RAIL MASS TRANSIT IN DEVELOPING CITIES: THE TRRL STUDY J M THOMSON Senior Consultant, Halcrow, Fox and Associates R J ALLPORT Director, Haicrow, Fox and Associates P R FOURACRE Principal Scientif ic of ficer, overseas Unit, Transport and Road Research Laboratory SYNOPSIS. Many Third World cities have recently built metros, usually encouraged by European manufacturers and their governments, but frowned on - because of their enormous cost - by the World Bank and some disinterested transport planners. How have these metros fared? Were they justified in either financial or economic terms? Or are metros really a luxury that poor cities should forgo? How can an aid agency, or a city government, come to a reasonable appraisal of a metro project without the cost and delay of a large feasibility study? The OverseasDevelopment Association recently sponsored a research study into these questions, and this paper summarises the findings. The Study Team went to 21 developing cities and collected data which was analysed on a strategic transport evaluation model. They enquired into the reasons why people wanted metros and noted a widespread misconception that the metro would cure traffic congestion. The Study found much to praise in the engineering and operation of the Third World metros, but much to criticise in their planning and financial management. It concluded that metros - in developing cities at least - cannot be finandially viable but can give good economic returns in the right conditions, which it endeavoured to spell out in broad terms. The Study noted the strong connection between the metro and the growth of the city centre; it concluded that very large cities must either build metros or see the growth of their centres frustrated by inaccessibility, with corresponding growth diverted to suburban centres. BACKGROUND 1. Mexico City opened its metro just over 20 years ago, following the mould of European and Japanese cities. Since then over 20 cities in the developing world have pursued a similar course and many others are planning to do so. This trend has been eagerly supported by some industrialised countries (France, Germany, Sweden, Japan, Belgium and others) but f iercely discouraged by some aid agencies and some transport planners. The World Bank, in its Urban Transport Policy Study (198.6), pointed to the high construction and operating costs, cost over-runs, over-estimates of demand, inadequate revenues and attempts to suppress bus competition, commonly associated with metro projects. The Bank generally preferred to see money spent on more cost-effective solutions, in particular, improvements to the capacity and operation of existing infrastructure. 2. This 'soft path' approach was widely criticised (by Dunbar and Rapp, 1986, and Henry, 1987, for example) and debated (by Ridley, 1986, and Rimmer, 1988, for example). The approach was alleged to overlook the wider and longer-term benefits of metros and to overrate th e ability of less costly means of transport to meet the needs of developing cities, many of whom continue to clamour for metros, whether light or heavy, above or below ground. The UK Overseas Development Administration (ODA) has been approached about six metro projects in the past two years. But there has been little research on the performance and impact of such projects. In mid-1987, therefore, the ODA, with the encouragement of the World Bank, commissioned a study by their transport advisers, the Overseas Unit of the Transport and Road Research Laboratory, who appointed Halcrow Fox and Associates to provide most of the staff input. 3. The aims of the-study were to: - review the experience of planning, building and operating rail mass transit systems in developing cities; - establishing a quick and simple method of appraising metro projects in developing cities; - identify the conditions under which investment in such systems might be justified; - recommend a role for aid agencies. 4. The work divided mainly into three parts: case studies of 21 developing cities with metros in operation or in prospect; establishment of a strategic traffic and evaluation model; and application of the model to some of the case study cities. The case studies provided data with which to furnish the model, and the model provided the tool for evaluating metros, either operational or projected. The application of the model to the cities completed the comparative analysis on which some general conclusions about metros in developing cities could be drawn. This paper summarises the findings and conclusions of the study. * ~~CASE STUDY CITIES 5. Case studies were made in 21 developing cities covering a wide geographic, demographic and economic range. All contained more than a million people and most 'more than 3 million, and they included many of the world's largest cities. They included Singapore and Hong Kong, which are no longer developing cities but were classed as such when they started planning their metros. Metros are in operation in 13 of the cities, are under construction in two and are planned in the other 6. One city, Tunis, has a light rapid transit system. 6. Figure 1 compares the cities in terms of population and income/head. The total income curves (i.e. population x income/head) show that all the cities with incomes above US$ 15 bn have a metro while most below US$5 bn do not. Transport Policies 7. A review of the cities'. transport policies was made in order to see what had been done, apart from building metros, to resolve their transport problems, and this is summarized in Table 1. Traffic management was universal; most of the cities had area traffic control, gyratories, lane markings, channelization, etc. Parking restrictions were common but generally not well enforced. Parking charges were sometimes imposed in central areas, but at very low levels. Apart from Singapore, there was no serious attempt to restrain the user of cars, but Hong Kong, and to a lesser extent some other cities, imposed high motor taxes in order to restrain car ownership. L-0n~ -0 0 .0 E 0 C (0 C~~~~ 0~~~~~~~~ Id .2~~~~~~~~ *CC 0~~~~~~~C 0~~ (99u6oiLeinSdoU011M PA 1743.1 c 0 U W 0 0 ,UC c 0) 0) 0 0 z 0 0 C,) E CC 0 Z7 0 .U 0 CD a. 0 Co U) 0(N Table 1 Transport policies in case study cities No of cities adopting measures All Cities with cities metros (21) (5 Policy Measures Traffic engineering 21 14 Bus priorities 5* 5 Traffic restraint: parking 3 3 other 2 2 Bus ownership: public 5 3 private 8 7 both 8 5 Paratransit 13 9 Bus fares control 20 13 Bus quality control 7 4 Investment Measures Busways 4 3 Light rail, at-grade 6 5 Metro, gradeseparated 14 14 Suburban rail, upgrading 13 9 Major highway construction 17 12 *a f urther 6 cities introduced bus priorities,,,-' which were ineffective **includes one city, Tunis, with LRT 8.. In 5 cities private bus operators were not permitted; 8 cities let the public and private sector operate side by side, while another 8 cities relied entirely on the private sector. Paratransit was found in 13 cities add often made a major contribution. Only 5 cities operated effective bus priorities, mostly in the form of bus lanes, which several other cities had also introduced but not enforced. 9. All the cities were investing a high proportion of their resources on transport infrastructure. New roads are essential in a fastexpanding city, and 17 cities were engaged in major road construction. Busways were a f eature in 4 cities. .Most of the cities were also investing in rail, of course, since the sample was selected because of that; but six of them had opted for light rail and many, in addition to metro building, were also upgrading suburban railways, with mixed success. Reasons for Building~ Metros0 10. The reasons for building a metro generally become obscure during the long, confused debate that invariably precedes it. Yet it is important to know why people want it. The findings of the study, which are summarised in Table 2, represent the arguments given in official reports and by officials questioned. The most persistent reason was to improve the quality of public transport; in every city the bus services were slow, crowded and uncomfortable. These conditions were invariably blamed in part on the overloading of the bus service - "the buses can't cope" - which was closely connected with another reason, namely, to increase the capacity of public transport in order to carry the forecast volume of passengers. Thus the dominant reason for the metro may be summed up as the need for a bigger and better public transport system. This was especially strong in Latin America. 11. The next most prominent reason was to relieve traffic congestion, by displacing buses and attracting motorists to the metro. For many people, especially in the Far East, the metro was the answer to traffic congestion, and this was its prime justification. 12. The corollary of improving public transport and easing traffic congestion was that people and jobs could be attracted to the areas served by the metro; thus the metro and its stations could be located to support plans for large-scale residential and comm~ercial development. This aspect was 0 Table 2 Reasons given for building metros: case study cities *The study team gave a score of 1 for a 'very important' reason, 0.5 for an 'important' reason especially important in Istanbul, Porto Alegre, Hong Kong and Singapore. 13. There were many other contributory arguments. By reducing the volume of traffic, the metro would also.. reduce accidents, pollution and energy consumption. In Korea, Brazil and India, it was seen as a stimulus to the domestic engineering industry. 14. An alternative, and more rational, approach was to consider the financial and economic returns on the investment. However costly, an investment which made a profit 'thereby proved its justification; and most of the metros were predicted to do just that. Why should government oppose a railway that was going to repay its costs, give a much better service to millions of passengers and at the same time relieve traffic congestion, curb accidents, ease pollution and save fuel? 15. A more sophisticated approach was to consider the economic return but this would normally have No of Reason cities Point citing score* reason It will raise the quality of public transport 17 15 It will relieve traffic congestion 14 11.5 It is needed to carry forecast volume of passengers 9 8.5 It will promote land use policy 10 8.5 It will be good for the environment 8 5 It will support local industry 7 4.5 It will support energy policy 5 4 It will save road accidents 1 0.5 it Will be financially viable 13 9 It will be economically viable 9 5 been considered superfluous if the project was expected to be financially viable. In 9 cities the expected economic return had been calculated and found satisfactory, but it appears to have made little impression on the authorities, who were more interested in the financial return. PLANNING, CONSTRUCTION AND OPERATION OF METROS Characteristics 16. The earlier systems - in Mexico City, Seoul, Hong Kong and Sao Paulo - were "full" metros in the European tradition, i.e. heavy systems largely underground and often called MRT (mass rapid transit). more recent lines include LRT (light rapid transit) in Manila, Tunis and Istanbul, and also as additions to the heavier systems in Mexico City and Hong Kong. There are now networks in Mexico City, Seoul and Hong Kong, but the other cities are limited to one or two, usually short, radial lines. Table 3 summarises the types of metro built or planned. 17. It makes little difference to the passengers whether they can travel by MRT or LRT; the cars are similar and so are the speeds. The table shows a range of journey speeds between 19 and 42 k/h, the main determinant in variation being station spacing; the high speed in Porto Alegre is assocated with an average station spacing of 1.9 kin, while the low speed in Tunis results from a station 'spacing of 0.8 km and an extremely winding alignment. 18. The systems vary in capacity from 14,000 to 75, 000 passengers per hour per direction (p/h/d) . The LRT systems are designed for 14,000-28,000 p/h/d as compared with 2*7,000-75.,000 p/h/d for the MRT systems. At the higher capacities, train headways are 2 minutes or slightly less. Planningr 19. Most of the faults with the metros studied arose during their planning. Feasibility studies were sometimes not made, or were made badly, or were irrelevant to the system eventually built. Poor alignments were chosen in eight cities, for bad planning reasons with sometimes distrastrous and irreversible results. The importance of fares structure and integration of bus and metro fares was generally not recognised. In 13 cities it was assumed that bus competition could and would be eliminated, but only two cities achieved it. Patronage forecasts were too high and cost estimates too low in almost every case. The vital programmes C),.. C-C 0 n(I 140 l- ~ CO I0'l) ~% 0 (1 ~~~(1~(1~'30 C'. C' C'.- (I Cl cl. o oCO 0 0 0 0 0 0 0 0 0 00 000 00 0 0 00 0 00 0 0 ,. 0 00 0 C '. 0 C'.0 0 0C'00 C o~a% tnco cO W).,3 '3 CtA 0 00 CD kocv-) - C4 N0N i w N r ('0 22N m .-4 l 40 (1C4(' ('L4 C~I 00'3ICl- C414 nN3 0 -1.4 .-4,-40 m-4.-4.4 .4 .4 -4 0 -4 0 0 -4 -4 -4 0 14~ ~~~~- V~~~~~~~~~~4 0.-LA ~-0 ~ 0 IOO C~0A.- (4W) 00 V d ~~~~~~~~~~~~~~~~~~cc I :2 00 . C)(0 ,S>. ,-4 ci~ ~ ~ ~~~~~~4cjc H 0 C~~~ ~~~~~~tTI ~ ~ ~ A caU 2 m 1 W ii 1 0 1 cl HZ2 ,for funding and property acquisition were sometimes not adequately prepared before construction started. Construction 20. Almost without exception the final engineering of the metros was good. But in about half the cases, important changes were made to the original plan, usually involving higher expenditure. Three-quarters of the projects were late in completion; the average construction time per line was over 7 years, implying an overrun of more than 2 years. Only the Far Eastern cities of Hong Kong, Singapore and Seoul kept to schedule. Other cities suf fered f rom unf oreseen service diversions, f ailure to acquire land soon enough, shortages of materials or funds, or arguments over the alignment. 21. For these reasons, as well as unrealistic costing, the capital costs exceeded budget, usually by a large proportion, in all but three cities. operations .22. Nearly all the systems were well operated and gave good service. But operating costs were consistently higher than predicted (in real terms) reflecting consistently poor estimation. Only one system (Hong Kong) had any chance of recovering its capital cost, and that only because of its unique population density and of the large profits made on property development above and near the metro. All other systems were subsidized; at least seven did not cover annual operating costs including depreciation. 23. Every system was planned on the presumption that bus services would be "integrated" - which usually meant "not. allowed to compete with the metro" - and that fares would be ajusted to enable the metro to attract the passengers who could benefit from it. But, rightly or wrongly, only Mexico City and Tunis have actually removed bus competition and only Sao Paulo and Porte Alegre have effectively integrated fares, (in both cases resulting in large deficits). 24.. Despite subsidized fares, patronage fell short of prediction in two-thirds of the cases, owing partly to failure to compete with buses and partly to unrealistic forecasts. PREFEASIBILITY MODEL 25. A key part of the Study was to produce a strategic model that could give a quick assessment of the impact of a proposed metro line. This was described as a prefeasibility model and was intended .to show whether a proposal merited the lengthy and costly investigation involved in a full feasibility study. Modified versions of Halcrow, Fox and Associate'1s standard traf fic model programs were used to form a model with minimal data requirements; it was designed: - to require little data of a detailed nature; - to use data normally available; and - where necessary data are unavailable, to use default values obtained from analysis of other cities. The model can thus be applied in any city within a short period of time. The more data available, the better will be the model, but holes in the data - other than absolutely basic facts - will not scuttle the model. The Study team applied it to 10 cities and applied a simplified version to three other cities. 26. The model was designed specifically for evaluation purposes and comprises two separate parts: a land use and traffic model, and an evaluation model. The former contains typically 20 zones, with their respective populations, average incomes and levels of vehicle ownership, and a rudimentary highway and public transport network, all focused on the main radial corridors where metro projects might be justifiable. The model was designed for strong manual control because it deals with the most complex and sensitive traffic situation, namely the centre of large cities where the capabilities of the roads and bus systems are often stretched to the limit. The traffic model produces estimates of daily passenger flows by mode: and average journey speeds and times, with and without the metro, for two selected years, and calculates the time savings resulting from the metro. 27. The evaluation model extracts the required traffic results, amalgamates them into simple corridor form, introduces costs and annual streams, calculates costs and benefits over a period of up to 50 years, and estimates the net present value and internal~ rate of return. It also analyses the costs and benefits by type (time savings to different groups, operating costs by mode, capital costs, consumers' and producers surplus) and by class of beneficiary (bus users, car/motorcycle- users, *generated passengers, bus and metro operators, and government). There is also a probability analysis which tests the sensitivity of the result to the principal variables in the model, assesses the probability of alternative values for each of them, and calculates a probability distribution of the expected economic rate of return. This helps to compensate for the uncertainties of the data and the predictions., IMPACT OF METROS Physical 28. The typical impact of the metro was found to be curiously at odds with the intentions of. its planners'. The metro was expected to relieve ,congestion, but nowhere has this been a notable *result. Consequently there has been little positive impact on, either the environment or energy consumption. 'But the prime intention was to provide better public transport. and more of it, and this aim has been dramatically realised, with two main results. First, the middle- and low~-income masses have been spared a great deal of wasted time and discomfort; in some cities, however, metro fares - and lack of integrated fares - have effectively excluded the low-income groups' and reserved the metro for the middle classes. 29. Secondly, in the biggest cities, the metro .has enabled the growth of activity in the city centre to continue, unchecked by declining accessibility. With a few exceptions, little -development has been positively promoted by government, or by the private sector, to exploit the --metro facili ties; the real impact is permissive, in0 that the metro permits the city centre 'to develop freely in response to market forces. Without the metro, the influx of workers, customers and clients would be checked, and the f rustrated growth would occur elsewhere, most often in sub-centres. This sort of enforced decentralisation is sometimes :considered to have attractions. it ' also entails costs in terms of continuing, chronic overcrowding of buses on the main radials, inconvenience to the .organisations shut out of the centre, and additional transport costs caused by cross-city traffic generated by the location pattern, which would encourage the growth of private transport. Financial 30. Financial performance has rarely matched expectations: both capital and operating costs have generally- exceeded, previous estimates,2 Qf-tenl by a large margin,, wh.,ile patronage and. revenues have .usually fallen: short. As already mentioned,- ,only ,Hong Kong shows .any. possibility. of making, a respectable return on the capital invested;, Seoul. is the only other metro to earn any contribution at all towards its capital cost. This is not to.-say that metros ought, to repay their capital cost. Profit ma~ximization (or loss minimization) ,is n ot the policy of most governments. The point is that most governments were led to believe that their, metros would be f inancially viable, which they certainly are not. ..E.conomic *31. The economic evaluations made by the.Study, using its prefeasibility model, assessed -the impact -past and future - of the metros relative to a dominimum scenario. It was, assumed that. the,, road capacity of the- af fected corridors. would increaseby 20%, reflecting progress in traffic management, ~.driving standards,,. vehicle performance, and road .standards. Thirteen metros. were evaluated, .-- ten-of them b~y the. full model, and the other three-without the traffic model--. The benefits- consist -of time sayings to metro passengers, bus passengers and -.other traffic,. vehicle operating, cost -savings, attributed "consumers' surplus', from .-generated ~traffic, anda proxy "bonus" representing the superior comfort of the metro relative-to the bus. -, Time savings---were valued at 40% of the average rate .of- earnings.' 32. Table 4 summarises the results. Three metros -- showed economic rates of return less than 10%, .three between 10% and 12%-, four between 12% 'and -15%, and three above 15%. Not included, however, were Lines -4-9 in Mexico city. which are poorly utilized and -.would produce a low rate; and some -lines which were *included, e.g. Line 2 -in Rio, Line 2 in Santiago, -band probably Line 3m'i Seoul and Line N in Tunis, .would also ~show low rates if evaluated, separately ..from. the other lines -in the same cities. --O-f the -three results above 15%, two are for Singapore and Hong Kong, which are no longer developing countries, -and the third,, for Cairo, is an exceptional case -where the metro results from the. linking and ,upgrading 'of two busy suburban lines. Thus with-the single exception ofCairo, no metro in a developing city is expected to produce an economic return greater than 15%. ..Table 4. Econo~mic Evaluation of Metros in Case Study Cities Notes: 1) capital costs tested; are in 1986 dollars and refer only to the lines 2 ) base year is fir-st full year of operation; evaluation year, _Js 20 years after completion of investment; i e Cairo 1988/2008 Calcutta 19841/2010; Hong Kong 1981/2007; Manila 1986/2005; Mexico' City 1970/2000; Porto Alegre 1985/2004; Pusan 1988/2007; Rio 1979/2010; Santiago 1.976/2000; Sao Paulo 1976/2008; Seoul 1975/2005; Singapore 1988/2009; Tunis 1986/2009; 3) base year estimates are only for the corridor~s) with metro *in the base year; evaluation year usually includes additional lines; 4) base trips are the estimated trips in the corridors with a metro before opening of metro; trips in evaluation year are prediction; 5) evaluation includes a period of 30 years after completion of investment, or at most 50 years after start of investment; ic the evaluation period extends beyond the so-called evaluation year. Trips/day in. metro corridors (OO0s) EIRR Capital- -- -. city cost bsyer evaluation year: $mn (without. total by metro) trips I metro Cairo 526 830 I 4963 2110 16.8 Calcutta j 684 736 992 400 2.8 Hong Kong 5051 2059 9121 3489 18.5 Manila 563 2250 3309 853 11.4 Mexico City 1974 .4056 .10184 6003 11.4 Porto Alegre 278 567 850 375 8.9 Pusan 680 2273 3616 664 14.2 Rio de Janeiro 2219 2100 4299 1700 f7.1 Santiago 940 2302 2700 900 I13.5 Sao Paulo 2280 2368 1253651 i10.7 Seoul 5240 1127 12705 2897 14.7 Singapore 2502 1391 3961 1260 20.5 Tunis 231 162 1728 700 12.4 3 3. Nevertheless the results do suggest that, despite the high costs, the economic return on metros can, be quite high when the. conditions are right. It has to be realised, however, that in the averge case the benefits consist mainly of nonworking time (62%) , relief of discomfort (7%) and the unidentifiable benefits from generated traffic (7%). "Hard" savings of re source costs amount to only 24% of the benefits, of which about half i s offset by the operating cost of the metro-itself. Furthermore, in most cases, the benefits are not yet "in the hand"; they depend on the assumptions of future growth of population and income. 34. The benefits are enjoyed largely by former bus passengers who now use the. metro, but* the remaining bus passengers also gain greatly from the relief. to the bus services. There. is little benefit for users of cars and motorcycles. Bus operators - -or their subsidizers - also gain 'when' fleet sizes !,have been adjusted to the lower demand. Government :loses, not only from the subsidy to the metro, but also from lower tax revenues from buses and cars. ;.There is almost, no foreign exchange benefits to offset the high f oreign exchange borrowings commonly incurred in metro construction. - CONCLUSIONS 35. A metro is an enormous investment for a -developing city. Clearly it should not be built until other ways of solving the same problems have .been properly studied. Bus lanes. and other bus priorities,,purpose-built buswaysj improvements to the road network (short of wholesale- expansion), light rail systems,- parking controls. and various types of traffic restraint: all can help to solve - at much lower cost - the problems of ineffective bus services and traffic congestion. (We have not included major highways as an alternative solution, because of their capacity limitations as carriers of public transport, because of their great cost and environmental disadvantages, and because no one in our case study cities suggested such a solution). 36. In very large cities, however~,- buses will sooner or later be unable to meet the growing demands of the city centre, and the alternative to a metro will be a change of direction in the development of the city: the growth of the centre will be stifled by chronic inaccessibility; offices, shops, places of culture and entertainment, all of the type that would normally gravitate to the city centre on account of its situation as the focal po~in~t, jof the~-.'t-tran'sport system, Will. be fot ced'~'to seek suburban location's becauise the city-cen'tre will have lost its ra-ison d'etre, its position of maximum accessibility. - TThh&307e.c.o no~mics of the city ..,centre.,-' -and 'the .implications o restricting' or restraining its 'growths, are ' subj ectd. of importance and 'complexity, and. they' lie .,beyond :.the 'scope of -this Study." 38. The largest of the developing' c-iti~e's,'.and. a few smaller ones, have already chosen to follow the European ;:pattern, ..aiid from '.their-' expoeri'~-nce the .'Study: can confirm':th'~g eneral belief thht~metros' in developing cities cannot be f inafibiali'y viable. But ,it -can al-so ,conclude. that "some 'of their' metros .appear.- 'economi'cally" re--spectableb; ahid some' of' the l~~ess. impressive :'lihes' coutl'd' have 'p;~-rformed.-"nt'ich better if they had been more efficiently planned--and implemented. As developing cities continue ,to grow *in size~ and .wealth', there will be- mariy new proposals f or., metro lieand, some' of them --wi1 'almost .certainly be. ecb-'onomi.bally- justifia'bl'e'.' '3,9:, Bcsexntoosts'do- not Vary sub~stantial with city income, while benefits do, it ;isno surprising to find economic viability strongly .correlate'd vith., city, income.. 'Of nthe higher income ;.developingj ..citits'"; aret metros- 'likely"- to ,generate .economic returns~ signif icantly-' abhove ~Governmnent.'Te-st 'discount rates' typically 10-12%. The--. exceptions .mnay - be --sho-rt scton ~ofr 'rute linking ,vexisti~ng underutilised lines, -o p'Ss11ly extensions to'lex~isting sys'tems,. -,40. -The Study has prodUced som rough gideline as to the prospects of justifying a' mie'tro in a developing city. They cannot,, of course, be ta]ken -,'a~is:.r cuunmivsetrasn-;caell s,- -r>ul:emse' ;. but sub-jec~t to: exceptional0 is not, 'l ikeily to' 'produce a good: economic return unless most of -the followin~g conditions hold~: '* . .a). Corridor size:: ,all successful lin'e's serv'e r!a-dial corridors to the city centre, with total corridor flows of over 700,000 trips per day before the metro opens.: This recognises-j the presednce. of'-at 'least one major ',-road,-,,with minor pa-rallel- 'roads, in: the corridor, and some turnover o'f passengers'.along the corridor. . It.-implies very high peak flows of close to 15,.00,0 bus -pasls~engers/hour/'dir ec'tio'n.', b) ~City msize: corridors o6f.this' ma'gnitude are found only i-n.:cities of. at least 5- million inhabitants unless the city is highly linear and contains only two mcain corridors,. p. q., Pusani, -NecellJcin',~.. Singapore has only three mnain.-.cori4iors. -uv-.A c) Income: clities with successful heavyj metros have ,average city, 4ncomes above. US $l8,00/head',i -except. for Cairo 'where the situation was unusuallyr .favourable -for a..metro.r, Cities-with incomes of,.th-is level-are unlikely', to.b e f un~ nlessithe natioQnal level' is.;at least us. $1000/head. -.. d~) Growth,.p-r-ospectsi: the 'economic~-viab~i.lity o,0f ,-mo"st of-~the metros tested, depgends ,.on fuur grwt - of popula.!tion and income. mass~ive influx-.of:-immigrant population. tends, to depress, a-veragd.e incomes, howe'yer';'.. soQ the ideal-,- .p-rospet.os ih steady popul~ation- growth, combined. -with ,-..strong~;-:eiconomic growth., e)City centre, growth:; -,growth of m-etro trafi i -1rectlyr assqciated with. growth of- the- city -. centre-. The-- i1deal cond-ition- i~swhen t-he .c-it-y,, is. a...nrati~onal capital and centre- of- a_-.large lppous!-~ -e.conomic .region.- : -- ~ tl)~~~~~~ ---- - ' -ty- ~) City management: -building.' akmetro through',-the middle, of a, "large congested city isa'prof-ound1ly demanding tas.Cos idex~abl~e expe-rtise -.is -required to hives-u.cce;ss ,.and, much can -go.: ~wrong.. - The .expertise of, -city -authorities -in achieving 1 small1 .I.Tprovements (eg in - traffic -mnemnt o conero~6l~ling bus op e rat i ons)-- p rov ides, a n.-.ind icato~r of. this institutional, capability, which:.is, es~sentia'l in impleeti~ng,,almetro. --, - -' g)-Metro. managrement:: once.oeainL ec-onomic, .success depends vital~ly-, on. good management., :,Most'. of the.. new' metros are :'rufl:- by~ n ew ~canies'", ocr .corporations-, independent of government departments, national railways or bus operators, and free to choos~e thei~r:own staf f and set up modern management systems. h)F~in'ancdal support: fares rnust~-,:be.. -reasonably ,co~mpe~titivewit 'the ,bilses~,, .which. means that -~they ~~usbte either- gra~~duated, .Qn both-..busesad~mto or -Intggrated (ie. with ,,through-tickets), .and -,the metro f'ares .must. be 'se~.t-.1jw-,-enough-to-:compete.: In practice t h i s' w-ill s el do m be possible without ,financial support, to--cover, caipital zco~sts- ,and at least. part of depreciation (asset. replacement). I,~ications -f-or-a -i&~ agencies 41, Metios-r.,,. often -riskcy proj~ects. ..,Much .n-eeds to zgo ri9jlt tahieve-- sxjpcess,,,.and -one or two bad mistakes cn ~.rdp alr. xtensi-on:,o existing metros are, howev er, likely to be less -..risky h-iita --hi nvestment.:,, ,T>is.-m ay-i weigh hi'avilyw~ith aid-'agenci.es.. - 4.. Ai4agenic~es, -are unlikely -to support~. metro _,proj-ects. that,,~- on-,a ~n-,objective evaluxation proifise -Iess~--tliakn-: 1OQ',.qconomicrrate of~.return.. 'Even-i~f. the return is 20%, they may considertthat-other p-roj~ects should take precedence. There is no way of proving beyond dispute the case f or a metro or any other public investment; but a good rate of return may be regarded both as a necessary condition and as an assurance that the resources will not be wasted. 43. Inevitably there wil be more developing cities seeking support from aid agencies for the construction of metros, and there will almost -certainly be some candidates that can pass the normal tests of justification. There may also be new f inancing arrangements designed to bring in more local contributions and more participation by the private sector., 44. The model and method of analysis developed in this Study are designed to help aid agencies, and also the cities themselves, to come to a first appraisal of the impact that a metro project would have on the city transport system, and of its .financial and economic viability. REFERENCES 1. DUNBAR F C and R T RAPP, 1986, "Urban transport economics: analysis- for development banks". Paper prepared for the First Annual Meeting International Mass Transit Association, Washington D.C. 2. HENRY E, 1987, Principales interrogations sur les metros. Journee specialisee INRETS, Joinville, France. 3. RIDLEY T M, 1986, "Metropolitan railways" in 'Moving people in tomorrows world', an Institution of Civil Engineers Conference. Thomas Telford Ltd, London. pp 19-49. 4. RIMMER P J, 1988., "The World Bank's urban transport policy: authorised version, revised version and the apocryphal'. Environment and Planning A, Volume 19, pp 1569-1577. 5. WORLD BANK, 1986, "Urban transport: a World Bank policy study". World Bank, Washington, D.C. The work described in this paper forms part of the programme of the Urba~h:'"..7T~r'anisp~o"'t ~~a'L fi .Manag~eumont - -56c-ti-on'~o f the.:'Ovio/e rAsis Un"it of -the ..T-ran-sport ~h od>e~rh~Do•6y t S -publ'is'h' -ba permissdion bof th&d--tiiectE61i. - Crcown Copy~r-ight ..7 Anyh iw ae are not necessarily tho:Ee)~rbf'tiie., ep a'rthmeht 'of :.Tra~nsport:' '.or the veT"Development Administr-_a~t-ioh ~xtt.n fom te ex hmi y b relrodtuced, except fo6 6erka p~r~6-V -ided :the..so~urcd- is-.acknbwledqeid.".