumbai (around 78%). Chennai s achievement in this regard is notable because it also had the lowest fares consistently of all metropolitan centres. Kolkata s fares were as low as Chennai s, but unlike Chennai, the city consistently had the lowest recovery (only 42%), because of significantly higher operating costs. Delhi had the largest increase in the rate of recovery (from 54 to 68% over the decade). Of the secondary cities, Thane, Pune and Kolhapur all had recovery rates of above 100% in 1990-91; while in Thane s case, this figure was 100% at the end of the decade, Kolhapur s dropped steeply to 73%. This happened because, although Kolhapur had the highest average fares (along with Mumbai s, of about 28 paise per passenger-kilometre) in 2000/01, it had by far the highest operating cost of any operation (38, as compared to 37 and only 13 paise per passenger-kilometre in Mumbai and Chennai respectively). INSERT FIGURES 4, 5, 6, 7 HERE 4.3.3 Breakdown of Bus Transit Costs An analysis of the various cost components (Figures 6 and 7) reveals, not surprisingly, that personnel costs are by far the most important, followed by diesel fuel costs. Further, both of these components registered a sharp increase in percentage share since 1995-96,

Badami and Haider 26 particularly in the metropolitan centres. These two components alone accounted for 85% of total costs (and 89% of operating costs, that is, total costs less taxes and interest) in these centres in 2000-01. Personnel costs accounted for a much higher share of total costs in the metropolitan centres than in the secondary cities in 2000-01; in the case of diesel fuel and other material costs, the situation was reversed. Of the metropolitan centres, Chennai appears to have held personnel costs steadily at around 52% of total costs through the decade, no doubt contributing to the low operating costs in this city, discussed earlier. On the other hand, the percentage share of personnel costs increased sharply in the other three metropolitan centres; in Mumbai, the increase was from 47% in 1990-91 to as high as 72% of total costs in 2000-01. In the last part of this section on financial performance, we consider the three most important components of bus operating costs, namely those relating to personnel, diesel fuel and other materials, on a passenger-kilometre basis (Figure 8). First, personnel costs are the highest, followed by diesel and other materials, in line with their percentage shares. Second, while there was an increase between 1995-96 and 2000-01 for personnel and diesel costs, the cost of other materials actually fell. The costs per passenger-kilometre were consistently higher for the secondary cities, except for personnel costs since 1995-96. This suggests increasing returns to scale with regard to operating costs. Whereas personnel cost increases on a passenger-kilometre basis were significantly higher over the decade in the metropolitan centre operations, diesel costs per passenger-kilometre increased slightly more rapidly in the secondary city operations. Since passenger-kilometres declined steadily over the decade in the metropolitan centres, and were only marginally higher at the end of the

Badami and Haider 27 decade than at the beginning for the secondary city operations, it may be concluded that while the secondary city operations were not very much better than their metropolitan city counterparts in controlling personnel costs, they were considerably less successful in controlling diesel costs. Chennai had the lowest diesel as well as personnel costs per passenger-kilometre of all the cities. Mumbai had by far the highest personnel costs per passenger-kilometre of all the cities in 2000-01 more than three times that of Chennai s in the same year, and more than twice its own cost in 1995-96. 4.4 Bus and Labour Productivity, and Fuel Efficiency The increase in operating costs, particularly since 1995-96, is likely due to increases in per unit costs; however, productivity is also an important factor. For example, fuel costs per passenger-kilometre are the product of costs per litre of fuel, and fuel consumption in litres per passenger-kilometre. Bus operations have no control over unit diesel costs per litre, but do have control over fuel consumption per passenger-kilometre, by way of bus technology, maintenance, efficient capacity utilization and route management, careful driving and so forth. In this section, we look at the effectiveness with which bus, labour and fuel assets have been employed, and changes over the 1990s. As discussed in our Methodological Approach section, we have calculated 10 3 passenger-kilometres per bus held per day as a measure of the effectiveness with which bus fleets are utilized to provide service. Fleet fuel efficiency has been calculated in terms of passenger-kilometres per litre, instead of using the kilometres per litre (Km/L) data provided by CIRT. Lastly, we have calculated labour productivity in terms of passenger-kilometres per employee per day, rather than using the

Badami and Haider 28 various measures for which CIRT provide data. Apart from capturing the efficiency with which bus operations employ various resources, these measures have the advantage, by not incorporating monetary measures, of being able to be compared over time without having to be corrected for inflation. 4.4.1 Bus Productivity Bus productivity in passenger-kilometres per bus held declined for the metropolitan operations by 17% over the decade (Figure 9). Since there was effectively no change in buses held, this decline was caused by a reduction by the same extent in passengerkilometres in large part due to a steady drop in passengers carried, and only to a small extent in revenue generating kilometres, as noted earlier. In the case of the secondary cities, the decline in this measure was about 11% over the decade, after having in fact increased by about 8% in its first half. This happened despite an increase of 27% in passengerkilometres over the decade, but this increase was offset by an even larger growth in the buses held (indeed, as noted earlier, though passenger-kilometres in these cities increased over the decade, passengers carried at the end of that decade actually fell slightly). The significantly lower bus productivity in the secondary cities as compared to the metropolitan centres was of course due to the significantly lower load factor in the former (Figure 4).

Badami and Haider 29 4.4.2 Labour Productivity As for labour productivity, there was a 7% decline over the decade for the metropolitan centre operations. For the secondary city operations, labour productivity declined only marginally over the decade, after having increased substantially until 1995-96 (Figure 9). Although passenger-kilometres declined steadily in the metropolitan centres, the number of employees also dropped, albeit to a lesser extent over the same time, with much of the reduction happening in Delhi. In the secondary cities, employees increased by the same proportion as passenger-kilometres. As in the case of bus productivity, labour productivity was significantly and consistently lower in the secondary cities than in the metropolitan centres. In 2000-01, it was about 30% lower. While passenger-kilometres were of course much higher (by a factor of 7.6) in the metropolitan centres, these centres had only 5.4 times the employees as the secondary city operations, suggesting scale effects as regards labour resources. However, while the secondary cities had far more employees per passenger-kilometre, these cities actually had fewer employees per bus held compared to the metropolitan centres (Table 5). This not only shows how notions of productivity are sensitive to the choice of measure, but also that the secondary cities have a lower labour productivity in terms of passenger-kilometres per employee per day not so much because they have far more employees than they need to vis-à-vis their metropolitan counterparts, but rather that they have been unable to get enough passengers on to the buses that they do operate.

Badami and Haider 30 Despite their significantly higher labour productivity, recall that the metropolitan centres have higher personnel costs per passenger-kilometre than the secondary cities. This is simply because of the much higher labour costs in the former (particularly in Mumbai). Indeed, daily employee salaries, which were only marginally higher than in the secondary cities in 1990-91, increased far more rapidly, and were as much as 53% higher in 2000-01, in the metropolitan centres (and in Mumbai, were more than twice the average salary in the secondary cities). INSERT FIGURES 8 AND 9 HERE INSERT TABLE 5 HERE 4.4.3 Fleet Fuel Efficiency Finally, fleet fuel efficiency in terms of passenger-kilometres per litre declined by about 20% over the decade in the metropolitan centres and secondary cities. The decline in the former cities may be attributed to the fact that, while the fleet size and kilometres remained substantially the same, passengers and passenger-kilometres declined significantly. In the secondary cities, both the fleet size and kilometres, and consequently fuel consumption, increased substantially, and to a much greater extent than passenger-kilometres. As with the other productivity measures, passenger-kilometres per litre was significantly lower in the secondary as compared to the metropolitan cities. This is because, whereas the metropolitan centre fleets had only five times the buses, and operated as many times the revenue