This paper is prepared for staff use and is not for publication. The views expressed are those of the author and not necessarily those of the Bank or the Association. INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOPMENT INTERNATIONAL DEVELOPMENT ASSOCIATION Economics Department Working Paper No. 73 A Framework for Urban Transport Studies March 27, 1970 This paper is a product of the continuing investigation of the Economics of Urbanization Division into methods for approaching urban development and, in particular, for appraising urban projects. After previous works on the interrelationship between land use and transportation are examined, an urban transport study method is proposed. On the project analysis level, the quantification of benefits is discussed. Economics of Urbanization Division Prepared by: Koichi Mera TABLE OF CONTENTS Page No. I. INTRODUCTION . . . . . . . . . . . . . . . . . 1 II. A REVIEW OF URBAN TRANSPORT STUDIES...... 5 A. Complexity of Analytic Techniques...... 5 B. Transportation and Iand-Use Models . 7 C. The Cost of Urban Infrastructure ......12 D. Modes of Urban Transportation . . . . . . . 18 III. STUDY METHOD . . . . . . . . . . . . . .......21 IV. BENEFIT MEASUR-EMENT . . . . . . . . . . . ... 26 A. Approaches to Benefit Measurement ......26 B. Economic Efficiency Benefit . . . . . . . . 31 C. Other Benefits . . . . . . . . . . . . . . 36 V. SUMMARY AND CONCLUSIONS . . . . . . . . . . . . 39 A FRAMEWORK TO URBAN TRANSPORT STUDIES I. Introduction 1. This paper presents a framework to urban transportation studies which could be adopted as part of a general urban development study for operational purposes as opposed to educational purposes. Such an urban transportation study is designed to produce a long-range transportation investment program for an urban area and to identify and evaluate high priority transport projects to be undertaken in the immediate future. A general urban development study covers a number of other sectors as well. However, it is maintained here that decisions on transportation are coor- dinated with those in other sectors in the general urban development study. 2. We shall assume that the objective of an urban development study is not the development of a particular urban area, but the general economic development of the country in which the urban area is located. There are at least three implications to this distinction. First, in most countries a large portion of income and wealth is created in urban centers and, therefore, urban centers are expected to play a significant role in the development of the country in the future. If an overriding objective of a country is e:conomic development, it has to be achieved, to a significant degree, within the urban area in question. The urban development program derived from the study must be consistent with the objective of increasing national income. Second, since urban areas are open to any citizen of the country, in contrast to countries which are generally closed, the develop- ment program has to take full account of its impact on national resource -2- allocation through market forces. It is just not possible to build and maintain a pleasant, highly capital intensive city when the rest of the population is substantially below the level of affording such luxury. A usual reaction, for example, is squattering next to glass and concrete palaces. Third, an urban development program cannot be made without regard to the resource availability. In most countries resource availability is significantly controlled by the central government. The central government allocates resources on the basis of national priorities. No development program for an urban area can be made useful without reference to resource allocation at the national level. 3. Two extreme viewqs are often expressed in regard to the method of urban planning and decision making. One states that an urban system is a complex organism in which all elements depend on each other and, therefore, a comprehensive analysis is required to derive a correct solution to any problem. According to this view, the decision to build a traffic signal at a specific intersection cannot be made until the child-rearing practice of migrants has been sufficiently studied. The other extreme view is that only one specific project needs to be studied; for comparison purposes all other projects can be represented by the opportunity cost of the resources. A project is analyzed in isolation by making appropriate assumptions, often of certeris paribus, in regard to all necessary information required for evaluation. 4. Consider the comprehensive approach. Two things must be considered: first, the amount of data required for a comprehensive analysis and, second, the accuracy of the analytic process used. Data on the developing countries is almost always insufficient to carry out an ideally comprehensive analysis. Therefore, the economic cost of such a comprehensive analysis could be enormous, merely in terms of the time required for collecting data and for completing the analysis. The extra cost of comprehensive analysis must be compared with the extra benefit it can generate. In terms of accuracy of analysis, there are equally profound questions. Although it is easy to state that every element depends on every other element, it is difficult to quantify the exact relationship for most of the vari- ables, particularly for those not related in any directly measurable way. As will be discussed later, the accuracy of the analytic processes itself must receive due consideration when any quantitative analysis is under- taken. 5. Any social science model is constructed according to our under- standing of the system involved. A model is not a mysterioue black box, but a set of interrelationships between variables in the mind of the model builder. Interrelationships are conceptualized only in a number of dis- crete ways. Therefore, even though the urban system is a complex system in which every element depends on every other elemeriT, we .an reason the impacts of a policy variable by following the stipulated interrelation- ships. For this purpose fairly simple analytic techniques are usually required. We should be cautious against the approach of building a gigantic model without clarifying the basic interreiationships beforehand and testing policies with the model at random. -I4- 6. An urban transportation study requires information on the pref- erence of individuals and organizations.with respect to the choice of loca- tion and the mode of transportation and on the policy tools available to the government in transportation programs. As will be demonstrated below, transportation policy decisions are not so powerful in determining the pattern of urban development as have been believed by many. We should recognize that individual preferences determine the location of different types of land use in a large part. However, such an objective is mis- directed for transportation studies, We should be concerned more with the efficiency of resource use than guiding land use development. For this purpose the objectives of such a study should be kept clearly in mind when the study is designed. The art of transportation studiies is in the selecting of key variables and key interrelationships which are rel- evant to the objectives so that the impacts of transportation decisions can be properly evaluated without spending a great deal of resources on the study itself. 7. A proper transportation study rests on the formulation of a proper methodology. Knowledge for this comes from previous empirical and theoretical research efforts. In the next section., a review of past efforts in the area of transportation will be presented, and some basic guidelines will be established for future urban transportation studies. In Section III, an approach to urban transportation studies is described on the basis of the findings in Section II. In Section IV, conceptual problerms in measuring benefits of transportation investment are discussed. Conclusions are pre- sented in Section V. II. A Review of Urban Transport Studies A. complexity of Analytic_Techniques 8. William Alonso casts serious doubt on the current practice of building complex models for predictive purposes.- He analyzes the accuracy of prediction in models of different degrees of complexity. He distinguishes between two types of' error which lead to error in prediction: error of measurement and error of specification. "'Error of specification arises from a misunderstanding or purposeful simplification in the model of the phenGmenon we are trying to re- present.... Measurement errors are those that arise from inaccuracy in assessing a magnitude."?/ He proceeds to demonstrate that the relative magnitude of the measurement error frequently increases as the variables are processed through mathematical operations and that the reliability of the predicted value declines as the number of the processes increases in the model. As far as the specification error is concerned, assuming correct knowledge on characteristics of phenomenon, the magnitude of error diminishes as the number of variables and equations in the model increases, i.e., as the complexity of the model increases. However, after a certain degree of complexity is reached, the reduction in the prediction error resulting from the speciiNcation error is more than offset by the increase in the prediction error resulting from the compounding effect of the measurement errors. Therefore, he argues that an increase in the complexity of a model does not necessarily improve prediction. According to his view, the optimal complexity of a model or of an analytic process depends upon the types of mathematical operations and the accuracy of data used. In general when data is less accurate, simpler analytic processes are recommended. Also, he discourages a long chain of analytic nrocesses as errors cumulate in the orocess. 1/ William Alonso, "The Quality of Data and the Choice and Design of Predictive Models," Urban_Development Models, Highway Research Board Special Report 97, Washington, D.C., 1968, pp. 178-192. 2/ Ibid., p. 178. -6- 9. Alonsots argument is made without reference to any specific subject. He does not discuss the bias in prediction associated with par- ticular processes, nor does he discuss the error of a predicted variable with respect to another predicted variable, which is often more important than the absolute error of a predicted variable. However, his analysis is particularly relevant to the study of urban transportation in cities of developing countries. Tn a number of respects, urban transportation in cities in developing countries is more complex than that in cities in developed countries. Take for example, the transportation modes seen in different cities. On the streets of U.S. cities, there are only a few typ.es of vehicles, whereas on the streets of Calcutta, there are auto- mobiles, buses, animal-drawn hand carts, rickshaws, bicycles, scooters, pedestrians, streetcars and sacred cows. In terms of the distribution of land uses, cities in developing countries present more complex patterns than developed country cities. In American or European cities, land uses are distributed in fairly discrete groups. However, in developing countries urban land ases are heavily mixed. Mixed uses are the rule rather thans the exception. Therefore, even when there is a fair amount of data about land use and transportation, it is more difficult to build a satisfactory pre- diction model for cities in developing countries than for those in developed countries. One obvious way to reduce this difficulty is to improve the data, but experience in the United States indicates that such improvement alone does not lead to a successful result unless a corresponding improvement is made in the analytic technique. -7- B. Transportation and Land-Use Models 10. Let us now consider the interrelationship between the transportation network and land-use pattern as it affects the overa'l efficiency and welfare of urban areas. In this connection U.S. experience is quite instructive. The U.S. Bureau of Public Roads and other agencies have been financing large scale transportation studies. From 1944 to 1961 more than two hundred studies were conductbed which involved large-scale home-interview surveys.!/ The objective of the earlier studies was to derive a transportation system which would satisfy the predicted transportation demand. The demand was predicted inde- pendently of the transportation system to be implemented. The task of the transportation study was conceived as fulfilling the transportation require- ment. The concept of choice among alternatives was not incorporated into the study.2/ 11. Since the advent of the contemporary planning doctrine around 1960, a number ox' studi-es have tried to evaluate alternative urban development plans and policies. David Boyce reviewed 13 major transportation planning studies conducted between 1959 and 1968 which are, to a large extent, based on the following beliefs: (1) in order to find a desirable development plan, a comprehensive analysis of alternative development plans considering both land use and transportation is necessary; (2) the cost of transportation can be varied by altering the mix of transportation modes and the pattern of the transportation network; 1/ Richard M. Zettel and Richard R. Carll Summary Review of Major Metropoliban Transportation Studies in the United States, University of California, Institute of Transportation and Traffic Engineering, Berkeley, Calif., 1962. 2/ A typical study of this kind is the Chicago Area Transportation Study, of which the final report was published in 1959. Such studies are discussed in Zettel and Carll, M. cit., and N.A. Irwin, "Review of Existing Land-Use Forecasting Techniques" Highway Research Record No. 88. -8- (3) the density of development is strongly related to the trans- portation facility and services, and therefore transportation policies cen be used as an effective lever for obtaining a more desirable physical pattern of development.-/ 12. The thirte-en studies covered the following metropolitart areas: Baltimore, Boston, Chicago, Detroit, Los Angeles, Milwaukee, Minneapolis- St. Paul, New York (2), Philadelphia, Pittsburgh, San Francisco and Washington, D. C. In terms of time, the shortest study took three years, and the longest, eight years so far, though it is still in process. In terms of cost, a typical study cost more than one dollar per i&habitant in the metropolitan area studied.L/ The plan-making process common to these studies is: (1) generation of alternative concepts, (2) elaboration of each concept into an alternative plan, (3) evaluation of alternative plans, and (4) the decision process with respect to the alternatives. Alternatives were con- ceived in terms of land use patterns such as "radial corridors," "spread city," "multi-towns," "linear city"; in terms of structural characteristics such as low density versus high density, concentration versus dispersal; or in terms of transportation systems such as highway versus transit, arterial versus freeway, grid versus radial, minimum transport versus maximum trans- port. In many of the studies, from three to four alternatives were evaluated. 1/ David E. Boyce and Norman D. Day, Metropolitan Plan Evaluation Methodology, Institute for Environmental Studies, University of Pennsylvania, March, 1969. 2/ Zettel and Carll, op. cit., p. 15. The cost of travel survey constitutes a large portion of the cost. In many cases alternatives were evaluated on the basis of performance in transportation services and cost of facility and service systems. 13. After reviewing the planning process and the evaluation process, Boyce concludes: (1) In those planning programs using urban development models to elaborate alternative land use patterns in which only transportation policies were varied, no significant differ- ences in land use patterns were identified. (2) In those programs in which land use patterns exhibiting significant differences were used as the basis for testing a single transportation system including both highway and transit, no large difference in network performance and cost was demonstrated for the land use alternatives. (3) In those programs using urban development models to generate land use alternatives in which both transportation policies and land use Dolicies were varied, the resulting land use patterns and transportation requirements were not sufficiently different to provide a technical basis for policy decisions. - The significant difference as, used by Boyce here, is a difference large enGagh to enable either the professional staff or the committee supervising the study to choose among the alternatives on the basis of the difference. The lack of sufficient difference in the outcome of the ubran development models used led the study groups to choose on the basis of non-economic criteria, such as the attractiveness of a strong metropolitan core. 1/ Boyce, 2. cit. p. 61 - 10 - 14. Boyce attributes the above observations to the following factors: (1) the land use and transportation policies tested were too conservative and too similar to obtain differences among the alternatives; (2) the models and methods available were too blunt to produce differences in alternatives in response to the policies assumed; (3) for the increment of growth and change and for the fore- cast period commonly used beyoncI the future base year, no significant variation could have possibly been obtained; (4) the size of the districts used for forecasting develop- ment and travel demand was probably too large to detect the assumed land use and transportation interaction; the ex- pected differences in spatial arrangements of aotivities and environmental quality r..' ,1 I2 ~.&.k')52I. - .......... tt !Ph ."II:. 6- .e- Si . I s.- a~- 1s I - i IIt a: : I -II .. . r'~I~I .: , -.rar- . ... - 1,--I ,.,- :'n.z-a> .- ---" ,L . .. . .. .. . C.: .. - 33 - 51. There is one type of benefit which is not explicitly considered in the Van der Tak - Ray analysis. That is what Morhling and Williamson call the reorganization benefit. This benefit results from the locational ad- justments of producers in response to reductions in transportation costs. They have demonstrated that such benefit can be measured as a part of the usual consumers' surplus. 52. On a close look, the Van der Tak - Ray analysis contains an element applicable to the organization effect. This appears in the analysis of rail- road and road transport as imperfect substitutes. The demand for two different modes is interpreted as two demands existing in two different locations. These two different locations are regarded as imperfect substitutes. Hence, the same analysis applies. For practical purposes, the sum of consumers' and producers' surpluses realized from the improved facility constitutes the entire economic efficiency benefit. 53. The Van der Tak - Ray analysis does not provide much clue for measur- ing benefits when marginal cost pricing is not practised. However, in the transportation sector, monopoly and non-marginal cost pricing is the rule rather than the exception. Particularly in the case of urban highway trans- portation, average social cost is the price perceived by private users. Marginal cost is not perceived by individual users at all. In the case of bus and train services, the prices are fixed by government agencies or by the monopolistic pricing policy of the operating firms. Under such cases the 1/ H. Morhling and H. Williamson, Jr., "Scale and 'Industrial Reorganization' Economics of Transport Improvements," Journal of Transport Economics and Policy, September, 1969. - 314 - measurement of benefit should be examined too. In the following we shall merely focus our discussion on the measurement of economic efficiency benefit under such non-marginal cost pricing in policy without placing much emphasis on the merit or demerits of such pricing policy. 54. Let us consider, first, a typical highway in which private automo- bile users perceive average social cost and not marginal social cost. In this case, suppliers are the individuals who demand the transportation. The resulting supply curve is in fact the average cost curve instead of the marginal cost curve. In general, such an average cost curve increases as the volume of traffic increases because of congestion effects. Therefore, the marginal social cost curve rises above the average cost curve. When the facility is improved, the average cost curve is lowered. The concept of con- sumer's surplus is still applicable in this case but the producer's surplus is nil because the total cost is the average social cost times the quantity of transportation: the users as a group pay the total social cost. See Figure 2, where MCO and MC1 are, respectively, the marginal cost curves without and with the facility improvement and AC0 and AC1 are, respectively, the average cost curves without and with the facility improvement. D Mo M a\/ ,X CA / NC1 0 T-f4 I'ran.sportati-.n Figure/ Notice in this case tIhe downward shift of the supply curve (theaverage cost curve) is irrelevant for measuring the benefit. The entire benefit consists only of consumers' surplus. Let us suppose that monopoly pricing prevails in the supply of transportation service. If the price charged does not change when the facility is improved, then there is no consumers' surplus, only producers' surplus. The producer's surplus is the entire economic benefit. Let us now suppose that the prices are set by the monopolist so that tŽe marginal rev- enue is equated with the marginal cost. In this case, the facility improve- ment will lead to change in the price. The situation is illustrated in Figure 3, in which there are both consumers' and producers' surpluses. AM C T T+it Transportation Figure 3 The facility improvement shifts the mar'g-..nal cost curve from MCO to MC1. The price is determined by the intersection of the relevant marginal cost curve with the marginal revenue curve) MR. The price is reduced from A to E and the traffic is increased from T to (T + t). The resulting con- sumers' surplus is the area ABFE, and the producers' surplus is EFGH. - 36 - However, the original producers' surplus ABCD is lost in this process. The net increase in the benefit is the dotted area DCBFGH. It is likely that the consumers' surplus is apt to be proportionally less than in the previous cases. 56. Incidentally, this diagram shows the net loss associated with monopoly pricing. The triangle FIG in Figure 3 is the loss of benefit due to monopolistic pricing. In this connection, the distinction between the financial and economic return can be pointed out. The revenue for trans- portation operator declines when monopolistic pricing is abandoned in favor of benefit maximizing pricing. Therefore, benefit maximization leads to a lower financial rate of return. The use of financial returns for economic justification should be carefully avoided. S. Other Benefits 57. As has been mentioned earlier, the economic benefits are distributed to various classes of people in the area. The distribution can be measured by income class, neighborhood basis, producers of transportation services versus consumers, landlord versus tenant, or producers of goods versus con- sumers. Generally, it is agreed among policy makers and academicians that more egalitarian distribution is desirable. But to what extent a specific distribution is desirable is a matter of value judgment. When distributional impacts are considered important, they must be examined and the factual dis- tributional effect must be presented as reference to policy makers. If we proceed to evaluate distributional impacts, subjective elements come into the analysis, but there are some methods which employ less subjective criteria. One method derives the implicit value judgment of politically elected decision makers with regard to the welfare value of a dollar in - 37 - 1/ different income classes.-/ Similar methods are advanced by Weisbrod, McGuire and Garn.-/ There are ways of integrating distribut.ional and economic efficiency benefits. However, at this stage of the art, such a mechanical technique should be used as a reference,not as the determining criterion. The basis of making recommendations should be the best judgment of the researcher. 58. The environmental effects of transportation investment encompass traffic safety, air pollutino;, noise and vibration, creation of unsightly and undesirable areas, and clime induced. The general strategy for evalu- ating these environmental impacts is quite similar to the one used for measuring the value of travelling time. In the case of noise frcm a highway, the depreciation of property value along the traffic route can be observed. That amount can be counted as a negative economic benefit. It can also be argued, as for the case of leisure, that most of these impacts are irrelevant for maximizing the growth rate. What is more difficult to evaluate is the benefit of increased traffic safety. There have been a number of attempts to attach a monetary value to lives saved, but the problem of valuing human lives which have other dimensions than economic benefit cannot be sidestepped. Therefore, the best the researcher can do in this area is to present objective estimates of each variable affected by the investment based on the behavior of the population. 59. Another type of environmental effect is the comfort associated with each mode of transport. Such difference in comfort can be reflected in the demand for each mode. If the demand is correctly estimated, there is no need 1/ Koichi Mera, "Experimental Determination of Relative Marginal Utilities," , August 1969, pp. 46h-477. 2/ Burton A. Weisbrod, "Income Redistribution Consequences of Government Expen- diture Programs," 81.B. Chase, ed., Problems in Public Expenditure Analysis, Brookings Institution, Washington, D.C., 1968; Martin C. McGuire and Harvey A. Garn, "An Experiment in the Integration and Efficiency Criteria in Public Project Selection, "r unpublished, 1968. = ; ;S j f|t &i^ 0 aa7 iYF t ' .i g& 7fzAita a &k$.>S9X;t! >3;§008R ' ....... S.Lil '' ..... . ;W k '.S>A.:.: i.Ml .:<"'. ..t' ....... S' Q .- . .....'.. . ... .!:-.- 38 - to consider difference in comfort further. In order to estimate demand correctly, the comfort difference between the modes of urban transport should be considered. For example, a train gives a stable ride in which reading is quite possible whereas a bus ride is not so smooth. However, for commuters, bus stops are more readily accessible than train stations. 60. In developing a long range land use development plan and, in parti- cular, predicting transport demand for specific routes, allowance for uncer- tainties should be given. Technology, the level of income and also consumers' taste change day by day. More significantly the art of predicting transpor- tation supply and demand and that of land use development are in the formative stage. We should not assume that we can predict the future with any degree of accuracy. When an unexpected development takes ?lace, the demand for specific facilities change and subsequent decisions must he changed accordingly. For this reason, flexibility of the system itself is highly desired. Such ttlexi- bility can be obtained in the stage of modal choice as well as in the allocation of space for specific purposes. Specifically, rail service is less flexible than road mass transit service since it involves a large fixed investment on the right of way which can only be used for specific limited purposes. W4e must recognize this additional benefit inherent to bus service over rail service. 61. To translate the amount of flexibility into economic benefit requires information on the probability distribution of future events and the preference for different events. Such information cannot be obtained for any specific area without spending substantial amounts for research. At this stage, i-t can only be said that flexible systems are preferred to less flexible ones. - 1 os .r . . r u F<.-;t ...: v ::e.1. :, ..... !. :e 1~.zeE - ...:R.sS-..U.Wsa ...... a |t f z A.. . . . .. .-. '. . .".. - 39 - 62. The above discussion shows that we have the firmest ground for economic efficiency benefit and rather weak ones for other benefits. Therefore, a convincing case can be made if the economic efficiency benefit alone is large enough to justify the cost of investment and other benefits are not negative. If a project is conceived primarily on the basis of redistribution benefit or environmental effect, some of the developing techniques must be used to justify the cost. V. Summary and Conclusions 63. A review of urban transportation studies has led to the following conclusions: (1) the urban transportation and land use simulation analysis which has been used frequently during the 1960's in the U.S. has been ineffec- tive in improving the planning of urban development; (2) the employment of a complex simulation model for planning urban transportation systems would require at least substantially improved data and analytic techniques which cannot be expected even in the developed countries in the near future. 64. The physical pattern of land use and of transportation network is not so significant in affecting the efficiency of resource use in transportation. Greater attention should be given to technological alternatives. For example, the transportation cost savings due to high density development are not likely to be large enough to compensate for the increased construction cost necessary to achieve high density. 65. Considering the above findings and the particular situation of the developing countries, the following points are made: (1) the duplication of the U.S. experience of building complex models should be avoided in developing countries and (2) in an analysis of a particular issue, significant variables and relationships should beexplicitly identified and the scope should be limited; a lengthy chain of reasoning should be avoided as much as possible. 66. The following three steps are recommended for an urban transportation study: (1) development of the general urban development program without particular reliance on any extensive simulation analysis, (2) comparison of different modes of transportation based on demand and supply conditions which are consistent with the predicted population and income levels and (2) the evaluation of priority projects which should be conducted with the full use of economic techniques. 67. The benefits of urban transportation investment can be grouped into (1) economic efficiency, (2) distribution benefit, (3) environmental effects and (h) flexibility to uncertain developments. Among them, the economic ef- ficiency benefit is most readily quantifiable. Some techniques are emerging for quantifying distribution benefit and environmental effects, but they have not yet reached the stage where they can be used for operational purposes. 68. There are two approaches for evaluating savings in traveling time: growth maximization and utility maximization. There may be some difference in the value of time savings depending on the approach used, but it does not seem to be significantly large. 69. The economic efficiency benefit of a particular urban transportation project can be measured by the variables associated with the project alone. In estimating the benefit, the pricing policy must be explicitly considered as it affects the composition of consumers' and producers' surpluses which are to be included in the benefit. The financial rate of return does not necessarily co- incide with the economic rate of return and the degree of deviation depends upon the pricing policy. 1l1as54MUBebiM¢4t1MbS>a+ S1BAtifF6= A't9g A¢F<+bAStsXvtev>u'f'?!Sa -s.ut sz1t,ez;wA ^.lsX+t^. ne.sB *jlxuq .... .. ..- -. .X.v- -