& w H Z w u Z O INTEGRATING PROJECT WORTHINESS INTO A COST-EFFECTIVENESS FRAMEWORK FOR ALTERNATIVES ANALYSIS ^OFT^v, SrATtS G» V ^3 S o Ph C/D z S RESEARCH NORTHWESTERN UNIVERSITY EVANSTON ?«ÎN01Si Technical Report Documentation Page 1. Report No. 2. Government Accession No. 3. Recipient's Catalog No. 4. Title and Subtitle Integrating Project Worthiness into a Cost-Effectivenes: Approach for Alternatives Analysis 5. Report Date February, 1981 6. Performing Organization Code 7. Aurhor'j) joggph l. Schofer, Mark A. Tumquist, and Steven E. Polzin 8. Performing Organization Report No. 9. Performing Organization Name and Address Department of Civil Engineering and The Transportation Center Northwestern University Evanston, Illinois 60201 10. Work Unit No. (TRAIS) 11. Contract or Grant No. IL-11-0026 12. Sponsoring Agency Name and Address Office of Policy Research Urban Mass Transportation Administration U.S. Department of Transportation 400 Seventh Street, S.W. Washington, P.C. 20590 13. Type of Report ond Period Covered 14. Sponsoring Agency Code 15. Supplementary Notes 16. Abstract Evaluating major capital investments proposed for urban transit systems is complex and difficult, not only because of the scope of alternatives and their im¬ pacts, but also because of the mixture of objectives held by various governments (local, state, federal) involved in the choice. This complicates the evaluation pro¬ cess itself and has resulted in the preparation of some planning documents which may not provide adequate support for decision making. This report describes some of thes«: difficulties, clarifies some of the conflicting goals held for Alternatives Analysis, and discusses the lack of adequate concern in the evaluation for project worthiness, the relationship between monetary benefits and costs of options. Recognizing that such worthiness analyses must be partial (incomplete) in nature, the report points out the value of such analyses for estimating partial worthiness and providing a basis for assessing the other, nonmonetary outcomes of alternatives. A conceptual framework for integrating cost-effectiveness and worthiness analysis is developed. Some prag¬ matic limitations of worthiness analysis are discussed. An integrated approach to the evaluation in Alternatives Analysis is presented and related to an interactive graphics computer software system developed for accomplishing this task. This system is des¬ cribed in detail and examples of its products are presented. The system can serve as a basic tool for evaluation and as a technique for creating graphical components for technical reports. Included in its capabilities are the analysis of time stream data the development of effectiveness matrices, and the plotting of two-dimensional trade¬ offs. A discussion of strategic approaches to making Alternatives Analyses more oriented to user nèeds concludes the report. A user's manual for the computer system is included as an Appendix. 17. Key Words Transit Planning; Alternatives Analysis; Benefit-Cost Analysis; Cost-Effectiveness Analysis; Transit Evaluation; Interactive Graphics 18. Distribution Statement 19. Security Classif. (ol this report) Unclassified 20. Security Clossif. (of this page) Unclassified 21- No. of P ages 22. Price Form DOT F 1700.7 (8-72) Rep roduction of completed page authorized INTEGRATING PROJECT WORTHINESS INTO A COST-EFFECTIVENESS FRAMEWORK FOR ALTERNATIVES ANALYSIS Joseph L. Schofer Mark A. Turnquist Steven E. Polzin Northwestern University Evanston, Illinois 60201 February, 1981 U.S. Department of Transportation Urban Mass Transportation Administration Washington, D.C. 20590 ACKNOWLEDGEMENTS This work was supported under the University Research Program of the Urban Mass Transportation Administration, U.S. Department of Transportation. Mr. Alfred Harf of UMTA provided important insights into the UMTA Alterna¬ tives Analysis process during the research. Mark Turnquist created an efficient and intriguing computer program and was instrumental in formula¬ ting the overall approach proposed herein. Mr. James Bunch, graduate student at Northwestern, assisted Mark Turnquist in software development and contributed importantly to the formulation of concepts. Mr. Joseph Perl also supported software development. Steven Polzin, then a graduate student, worked on software, argued strongly for clearer concept develop¬ ment , and contributed many of the ideas in Chapter 6. Guido Timmermans, Maureen Schultheis, and Janet Hopper pursued a thorough test of the soft¬ ware system and produced the hard copy graphics. This report, most opinions and all errors in it, remain the responsibility of the first author. i Technical Report Documentation Poge 1. Report Ne. UMTA-IL-11-0026-81-1 2. Government Accession No. PB 82-113895 3. Recipient's Cetelef No. 4. Title end Subtitle Integrating Project Worthiness into a Cost- Effectiveness Framework for Alternative Analysis. 5. Report Dete February 1981 6. Performing Organisation Cede 7. *u»ty be put to better use if planners and decision makers put more energy into strategic scanning of local issues, problems, and interests so as to enter into CEAA earlier, prior to the point where actions commitments begin to develop. This suggests a need for a change in the approach to planning, perhaps by making a stronger, conscious effort to use the long range planning process to search out such opportunities for analysis and study in CEAA (Schofer and Stopher, 1979) . It may also suggest the need for the federal government to provide stronger encouragement and resource support for such preliminary, broad brush studies. The fear that a particular community may not get what it wants out of CEAA can be approached by making the process more useful to decision makers and less frightening to planners. This may still mean that a city is unable to implement an action which was preferred at an early stage of the decision process. On the other hand, it may help local officials see that they can, in many instances, find good alternatives which in some sense are both worth their costs and implementable given available resources. Finally, a more positive approach to CEAA may help a community find local resources to im¬ plement options which are best for them but are not compatible with current federal policies. The San Diego light rail transit system, which is locally funded, may exemplify this outcome. What are the implications of this discussion? First, implementing federal policy at the local level through regulation and discretionary funding is difficult because of the tensions between national and community interests. This is especially evident in the field of transportation, which appears to be so well circumscribed, but where there exists the most 11 imperfect knowledge of its broad spectrum of impacts. Thus, it is entirely logical to find cases where one major actor wants to maximize the cost-ef¬ fectiveness of transit service, while another sees the objective in terms of maintaining downtown property values, yet both explore the same set of alternatives. The conflict arises when each sees a different alternative as preferred; that conflict is heightened when one actor urges his or her own measures of cost-effectiveness onto the other. There are at least two possibilities for alleviation of such conflicts. First, if the state of the art of transportation impact assessment is advanced sufficiently, and given enough time for facts about impacts to drive out the myths, it may be possible to achieve a shared view, at all levels of decision, about what transport investments can and should achieve. Second, if the federal government can act to clarify in narrow, transportation terms, what it wants transport investments to achieve, much of the ambivalence experienced by non-federal officials may be reduced. At the present time, however, local officials are logical in their efforts to meet whatever needs they perceive through federal transportation funds. Federal transit officials may or may not wish to focus the resources they allocate along narrow transportation service lines. Yet the significance of federal tansit dollars is sufficiently great to encourage more broadly- oriented local officials to push the national choice to a higher level, where policy-making may well be more concerned with investing in cities in general, rather than in transportation in particular. Of course, the latter approach maybe entirely appropriate. Yet it leaves the transportation planner and tool developers in a difficult position, because the ambivalence of policy means it is not clear what technical pro¬ cedures should do. At the level of specific applications of transportation planning, this appears to be an ethical choice. In this research, the 12 choice was made to develop some useful approaches to evaluation, focused principally but not exclusively on CEAA, which may be of value where the local interest is to find the most attractive options in a relatively ob jective sense, and largely in terms of transportation dimensions. Even where local interests are broader, the ideas in this report may be useful at some point in the comparison of alternatives, if only in understanding some of the operative trade-offs in moving from the best alternative in a narrow, transportation sense, to the option perceived as best for the com¬ munity as a whole. Politics and Planning: A World View It is important at the outset to recognize the interrelated roles of planning and political decision making. Some analysts view politics as an evil, unstructured, illogical process which moves in rapidly to fill any gaps left by inadequate or incomplete planning. Politics can be a pejorative term which includes decisions made without regard to technical findings. The result of this view can be an unwillingness to communicate meaningfully with elected officials, an inappropriate sharing of responsibility between politics and planning, and ultimately, the failure of planning to support the political choice process. The consequence is that politics, sometimes in the worst sense, comes to dominate social choice, and planning is ineffective. This perspective is a self-fulfilling prophecy: "politicians are evil, we won't deal with them, and (therefore) they ignore the results (unresponsive) of technical analysis." The reality, in most cases, is different. Politics is the process through which duly elected or appointed representatives of the public, who are accountable to their constituency, attempt to represent and respond to the needs and values of the community through the management and development 13 of public facilities and services. Politicians have the responsibility of making the major public action choices in society, taking into account available information on needs, alternatives, and community values. In an idealistic situation, politicians and politics can focus on true value choices, where technical analysis can make no headway. Even the most advanced and sophisticated social science techniques for deriving social values are inadequate for this task because, in our representative demo¬ cracy, it is assigned to accountable, elected officials. One major attribute of the system is that, unlike analysts and their tools, public officials are obligated to stand up for their actions, to justify their choices, and, in some cases, to suffer the consequences of job loss when they fail to meet • the requirements of the majority. In such an ideal world, there would be a clear division between issues of value and issues of technical fact. While the politicans wrestle with values, analysts can be focusing on developing and refining the facts, and in that process, the latter can narrow the scope of most decisions rather con¬ siderably. Where the relationship works effectively, priority items for analysis are specified by politicans, supported by information from planners, who then pursue technical studies and recommend one, or a few alternatives to decision makers. At this point the decisions to be made are largely based on issues of value, and politicians, who hold this responsibility, must and do take them. The merits of the political system of today are clouded by the fact that issues of value and issues of fact are not so clearly differentiated. In part, this is due to fundamental uncertainties associated with alternative actions, their impacts, and the criteria for choice. It is also due to a failure on the part of some analysts to perceive the value implications of 14 seemingly technical choices which must be made. For example, decisions about which corridors to study in CEAA, which, alternatives to consider seriously, what rate of inflation to use, which land use forecast to apply, which land use model to select, etc., all may have, value elements in them. Planners may be able to increase their effectiveness by recognizing and respecting the vital role of politics in decision making. The work presented here assumes that political decision making will determine the actions to be taken and, indeed, many of the assumptions which lead to them. No model, no evaluation tool, no computer program can or should change that. At the same time, analysis which is of good quality, and which is presented in an under¬ standable and useful way, can help improve and maintain the quality of those political decisions. It is only when planner and decision maker work effec¬ tively in consort that society can get the most out of both processes. Recognition of the philosophy that planning should be strongly suppor¬ tive of political decision making is important in this work particularly be¬ cause it attempts to deal rather directly with the question of worthiness analysis. Where all of the assumptions and requirements of economic evalua¬ tion can be met, well-developed decision rules (such as cost-benefit analysis) can be applied, which produce a clear and strong decision about the most worthy alternative. A weakly structured cost-effectiveness analysis alone essentially leaves all matters of choice to the political process. By pro¬ posing to increase the structural element of cost-effectiveness, and to in¬ clude an explicit, though partial, consideration of the worthiness question, this report intends to strengthen the role of the analyst in CEAA for the purpose of improving the process. This does not, nor can it, mean de-politicizing the decision, for the issues of value are too strong and pervasive to permit this. It does mean that the analyst may become more influential by presenting more useful and 15 more understandable, information to his or her clients. The analysis system presented here is not intended to make choice. It is intended to insist that decision makers examine certain issues deemed to be of importance, many of which are predefined in this report, and others of which may be defined by the on-scene planner. It makes sense for that planner to try to work closely with the client to make sure that decision makers and the community have ample opportunity to participate in the definition of the important issues. There is, of course, a not unsubstantial risk that, by presenting a more comprehensive information package about the alternatives, and by ex¬ plicitly considering the worthiness question, favorite alternatives of the community and its decision makers will be shown to be unattractive. Where this is true because the CEAA process has ignored other, important factors, appropriate ways must be found to include those factors in the resultant products. Where it is true because those alternatives are indeed unattrac¬ tive, it behooves the planner to let this be known not only to help his or her community find the best alternatives, but also to increase the likeli¬ hood of federal support later in the decision process. A good example of both the risks and benefits of this philosophy may be found in the approach to worthiness analysis included here. It is pro¬ posed that a user benefit study be conducted much the same as such studies have been done in the past for highway investments. It may turn out that, with few exceptions, it is impossible to find transit alternatives which can be justified strictly in terms of their user cost savings alone. This does not mean that no such alternatives should be implemented although this outcome may seem threatening to community preferences. Knowing the degree to which traveler cost savings offset investment costs can be useful, even though — indeed — especially because, the remaining choice will be value laden. The net difference between costs and traveler 16 savings represents the amount of unoffset cost which must be balanced by other benefits not measurable in money units if a project is to be considered a worthy investment. This information should encourage the search for those benefits, as well as helping the decision makers to make the ultimate judg¬ ment on project worthiness. Through this approach, the value-related un¬ certainty is by no means eliminated, but it may be reduced to the point where the choice becomes somewhat clearer. The role of the political process will remain strong, but the separation of issues of value and fact becomes clearer. Still, it would be naive to expect that this or any other planning strategy would eliminate the tensions between politics and analysis. The basis for those tensions is, ultimately, the competition for decision making power. While in the ideal one might argue that all of the decision power truly lies with elected officials, the nature and content of the analysis process will influence critically important issues, such as the alternatives considered and the measures of their costs and effectiveness. Managing these tensions between planning and politics is a challenge, but the tensions them¬ selves are a fundamental attribute of the relationship. The initial political response to the CEAA concept at the local level reflected these tensions, for the CEAA guidelines drive and constrain the analysis process, and thus establish some limits on the power of politics. In the end, however, a clear and strong argument can be made for creative, objective analysis of alternatives such as that called for in CEAA, and par¬ ticularly for the explicit consideration of the worthiness of competing pro¬ jects . The response to a proposal to establish a similar analysis process for urban highway investments represents a case where this competition for power produced a different outcome (Federal Register, December 7, 1978). The strong reluctance to give up the local flexibility found in formula allocation of 17 highway funds led to sufficiently powerful opposition to the proposed analysis requirement to lead to the death of that concept. Of course, the same argument for the merits of quality analysis could be made in the case of highways, along with the argument that it makes sense to treat potentially competing highway and transit investments in a corridor in the same way. Per¬ haps over time, and through advances in the state of professional practice, the results of CEAA can be used as a good example to move toward a consistent pattern of analysis and evaluation across all modes. Indeed, such a proposal, modified from the 1978 policy, was reintroduced in 1980 (Federal Register, October 30, 1980), only to be postponed pending regulatory review (Federal Register, March 30, 1981). While such a regulation is unlikely to be imple¬ mented in the foreseeable future, the fact that it continues to be proposed suggests the perception on the part of some of the value of more consistant intermodal investment analysis. Chapter Two CONCEPTUAL FRAMEWORK FOR WORTHINESS ANALYSIS Introduction Project worthiness analysis, as the term is used in this report, is concerned with determining, or helping to determine, whether or not proposed projects are sufficiently attractive to justify their associated cost. More specifically, worthiness analysis explores the extent to which the returns, or benefits, from a project are, in the eyes of the investor, large enough to make its total investment costs worthwhile. The information contained in a successful worthiness analysis, one which accurately captures in a common metric all of the relevant benefits and costs, is of unique value to planners and decision makers, for it should make clear which alternatives are truly worthwhile in the narrow sense that the investment required to implement them will bring a profit to the investor. This informational product will be termed "absolute" worthiness. Most cost- effectiveness evaluation studies, for several reasons, are useful only for determining the relative attractiveness of alternatives. Thus, it is pos¬ sible to determine which alternatives are preferred, but not which ones are profitable in terms of providing an excess of benefits over costs. In a case where only relative project worthiness is known, it is cer¬ tainly possible to make a selection, but whether or not a project is worth its costs is determined implicitly by the outcome of the choice process, not by analytic results. Thus, the absolute worthiness of the selected alter¬ native is an assertion on the part of decision makers which cannot be justi¬ fied before implementation. Furthermore, rarely is the effort made to document absolute worthiness of an implemented project after the fact. 18 19 Worthiness and Cost-Effectiveness Analysis Figure 1 relates these concepts to established tools and procedures. Absolute worthiness analysis is addressed by the methods of economic analy¬ sis or engineering economy, including benefit-cost analysis and related tools. Such methods bring together comprehensive estimates of all relevant benefits and costs, measure them on the same scale (e.g., monetary units), and aggregate them to produce a unitary or scalar measure of project worthi¬ ness. This unitary worthiness measure is on a cardinal scale, on which it is possible to illustrate how each project is related to a predefined absolute break-even or zero point (e.g., benefits = costs), as well as how each alter¬ native is related to all other options. Knowledge of where a project fits relative to the breakeven point per¬ mits us to know whether or not each project is worthy in an absolute sense. Knowledge of the relationship among projects permits the selection of the most preferred alternative in the sense of measured benefits and costs. Of course, it may be quite reasonable for decision makers to select any of the worthy projects (those on the favorable side of the breakeven point), using cri¬ teria not included in the worthiness analysis, with the assurance (within the range of accuracy of the analysis) that the chosen option is worth its costs. However, if decision makers are willing to accept the limitations of the measurement framework of the worthiness analysis, choosing the most worthwhile (profitable) option which gives an aggregate return greater than costs would be logical. Figure 2 illustrates the concept of absolute worthiness analysis. The do-nothing is normally defined as producing no incremental benefits and incurring no incremental costs. In this example, alternatives B, C, and E are worthy, in that their benefits equal or exceed their costs. Alternatives 20 Analysis Inputs Project benefits and costs ABSOLUTE WORTHINESS ANALYSIS Methods of Economic Analysis (Engineering Economy) Benefit-Cost Analysis Rate Of Return Net Worth Benefit/cost Ratio Analysis Outputs e Relationship between benefits and costs for each alternative • Worthy Alter¬ natives « Preferred (most worthy) alter¬ native ( s ) Analysis Inputs Measures of costs- effectiveness RELATIVE WORTHINESS ANALYSIS Cost-Effectiveness Analysis Analysis Outputs • Relationship between costs, effectiveness for each alternative • Relative attrac¬ tiveness of alternatives Figure 1. Comparison of Absolute and Relative Worthiness Analysis 21 Figure 2 Graphical Illustration of Absolute Worthiness Analysis 22 A and D, in terms of the information in Figure 2, are not worthy, although they may involve benefits not included on the graph which make them viable. It may be argued that, all other things being equal, alternative C is pre¬ ferred to alternatives B and E in that C provides a greater excess of bene¬ fits over costs. Figure 3 illustrates, in the simple case of two dimensions, the concep¬ tual results of a cost-effectiveness analysis of relative worthiness. Here the trade-offs between costs and effectiveness among the alternatives are shown. These results may lead to the rejection of D, indifference between A and B, and perhaps the preference of C over E. Yet there is no assurance that any of the alternatives is worth its costs. Further, even given uncer¬ tainty and the difficulties of monetary measurement, there is no signal of the degree of separation of benefits and costs of the alternatives. Thus, there is a serious risk of choosing an attractive, but unworthy, alternative. Although the absolute worthiness procedure seems simple enough, it is rarely applied in the case of major (urban transportation) projects because it is impossible to perform a sufficiently comprehensive analysis to pro¬ duce results in which decision makers will have sufficient confidence to make a choice in the absence of additional information. Among the reasons for this are the following: (1) All of the relevant benefits and costs cannot be measured in ob¬ jective units. For example, aesthetic impacts are by definition qualitative. (2) All of the relevant benefits and costs cannot be valued in commen— surage (e.g., monetary) units. For example, air quality improve¬ ments are not readily evaluated in acceptable money units. (3) Estimates of many benefits and costs are highly uncertain. A good example is land development benefits. Effectiveness " Do-Nothing -1 L. Cost Figure 3 Graphical Illustration of Cost-Effectiveness Analysis 24 (4) Even where the requirements stated above are met, the resul¬ tant unit values for benefits and costs may not reflect decision makers' perceptions or values. For example, the political bene¬ fits of construction employment may exceed the monetary worth of that employment. The result of these and other limitations of economic worthiness analy¬ sis make it dangerous to force project evaluation into this narrow framework alone (Thomas and Schofer, 1970). Simply stated, a feasible absolute worthi¬ ness analysis can, at best, indicate the relationship between some of the costs and some of the benefits. Such a limited information set precludes the use of this approach for selecting the preferred alternative or for assessing that any of the alternatives are worthy investments in an absolute sense. If this were the whole of the argument on this issue of project evalu¬ ation, the best possible advice would be to use some version of cost- effectiveness analysis, comparing a set of measures of costs and effective¬ ness for each of the alternatives. This comparison could — and does — take advantage of a variety of techniques for exploring the interrelationship between alternatives (Beimborn, 1976). Although the absolute worthiness of any option couldn't be known with certainty, decisions would be feasible and some good choices would be made. The operative and potentially serious risk is that important and useful indicators of worthiness, as differentiated from true measures of worthiness, would be lost. The philosophical approach proposed here is that it may be beneficial to pursue a partial worthiness analysis, within the frame¬ work of cost-effectiveness, to support decisions which explicitly recognize the importance of the worthiness question. 25 This conceptual approach is illustrated in Figure 4. In this approach, there is a definite and explicit exploration of absolute worthiness, using tools of economic evaluation (e.g., benefit-cost analysis). A comprehensive cost-effectiveness analysis is conducted: as a component of this, a neces¬ sarily partial absolute worthiness analysis is also pursued. The degree of ambiguity of the latter will vary with the context, but in all but the sim¬ plest cases the worthiness results will not be definitive. They should, however, provide a useful information set regarding the bounds within which uncounted benefits and costs must fit to assure worthiness. In particular, the results should suggest the degree to which some of the costs are offset by some of the benefits. The relative worthiness of alternatives is con¬ sidered in the context of an absolute worthiness analysis, so that it becomes somewhat clearer how large uncounted benefits, or unmeasured costs, must be to make a proposed project worthy or unworthy. This concept is the basis for the analysis approach developed in this report. It can be implemented by combining both a cost-effectiveness and absolute worthiness analysis in parallel, as suggested by Figure 4. The choice of an alternative for implementation is still clearly sub¬ jective, but, through this approach absolute worthiness is explicitly con¬ sidered and at least some of the uncertainty regarding this key aspect of alternatives should be eliminated. Structural Formulation of Absolute Worthiness Analysis The various methods of project worthiness analysis are based on the straight-forward, private sector investment criterion requiring that a preferred investment be profitable: that it return to the investor a pack¬ age of benefits equal to, or greater than, its costs. When, in the simplest case, benefits and costs can all be characterized in equalivalent monetary 26 Analysis Inputs Cost-Effectiveness Framework (Relative Worthiness Analysis) Measures of Costs, Effectiveness e Dominance and Sensitivity Analysis o Efficient Frontiers and Trade-off Analysis • Partial Cost-Effectiveness Analysis • Partial Worthiness Analysis - Benefit Estimates - Cost Estimates - Partial Net Worth - Determination of Unoffset Costs -Exploration of Project Worthiness Analysis Outputs Relationships Between Alternatives Apparent Worthi¬ ness of Preferred Alternatives Figure 4 Integration of Cost-Effectiveness and Absolute Worthiness Analysis 27 terms, this requirement can be written as b. > c. ; x — i Where b^ = Aggregate benefits of project i c^ = Aggregate costs of project i. Furthermore, when several mutually exclusive projects are considered, the logical decision criterion is to select that project with the greatest excess of benefits over costs, or: Maximize (over all projects i) [b^ - c^] Subject to the constraint that, for the preferred project, b. - c. >0. i x — This is known as the net worth criterion, for b. - c. is the net worth of x x project i. It can, with equal validity and equivalent result, be formu¬ lated in terms of discounted present worth or equivalent annualized worth. While there are alternative algebraic formulations which will produce equivalent results, there is a powerful underlying logic to this generic approach: the preferred investment is that which returns the largest (non- negative) profit to the investor. In the private sector, a firm cannot continue to invest in actions which do not meet this requirement. Application of this conceptual approach in quasi-private sector situations, where prices of services can be adjusted to approximate marginal costs, may be simplified without loss of logic by interpreting revenues collected from those who purchase service as lower bound indicators of benefits. Thus, the criterion reduces to measures revenues net of costs. In the more typical urban transit case, where prices are not equated to marginal costs, this simplification cannot be applied. 28 Characterization of Benefits in Public Sector Analyses This generic approach has for several decades been translated to public sector applications, in many fields but principally in water resources and transportation, with only a few changes in its orientation. The most impor¬ tant change is the recognition that the investor must be broadly defined to represent society as a whole, rather than a single firm. This suggests that there is a need to search out all entities or actors who benefit, and all who bear the costs, in order to arrive at a complete tabulation of aggregate net worth. For example, the classical benefit-cost analysis of the Victoria line subway in London (Foster and Beesley, 1963), examined travel cost impacts of the new line on the following groups: 1. Travelers diverted to the Victoria Line a. from other subway lines b. from commuter rail lines c. from buses d. from autos e. from walking 2. Travelers remaining on their original modes a. other subway lines b. commuter rail c. buses d. auto 3. Generated (induced) trips on the new line. Similarly, investment costs must also be sought out and accumulated across all relevant entities. These must include all incremental costs to provide and operate each option, regardless of who pays them. The more global view of the public sector analysis also requires, or 29 permits, the consideration of only net impacts as a whole. As a result, some costs and benefits which are merely transfers, flows of funds or values from one pocket to another, can be ignored. For example, taxes or fare payments collected by a public agency may be viewed as a cost to those who pay and a benefit to those who collect, but they are equal and offsetting, and may thus be ignored in this aspect of the analysis. Simply stated, expanding the boundaries of the analysis causes some important value flows to become irrelevant as long as they remain within the system of concern and are exactly offsetting. The public sector view also requires the consideration of benefits and costs even where no money changes hands. This begins to complicate the analy¬ sis considerably. For example, considering a realistic demand relationship, through which consumers display some sensitivity of their behavior to price, a service improvement is likely to leave at least some travelers with excess value, above and beyond the price they pay for their trip. This is illus¬ trated in Figure 5. The proposed introduction of a new facility would be expected to reduce the equilibrium price from p^ to p , which would be expected to increase the demand volume from v^to v^- Traveler r experiences a direct cost saving equal to p^ - p^, and this is her benefit. Traveler s was not traveling, and thus paid nothing, under the existing system. He pays, under the im¬ proved system, p . While his travel cost has increased by p2# we may pre¬ sume that the trip he makes on the new system has value to him; in particular as this example suggests, that value is at least p , the price he was willing to pay for this trip. Yet he only pays (

and thus he is left with a net benefit, called consumer surplus, amounting to Ps - p . 30 Price of Service* Demand Relationship ^Traveler r Equilibrium (Price, Volume) under existing system Traveler s Traveler t Equilibrium (Price, Volume) under proposed system 1 2 Number of Travelers per Unit Time Figure 5. Illustration of Monetary Measurement of Traveler Benefits (*Price, here, should be interpreted as the full price, in time, money, etc., valued in money units) 31 Thus, even though traveler s is not involved in a cash transaction other than paying the travel price p^ » he receives a consumer surplus which must be accounted for in a public sector, or social, worthiness analysis. It should be noted that traveler t is paying an amount p2 which appears to be equal to his willingness to pay for (or value of) this new trip. His consumer surplus is p2 - = 0. while some have argued against considering consumer surplus in trans¬ portation project benefit estimation, it seems essential to include it where there is real potential for increased travel (new trips) for otherwise these trips would show up on the debit (cost) side of the ledger. This is illogical, for it suggests that the trips have no value which implies an irrational traveler, which in itself makes no sense. The need to pursue consumer surplus seems especially great in transit alternatives analysis, where significant numbers of new trips may be quite likely. While a public sector analysis must search out consumer surplus, it is generally not necessary to know the willingness to pay of each traveler, or the precise shape of the demand relationship. If we assume that the de¬ mand function has the general, well-behaved shape shown in Figure 5, and if we assume that it is approximately linear in the region of interest, then the traveler benefits (cost savings) may be shown as: v (p - p ) benefits to existing travelers (real cost 2 savings) (v - v ) (p ~P2) benefit to new (generated) travelers 2 1 (consumer surplus) or ^(v + v )(p ~P2) total benefits to all travelers on 2 the facility in question. 32 Multiple Counting and Measurement Choices Of course there are likely to be many other categories of benefits, and costs, of importance in the assessment of a major urban transit investment, and this approach calls for them to be identified and included in the analy¬ sis. A significant danger in the search for benefits is multiple counting. If improvements in service produce traveler benefits and increased accessi¬ bility and enhanced land values, care is required to avoid counting the same benefits several times. For example, if (or to the extent that) accessibility improvements are reflected in more or cheaper travel — the principal reason they should be of value — than direct traveler benefits, savings in costs and increases in consumer surplus should measure some (or all) of the accessibility benefits. Putting a separate and distinct value on accessibility (including using it as a non-monetary measure in cost-effectiveness analysis) amounts to double counting of benefits. If, of course, an increase in accessibility over and above that reflected in usage of the transit system is valued, this accessibility increment can be included among the effectiveness measures. Such a choice, however, should be made carefully and explicitly. Virtually the same argument can be made for including both estimated demand volume and an accessibility measure such as number of regional employment centers served in cost-effectiveness analysis. While in some cases it may be legitimately argued that it is valuable to the community to have a bus running down a street with no passengers (accessibility without traveler benefits), this argument is best made openly rather than by implication. Similarly true increases in land value which result from transportation service improvements (cost reductions) are likely, in the long run, to be traceable to real traveler benefits. While it is possible that land market 33 values may increase at the promise of a major transit improvement, it seems unlikely that such benefits will be sustainable over time if the potential ridership, limited by traveler benefits, does not materialize. This sug¬ gests that a conservative, and perhaps realistic, approach to benefit estimation will be focused on expected traveler benefits (reductions in total costs of travel and increases in consumer surplus). In this example, it should also be recognized that accessibility improve¬ ments would be expected to be strongly collinear with land value increases, and thus using both of these measures is another source of multiple counting. Where improved access, as distinct from increased travel demand, is valued by the community, it may logically be included in some fashion in the analy¬ sis. This is not a measurement (technical) decision, but a policy choice that needs to be made with a clear understanding of its implications. Figure 6 illustrates a logical framework for making such measurement choices to avoid multiple counting of costs and benefits.- It is based on the notion that, in the face of numerous rounds of causally-related out¬ comes , the analyst should attempt to cut each causal chain only once with a measure. The location of the cut is less important than assuring that the value of the impact associated with any input or output is entered into the accounting scheme only once. This is particularly important in an absolute worthiness analysis, where all of these values are aggregated to estimate a net worth. In cost-effectiveness analysis, where a dis¬ aggregate display of the attributes of alternatives is presented, multiple counting is less deceptive. Still, it would be desirable even in this case to identify the possibility of multiple counting, since the decision makers may be attempting to perform a mental aggregation of tabular cost-effectiveness displays. Second Round of Effects First Round of Effects First Round of Effects Second Round of Effects Measurement Boundary Figure 6 Making Measurement Choices Without Multiple Counting of Outcomes 35 While it is important to attempt to capture all of the (important) benefits and costs, avoidance of multiple counting is possible by under¬ standing the chain (or rounds) of effects and "cutting" each path with a measure only once (Schofer, 1978). Generally, on the benefit side, traveler benefits should be the prin¬ cipal focus in the absolute worthiness analysis. Primary environmental benefits (reductions in air pollution, noise, etc.) must also be pursued, although monetary measurement problems may preclude their inclusion in the worthiness computations. These are good candidates for treatment in the cost-effectiveness analysis. Subsequent impacts likely to derive from direct improvements in travel service and costs, and from direct environmental changes, are usually best left out of the aggregate worthiness analysis. Outcomes or effects measured close to the alternative action, direct inputs and outputs, may be termed "nearside" measures, while those impacts occuring farther down the causal chain can be described as "farside" measures. It is advantageous to concentrate on nearside, rather than farside, measures not only to avoid multiple counting. The principal argument here is that nearside measures are likely to be more accurate and more certain. Nearside measures will typically be more obvious in the planning stage and thus are predictable with greater accuracy than subsequent effects. Farside measures (especially changes in land values and use) are less certain because they are increasingly affected by factors other than the transport improvement as one moves further along the causal chain. For example, Knight and Trygg (1977) propose a useful conceptual model of the factors affecting land use impacts of transit investments which suggests that such outcomes are influ¬ enced by the land market, the money market, and government policies, inter alia, in addition to the expected change in transit service. 36 In some cases, of course, it is the farside measures which best cap¬ ture some of the key policy effects toward which major urban transit investments are directed. Two logical options are available to the analyst in such cases. First, efforts can be concentrated on these selected farside measures, eliminating or attenuating consideration of the causally related, "upstream" nearside measures. Second, the absolute worthiness analysis can be focused on the relevant nearside measures, while the cost-effectiveness analysis can be used to highlight farside measures. This brings back an issue discussed in Chapter One, the true goals for which urban transit investments are pursued. In the many cases where these goals do not focus on providing primary transit user benefits, some of the key controversies which arise over such investments may be linked to the uncertainty which arises when one focuses on risky, farside outcomes. Some agencies have chosen to avoid a serious study of the user benefits of pro¬ posed investments because of the expectation that such benefits may be too small to offset even a reasonable fraction of the costs. In the face of such threats to preferred alternatives, and given a strong desire to achieve non-transport benefits, it may be reasonable to reconsider the degree to which transit funds, may be spent on farside benefits. In such cases one may find the seeds of an argument to focus transit funds more narrowly. The adoption toward the end of the last decade of the short, 15-year time horizon for investment analysis suggests the existence of such an implicit policy at the federal level. Closure There is much useful information to support decision making which can be developed in an economic worthiness analysis. At the same time, there are important conceptual and pragmatic limitations associated with this 37 approach, especially when it is applied alone. The next chapter discusses some of these issues and offers some approaches toward dealing with them. Chapter Three SOME PROBLEMS ASSOCIATED WITH ABSOLUTE WORTHINESS ANALYSIS Worthiness analysis, in the form of traditional economic analysis tools, was widely used in the evaluation of major highway capital invest¬ ments in the late 1950s and the 1960s. During this period, numerous problems associated with (though not necessarily caused by) this evaluation strategy arose, which led to a virtually complete rejection of the use of economic evaluation techniques for urban transportation evaluation in more recent years. This section discusses some of these issues, explores their causes, and poses some solutions to them. Problems of Measurement and Valuation Economic evaluation is inherently narrow in its consideration of impacts and costs of investments because it requires that all items included be characterized in monetary units. This imposes the dual necessity of quanti¬ tative measurement and monetary valuation. As a result, a number of poten¬ tially important impacts do not fit into this analysis framework. Aesthetic impacts represent a classic example of outcomes not subject to objective measurement. There are other outcomes, more significant than aesthetic changes in most cases, which are not subject to unambiguous, readily accepted, objective measurement. These include land development impacts, changes in accessibility, shifts in mobility for particular population groups, and environmental impacts of various types. While measure can be proposed and applied in each of these categories, a multiplicity of measure¬ ment options exist, and there is no clearly correct metric. The challenge of monetary valuation is even more difficult to deal with• Referring to the impact categories listed above, putting an acceptable money value on measures is complicated because, in most of these cases, no effective 38 39 money market exists for estimating values. There are a number of non-market, artificial approaches to valuation which can be, and have been, utilized to deal with this problem. Some examples may serve to illustrate these approaches : Valuation of human lives (safety benefits of transportation investments resulting in mortality reductions): Here the most common approach has been to estimate the cost of a death, in terms of pre-death medical care, adminis¬ trative costs of insurance and the discounted present worth of the expected income stream of persons killed, etc. (Fagin, 1976). The last item is con¬ troversial, in that some argue that the income stream does not capture the true worth of an individual, and others call for the cost of the consumption of the individual to be subtracted. The literature suggests that different approaches to human life valuation have led to radically different estimates. Furthermore, there is a clear political preference for avoiding setting, a priori, any money value for a life, because this tactic appears callous, and because — in the abstract — many would argue that each life is invaluable (Brown, 1980). On the other hand, every public sector choice regarding investments which save — or fail to save — lives sets an implicit boundary value on a life. In the absence of an a priori value set by policy, it is common to find that different projects, funded from the same ultimate source, are committing widely varying amounts of money to save a single life (Mendeloff, 1980). Thus, from an economic, rather than a political, perspective, setting some fixed value on a life, derived from rational but judgmental cost estimation, seems reasonable. Valuation of health improvements due to environmental and safety impacts : Here, again, in the absence of a real market for such benefits, a 40 cost estimation approach has been followed (Lave and Seskin, 1970) . Usually this has meant adding up the costs of medical care and lost wages due to various sources and periods of morbidity. As in the case of a human life, there remain serious questions regarding what costs should be included in the accounting scheme. Furthermore, estimates of changes in morbidity tend to be highly uncertain, particularly in the case of changes in concentrations of environmental pollutants. Finally, such approaches fail to account for the consequences for the quality of life associated with various illnesses. Valuation of park land, recreational facilities, and historic places: In these cases, where user charges are levied on such facilities, it is pos¬ sible to take a consumer surplus approach, which amounts to imputing the value of the experience in terms of the (presumably) lower bound estimate provided by the fee structure. Thus, if each day 1,000 people pay $1 each to use a park , one may presume that the minimum value of this facility is $1,000 per day. Where no user charges are levied, the Commission on the Third London Airport (CTLA, 1971) presumed that the price could be imputed from the trans¬ portation costs people were willing to accept to gain access to the facility. Thus, if the person traveling farthest to the park paid a transportation cost of $5, the presumption is that such an individual accrues no consumer surplus, while other users do retain such surplus in the amount of $5 less their actual travel costs. This is, at best, a gross approximation, in part because it presumes that all users value the facility at the same level, and that people select residential locations independent of their preference for recreation. Valuation of community facilities in terms of cost of provision: Mao (1966) proposes that one way to value public facilities and services which 41 are not directly priced is to assess them at their direct cost of provision. Thus, a new school or community center might be valued at the cost of pro¬ viding it (land acquisition, capital cost, discounted present value of operating and maintenance cost). This presumes that there is no consumer sur¬ plus associated with such investments, or stated differently, that their net worth at the time of opening is zero. One would hope that this valuation basis produces a significant underestimate of the value of such facilities and services, and thus this appears to be a poor alternative. Cost avoidance as a benefit measure: Some studies have included as a measure of benefit the costs of other alternatives which would not need to be built as a result of the choice of a preferred option. The most typical example is the case where the costs of building an urban freeway, as an alternative to a transit investment, are counted as a savings (benefit) to the transit option. This is an erroneous approach, in that it presumes that the highway would certainly be built were the transit option not chosen. This is a dangerous assumption in most cases and, where it is used, it is only rational to count the benefits associated with the highway investment as a cost to the transit option. This approach abounds with poor logic, and it clearly serves to bias the outcome (Schnaiberg, 1980). It amounts to defining (in this case) the highway option as the null or base¬ line alternative, which is generally incorrect. Travel time savings valuation: Since the beginnings of the use of economic evaluation in the transportation field, travel time savings have been valued, at various levels and on different bases, in monetary units. The principal approach has been to derive an implied value of time from behavioral observations, where travelers face the option of spending more money to save time. Results have indicated some significant variations of 42 implied time valuation across trip purposes and income categories (Stopher, 1975). Typical products have thus been in the form of time values by trip purpose as a function of wage rates. Some researchers have offered the hypothesis that individuals' budgets are essentially fixed in magnitude, and thus transportation policies which reduce travel time and costs for some trips may lead to increased expendi¬ tures on other trips (Zahavie and Talvitie, 1980). This could happen, for example, if travelers experiencing a savings due to the implementation of a new fixed guideway transit line were to decide to move further from the central city, maintaining the same travel time but collecting their bene¬ fits in other dimensions (e.g., improved residential amenities, better municipal services, etc.). However, even if the hypothesis of a fixed time budget were true, that a traveler who experiences a reduction in travel time or cost for some trips receives a tangible benefit. This increment in value could be retained as a savings, essentially increasing income, or it could be spent on other commodities or services, including travel. The result, no matter what the outcome, is an increase in utility to the traveler who receives the time/cost savings. It is this increase which should be measured and credited to the transportation investment under study. The only issue to be resolved is valuing these savings. Because of these variational patterns, and because of uncertainties in data and behavioral theories (e.g., is the value of 60 people saving one minute the same as one person saving 60 minutes?), there has been a growing reluc¬ tance to apply monetary values of time savings in urban transportation invest¬ ment analyses. Instead, the emerging pattern seems to involve an estimation / / and evidence is available to the contrary (Downes, 1980), 43 of total time saved without monetary valuation. In some cases, policies have been established to ignore small time savings (e.g., 5 minutes or less) on the presumption that there exists some threshhold level of time savings below which travelers either do not perceive the savings or do not respond to the benefit. On the other hand, it is reasonable to argue that, in this case, there really is an operative market, for travelers commonly face the economic decision regarding spending money to save time. Such decisions are made and, if it is possible to hypothesize economically rational behavior (which is an underlying assumption of travel demand forecasting in general), then the imputed values for time may be reasonable. Furthermore, to the extent that the results of travel forecasting models are acceptable, and indeed, they are the fundamental basis for planning large scale transit investments, it is quite feasible to predict travel time savings. The degree of uncer¬ tainty in these estimates is on the same order as that associated with any of the products of demand forecasting. Getting an accurate esti¬ mate of time savings may require somewhat more careful accounting schemes to identify classes of travelers, trips, and magnitudes of savings (if the threshhold theory is accepted), but existing modeling procedures can, in fact, support such analysis. The uncertainty associated with estimates of the value of travel time can be dealt with relatively easily through the use of sensitivity analysis. For example, three values of time (for each traveler/trip class) can be applied, a middle value, and values 25 percent higher and lower. The ranging itself is easily accomplished computationally, since it merely amounts to factoring aggregate travel time savings benefits by (in this example) 1.25 and 0.75. The result will be three levels of time saving benefit estimates. 44 Generally speaking, sensitivity analysis represents a constructive way to deal with uncertainties in any benefit and cost measures. This approach becomes increasingly less effective as the number of parameters to varied increases beyond two or three. One constructive option for dealing with this (utilized in the software package produced in this research) is a simplified scenario approach, wherein a selection of the possible combina¬ tions of parameter values is utilized to reduce the number of estimates provided and to span the space of reasonable combinations. For example, if the sensitivity of user benefit estimates to assumptions about values of time and vehicle operating costs is of interest, the fol¬ lowing levels of parameters (and resulting combinations) may be considered: Average Value of Operating Costs $/mile Average 0.15 0.20 0.25 Value of 3.75 B.,, B B 11 12 13 Time 5.00 B^ B^ B^ $/hour 6.25 B31 B^ B33 In this table, B represents the benefit estimate for a given unit cost assumption. Nine benefit levels is far too many to work with. Possible compromise selections which appear sensible include: B , B B B 11 13' 31' 33' which is a "corner point" selection; and B^, B^, B , which is a "high, middle, low" selection. 45 In essence, the cojnbinatorial problem can be ameliorated (but not eliminated through the choice of a small number of logical combinations which will illustrate the likely range of the output measure, in this case, traveler benefits. Measurements and Uncertainty*: Problems associated with the measure¬ ment of benefits (and costs), particularly in cases where measures are avail¬ able but their validity is questioned, might generally be treated in terms of the gross uncertainty associated with the forecasts of any measures of interest in CEAA. Forecasting errors are unavoidable, and they are likely to be relatively large, with their magnitude increasing with both the time span of the forecast and the magnitude of the technological and service changes planned, while the nature of these errors is not yet well under¬ stood, recent studies of travel behavior suggest their magnitude (Koppelman, 1975) and some have argued that certain kinds of errors (especially under¬ estimates of costs) are not unintentional (Wachs and Ortner, 1979). A useful and simple tactic for dealing with such uncertainty problems is to establish a gross, flat rate error range, and to ignore differences between alternatives within this range. The Table 1 illustrates this notion. If, for the sake of this example, the expected error rate is + 10%, which nominally is a low value for this parameter, the range of estimates is shown in Table 2. The concepts in this section are not, of course, limited to absolute worthiness analysis. 46 Table 1 Example of Measures of Costs and 3cnefits Measures Annual User Travel Time Savings (Smillions) Annual User Operating Cost Savings ($millions) Total Capital Costs (Smillions) Annual O&M Costs ($millions) Total Annual Costs ($millions) Net Annualized Worth (Partial, ?millions) Alternative Al A2 A3 45 50 30 40 48 30 580 600 650 36 40 50 88 95 100 -3 +3 -40 47 Table 2 Example of 10% Error Range of Measures of Costs and Benefits Alternative Measure Al A2 A3 User Time Savings [40.5,49.5] [45,50] j^s0s-r User Operating Cost Benefits [36,44] [43.2-52.8] Captial Costs [522,638] [540,660] [585,715] O&M Costs [32.4,39.6] [36,44] [45,55] Total Costs [79.2,96.8] [85.5,]04.5] [90,110] Net Worth [-20.3,+ 14.3] [:16^ l-T-^T 48 Based in this information, it may be reasonable to argue that Al and A2 are not measurably different from each other. Both, however, seem superior to alternative A3. One might argue that Al and A2 are simply equivalent alternatives. A more constructive statement might be one of the following: (a) The choice between alternatives Al and A2 should be made in other explicit measurement dimensions, since the measures shown in Table 2 suggest these options are otherwise equivalent. (b) The decision makers should feel comfortable choosing, from Al and A2, whichever suits their interests and needs, explicit or otherwise. Decision makers may find it comfortable to accept such guidance. More importantly, it seems ethically appropriate to highlight such cases of un¬ certainty-based indifference, which is often contrary to common practice. That is, it is not unusual to find situations where a major issue is made over alternatives which differ by only a few percentage points in the key measurement dimensions (Litchfield, 1971). It should be noted that, in the context of cost-effectiveness analysis, this notion of uncertainty-based indifference bands may provide the basis for deleting some measures of effectiveness and cost which fail to differ¬ entiate between alternatives. Deletion of such measures serves to reduce the information burden on decision makers, allowing them to concentrate on the measures which are more useful in the choice process. Treatment of Measurement Problems — Closure: While the measurement and valuation problems faced in the application of absolute worthiness analysis are important, we submit that they are not insurmountable. Where reasonably acceptable measures and values can be established, at least within the range of uncertainty of other analytic products, they should be applied, and the appropriate caveats should be displayed. 49 Where measurement is possible but more uncertain, and/or where unit impact values are uncertain yet subject to estimation, creative use of sensitivity analysis offers a reasonable avenue to the pursuit of the worthiness analysis. Finally, where either measurement or monetary valuation is impossible, the impact category must be excluded from the worthiness analysis. This need not mean that the impact is ignored. Instead, if it is important, it should be treated within the broader context of cost-effectiveness analysis, which can accept nominal measures and measures in "natural" but non-monetary value units (e.g., tons of air pollutants, number of dwelling units taken, etc.). As long as the cost-effectiveness portion of the evaluation is given appropriate weight in the presentation of analytic products, this offers a relatively safe approach to retaining the absolute worthiness analysis, though it will certainly remain a partial estimate of net worth. Thus, it may be argued that, particularly in the area of measurement and monetary valuation, some of the major errors made in past uses of absolute worthiness analysis have been in the application, rather than coming from the technique itself. Careful choice of measures and values, recognition of uncertainty, and the presentation of important non-monetary measures in the cost-effectiveness analysis, should help to alleviate many if not most of these errors. Lack of Explicit Policy Orientation A kèy limitation of absolute worthiness analysis is its lack of explicit policy orientation. That is, the technique is structured to search for (monetary) benefits and costs, and to provide an assessment of options in those terms alone. However, it is most common for communities to consider major transit improvements in order to achieve a variety of other benefits, 50 as a matter of policy, including inter alia downtown preservation and re¬ development, corridor densification, and the expansion of employment. A problem arises to the extent that achievement of these essentially non-transport goals is not reflected in increases in traveler benefits. In such cases absolute worthiness analysis cannot be relied upon alone to select the preferred alternative. It should be evident, of course, that traveler benefits are sometimes useful indicators, if not direct measures, of the attainment of some key (non-transport) policy goals set for transit investments. This may be particularly true in the case of urban development, redevelopment, and preser¬ vation goals, for it is reasonable to expect that attainment of such goals would be correlated with increases in transit ridership and thus traveler benefits. (Of course, the presumption is that the rational traveler will utilize an improved mode, or benefit by remaining on an existing service which improves due to the introduction of a new one, because some service improvement — traveler benefit — is derived from this choice. Behavioral travel demand models predict outcomes of service changes based on this assumption). A serious limitation of the use of traveler benefits as indicators of policy success lies in the uncertainty associated with predicted develop¬ mental (and other non-transport) outcomes. For example, it is fairly clear that our ability to forecast urban spatial development impacts of fixed guideway transit investments is rather limited. Here there are at least two promising options. One is to adopt a scenario-based approach which amounts to performing a sensitivity analysis on the level (and arrange¬ ment) of expected development impacts. The other calls for maintaining a strong federal policy of seeking (virtual) guarantees, through policy 51 commitments and promises of delivery of other required, supportive actions (tax policies, zoning changes, etc.), that expected developmental impacts will occur. A proposed investment oriented primarily to producing such impacts, however , may not be particularly attractive in terms of its user benefits, or in terms of its net worth. An underlying cause of this phenomenon is that at least some of the investment costs are, in reality, allocated to producing the development (or other) benefit rather than to causing an increase in traveler benefits. Thus, if the net worth analysis shows a result such as : net worth = traveler benefits - total costs -$33.0 m $25 m - $58.0 m this may simply mean that some of the $58.0 million cost is not truly allocatable to the production of the $25.0 million traveler benefit. Put differently, the same traveler benefit may be achievable with a smaller investment which (presumably)does not produce the desired non-transport benefit derived from the proposed action. Obviously it is critically important that this likelihood be recognized explicitly. The net worth (in this case) of -$33.0 million represents the degree to which traveler benefits offset investment costs. The question still to be answered is whether or not other, non-monetary, and relatively uncertain, benefits offset the remainder of the costs. The approach of imbedding the absolute worthiness analysis in a broader cost-effectiveness analysis offers a way to deal with this question explicitly. To accomplish this, it will be essential to specify the non- transport (e.g., development, employment) policies set for the proposed transit investment in clear terms (see Chapter 6). That is, it will be 52 important at the outset of the analysis to state the goals to be achieved, so that a logical and directed evaluation can be conducted. Such clear statements of policy intent should be supportable, and supported, by sharp measures of effectiveness which will facilitate the evaluation and decision process. Such measures, in the case of urban development, should focus on such readily quantifiable issues as the magnitude, type, and location of expected developmental impacts. It then becomes feasible to include measures of the attainment of key policies in the cost-effectiveness analysis itself. The absolute worthi¬ ness analysis can serve the function of abstracting some of the key benefits, and perhaps all or most of the costs, from the cost-effectiveness presentation to develop a partial worthiness assessment. The result, in most cases, is likely to be a rather clear specification of unoffset costs (negative net worth) which must be counterbalanced by other effectiveness achievements for a proposed action to be worthy of implementation. The choice will ultimately be subjective, and political in the best sense, but the boundaries of the choice issues should be more clearly delineated as a result of this approach. Thus, absolute worthiness analysis can contribute to sharp policy analysis, but only when it is strongly supported within a cost effective¬ ness framework, within which the policies themselves, and the measures of their achievement, are indeed sharp and explicit. Treatment of Distributional Effects It is well known that absolute worthiness analysis, because of its aggregate nature, covers up the distributed pattern of benefits and costs. This is particularly troublesome, since transportation investment choices almost always involve important decisions about where (or on whom) to focus 53 benefits, and how (to whom) to allocated costs. Yet absolute worthiness analysis, in order to estimate a total net worth of alternatives, treats equal benefits (or costs) to different individuals or groups as if they were equally valued, which may not be the case. Furthermore, it commonly makes a difference to political decision makers if the groups who benefit and the groups who pay are not the same. Finally, it is becoming increasingly important to assure that public sector programs promote equity in the delivery of services and availability of facilities, as reflected in policy statements, major conflicts, and court cases (Jones, 1979). Equity describes a particular kind of distributional pattern, under which all groups are made (in some sense) equal. Equality may reflect a longer term goal, in which case the concern may focus on making an existing distributional pattern more equitable. A project which achieves this will likely be, viewed alone, quite unequitable. Alternatively, equity may be measured in terms of each separate project, in which case the overall pattern of distributed effects will (probably) not be changed by the investment process. There are several ways in which distributional effects may be viewed and equity assessed, and each has different ramifications for the trans¬ portation system (Savas, 1978). Among these are the following: (a) Output measures, such as magnitude or quality of service. For example, equity might be defined in terms of equal waiting times at or walking distances to transit vehicle access. A variation of this might be to measure service relative to demand, e.g., vehicle-miles per capita by subarea. (b) Input measures, such as investment costs. For example, equity might be characterized in terms of equal expenditures for transit in each subarea, or on a per capita basis. The cost allocation accounting problems in using this approach will be considerable 54 and, because of widely varying conditions (e.g., population density), service quality may vary widely under this criterion, (c) Impact measures, such as ridership or user benefits. Here, equity might mean tuning service to achieve the same transit share of the travel market in each subarea. Theoretically, absolute worthiness analysis deals with distributional effects by proposing that transfer payments be made, preferably in reality, but optionally only in the analysis, to compensate for inequities in the distribution of benefits and costs. This is normally not done in practice because of the complexity of the accounting activities required and, more importantly, because of the difficulty associated with placing values on some of the key impacts calling for such transfer payments. On the other hand, evaluating, and selecting the distributional pat¬ tern of benefits and costs has been established as the traditional role of the representative political process in the United States. Elected officials hold the responsibility for such allocative decisions, and, quite desirably, their job security is based on an ability to deliver an acceptable distri¬ butional pattern of benefits and costs. Of course, acceptable means that the majority of the electorate (or in some circumstances a plurality) finds the overall pattern reasonable. Major inequities remain, although it is not clear that any planning effort can effect such situations beyond dimply displaying the resultant or expected patterns. The point remains that politics can, and must, make the fundamental allocative decisions. While absolute worthiness analysis cannot respond to this issue, analysts can and should produce other useful information to support such choices. It should be recalled that the political process typically will not attempt to achieve anything like true equity (however defined) in a single 55 transportation decision. Instead, elected officials are likely to be more interested in the distributional effects of a series of actions covering various investment sectors. Thus, it may not be productive to be overly concerned with equity itself in evaluating a major transit investment. Instead, it is probably more important to report distributional patterns so that they may be assessed explicitly, rather than to try to characterize equity, where the criterion itself is likely to be a serious policy matter, or to try to achieve a condition of equity in a single action. This does not mean that specialized equity goals, such as providing equitable service to the elderly and handicapped, will be unimportant. Indeed, using major public investments to rectify selected inequities is common. Here, again, it will be important to specify such goals at the outset, and subject them to public scrutiny in the design, evaluation, and decision process. In general, the nature of the distributional issue, as well as the limitations of absolute worthiness analysis, demand that this question be resolved in the political process. This is best supported by reporting selected distributional measures in the cost-effectiveness analysis. This can be accomplished by carefully defining the appropriate distributional criteria, selecting a small set of measures of effectiveness, perhaps only one or two. It will also be necessary to select the aggregation units or incidence groups. The choice of groups and measures must in most cases be made jointly, since they will together define data requirements and both will be influenced by the attributes of the forecasting process. Probably the most useful and defensible measures for CEAA will be mea¬ sures of service magnitude or quality (e.g., vehicle miles, seat miles, both related to a service period such as a day or the peak period). The 56 easiest aggregation units will be geographic, which can normally be related to political jurisdictions (wards, neighborhoods, city boundaries), which in turn are likely to be of interest to elected officials. Describing distri¬ butional effects for each socio-economic group, particularly income groups, is also likely to be of importance, though from an analysis perspective this may be more difficult. However, since much of the travel demand forecasting process uses income as a variable, ridership measures may be reported by income without much conceptual difficulty. Service measures will be somewhat more difficult to report by income because service is usually more geographically defined. The spatial patterns of socioeconomic groups in U.S. cities today tend to be relatively well-defined in a geographic sense, making it feasible in most cases to work with geographic areas in distributional analysis while still providing an indication of the socioeconomic impacts. Patterns of distributional effects are probably most easily understood if presented in graphical form (e.g., maps). Tables offer an option to this, but they may not promote a clear understanding of the differences between service quality in each subarea. While aggregate measures of distributional patterns have been suggested, it is likely that, in inter¬ actions with political decision makers and their constituents, such measures may defeat the purpose of distributional analysis because their meaning may be relatively unclear to most laymen. One advantage of pairing the graphical or tabular, disaggregate dis¬ tributional analysis with absolute worthiness analysis is that this may help to illustrate the costs (in terms of net worth) of achieving certain distributional patterns. That there are indeed such costs is highly likely, particularly if some alternatives are derived from particular, a priori 57 distributional objectives. This is because, in concept, the worthiness analysis will seek out the option with the highest aggregate net worth. Establishing an a priori distributional requirement amounts to introducing additional constraints on the "optimal" solution, which can be expected to reduce the aggregate net worth (objective function value). One must recognize, of course, that to the community and its elected officials, certain distributional patterns may be preferred, and thus offer an added value not reflected in project net worth. While at the outset this added distributional value will generally not be known, it may be possible for decision makers to judge whether a particular distributional problem is worth its costs in net worth reduction from other options with less attractive distributional outcomes. Absolute Worthiness Analysis and the Reality of Cash Flows Although absolute worthiness analysis deals with monetary measures, it does not, as stated earlier, deal exclusively with real cash flows. While some of the benefits and most of the costs (capital, operating, main¬ tenance) represent real cash exchanges, a major share of the benefits do not represent real, cash transactions. Instead, monetary measures are used to signify the approximate value associated with such non-cash transactions. This notion can be confusing to decision makers and their constituents, since no. true cash accounts will change in the magnitude indicated by the absolute worthiness analysis. It should be recalled, and pointed out to users of CEAA results, that monetary measurement is pursued to put as many benefits and costs as pos¬ sible on a common (money) scale. At worst, this is a convenient analytic artifact; at best, the chosen monetary unit values are useful reflections of true social worth of outcomes of the alternatives. 58 One popular way to avoid this potential confusion is to report results of the absolute worthiness analysis in terms of the benefit-cost ratio rather than net worth. The ratio form has similar information content in the limit, as shown below: B. - C. > 0 l x — B. > C. 1 — x B./C. > 1.0 l x — Where: B^ = aggregate benefits of project i C = aggregate costs of project i. while reporting the ratio instead of net worth avoids stating a monetary amount which only indicates project worth, the ratio format has serious deficiencies. For example, the benefit-cost ratio only has significance in terms of its relationship to 1.0. If the ratio is 1.0 or greater, the pro¬ ject is a worthy investment (all other things being equal). If the ratio is less than one, in the absence of other information, the project is not a good investment. The size of the ratio (greater than 1.0) is not necessarily an indicator of relative project worth. Consider the following example: Benefits Costs B/C B-C Option A 50 25 2 25 Option B 82 47 1.7 35 Option A has a higher B/C ratio. Yet this does not assure that it is preferred to option B, which has a B/C of only 1.7. The appropriate decision rule requires computation of the incremental benefit-cost ratio: Incremental Incremental Incremental Benefits Costs B/C Option A 50 25 2.0 Option B 32 22 1.45 59 This result indicates that the extra investment required to select option B over option A is a worthy one. Examining the simple benefit- cost ratio alone offers a high probability of giving deceptive advice, for it is convenient to believe that a higher simple benefit-cost ratio means a better project. While the pursuit of the incremental B/C will assure finding the best answer, maximizing net worth alone will always guarantee the choice of the most worthy option (all other factors being equal) in one step. In addition, the net worth criterion offers a powerful and obvious logic: it is best, when possible, for society to walk away from a decision with the largest profit. Furthermore, in cases where a series of non-exclusive options is being considered, the value of the overall program can be signi¬ fied by the sum of the net worths, but not by any aggregation of the benefit- cost ratios. Thus, the net worth, benefits less costs, has substantive meaning in its magnitude, gives a clearer choice of mutually exclusive alternatives, and has meaning in programming decisions (non-exclusive alternatives), all of which are not true for the ratio B/C. Perhaps most important for this report, the use of net worth as a worthiness measure provides a more natural link to the remainder of the cost-effectiveness analysis, for it gives both analysts and decision makers a clearer indication of what the cost-effectiveness analysis must — or may not — show for an alternative to be acceptable. For example, a negative net worth of $40 million says that the net non-monetary benefits must be worth more than $40 million for the alternative to be a worthy investment. A positive net worth of $35 million says that the net non¬ monetary worth can be no less than - $35 million for the investment to remain 60 acceptable. This is surely more useful than the benefit-cost ratios for such projects, which give us no clues about what the cost-effectiveness infor¬ mation may or must indicate. Stated differently, it is quite unclear to the decision maker how to weigh the ratio against cost-effectiveness information in the following table: Option Measure Baseline A B Annual Riders (millions) 65 84 102 Pollution Reduction - 220 2260 (tons /year) Residential Units - - 162 Taken Total Annual Costs $32.87m $44.97m $66.23m Annual User Benefits - $13.55m $37.63m B/C - 1.12 1.13 With an explicit net worth measure the picture is clearer: Option Measure Baseline A B Riders 65m 84m 102m Pollution Reduction - 220 2260 Residential Units - - 162 Annual Costs $32.87m $44.97m $66.23m Annual Benefits - $13.55m $37.63m Net Annual Worth -$32.87m +$1.45m +4.27m This suggests that option B gives a significantly larger net worth (though not particularly large in itself), which, along with its pollution benefits, may easily outweigh the reported relocation impact. (In this example it should be noted that the relevant costs of A and B are the incremental costs over the baseline alternative, not the full costs. The negative worth of the baseline alternative suggests a poor choice of the baseline, which requires a substantial annual investment. Surely this option must also produce some user benefits, though none are 61 reported. The implication is that the $32.87 million annual cost of the baseline is unavoidable, which is probably not the case. The result is that options A and B are made to look better, in terms of the absolute worthiness analysis, because of the high cost baseline alternative. It would be desirable to avoid such an approach to the definition of the base--' line alternative.) Returning to the original theme, the net worth is not a quantity which can be put in the bank, but retaining it in monetary units is desirable from the perspective of maintaining a high level of meaningful information in the presentation of the results and providing a sensible link back to the cost-effectiveness analysis. Working with monetary net worth also introduces the possibility of confusion between the absolute worthiness analysis and financial analysis, while the latter is not within the scope of this project, it is important to recognize that these two types of analysis are significantly different from each other: o Worthiness analysis is concerned with determining the (partial) net value of the proposed investment to society in money units. o Financial Analysis is concerned with the total real cash out¬ lay required to implement and operate the investment. o Financial analysis estimates the real bills to be paid, in current (inflated) dollars, for that is what creditors will demand. o Worthiness analysis can (usually) safely work in constant (uninflated) dollars, for it does not deal with real cash flows, but with equivalent monetary worth. Inflation becomes an issue only when the rates of inflation of benefits and costs are predictably significantly different- 62 o Fares are a key issue in financial analysis, for they are usually a major source of funds from which bills are paid, o Fares are irrelevant in worthiness analysis of public systems, since they are transfer payments from one of society's pockets to another, o A project could be financially feasible but unworthy, or worthy but financially infeasible. The outcome will rarely be obvious. o In financial analysis it makes a difference where funds come from. The local share is extremely important relative to local budgets; external shares (e.g., federal) are important in their amount but not their source, o In worthiness analysis, the important issue is costs, not where the money comes from. A dollar of federal money spent has the same value as a dollar of local money. Total costs must be included. A worthiness analysis considering only local shares is nonsense. Closure Despite the evident draw backs of absolute worthiness analysis, some of which have been discussed here, this approach, partial though it may be, can contribute in important ways to the assessment of the worthiness and cost-effectiveness of alternative transit investments. Avoiding worthi¬ ness analysis has perhaps made it too easy for planners and decision makers alike to ignore this important aspect of investment decision making. Answering, or attempting to answer, the question: "Is this option worth its costs?" becomes increasingly important in the face of tightening finan¬ cial constraints in the public sector. These constraints are becoming par¬ ticularly relevant in public transportation. 63 Thus, supported by appropriate caveats, sensitivity studies and — most importantly — by strong and clear cost-effectiveness analyses, partial but explicit worthiness analyses may help communities find alter¬ native investments which are not only best for them, but also defensible in the effort to secure financial support from other government entities in the face of competition for scarce resources. Chapter Four RECOMMENDED APPROACH TO CEAA EVALUATION Strategic Approach - Reducing the Evaluation Burden Evaluation and choice in CEAA should be a difficult process for analyst and decision maker alike, presuming that a firm choice at the local level has not been made prior to entering this effort. It is the role of the analyst to simplify this process as much as possible, not only by presenting the most useful, and simplest, information set, but also by structuring the search process so that it is efficient. At the very least, this calls for a strategic screening approach that begins with a broad range of generally described alternatives, few in number, and applies an increasingly more complex analysis and evalua¬ tion procedure as the process proceeds. This means that the number of broadly defined alternatives is small at the start, and becomes progres¬ sively smaller with time. The number of detailed, but not radically different, alternatives may be allowed to increase gradually, but ulti¬ mately these must be reduced in number near the completion of the effort. The complexity and detail of the evaluation should increase as the CEAA process goes on; that is, simple, but powerful measures of costs and effectiveness ought to be applied early to generalized alternatives (i.e., not to variations on the same theme). The analysis cost to dis¬ card early alternatives should be low. This cost per alternative ex¬ amined should be allowed to increase through the study as level of de¬ tail, and the number of salient measures, goes up. Figure 7 con¬ ceptualizes the flow of this process. Adopting an approach of this type should make the CEAA process 64 65 more efficient, focusing scarce analysis effort and decision maker atten¬ tion on more detailed issues NUMBER OF ALTERNATIVES * General Alternatives A. ***** * NUMBER OF MEASURES OF COST, EFFEC¬ TIVENESS START CEAA PROCESS PROGRESS Figure 7: Strategic Screening of Alternatives END and problems. Some CEAA studies have generated extraordinarily large numbers of alternatives early in the process, and have evaluated each in considerable detail. This likely to be a waste of resources, in that too much effort is spread over too many alternatives. Some options which could have been rejected easily received a large amount of attention, while others which required special studies to differentiate between them were not treated differently. Another way to understand this issue is illustrated in Figure 8. Here generic alternatives are defined in terms of their major themes, in this case the predominant line haul technology. More detailed options represent variations on the principal themes, in this case differences in alignment, collection/distribution services, etc. It should noted that, in some contexts, the principal themes may be significantly differ¬ ent aligments, while technologies way represent variations. 66 Rather than focusing the most sophisticated evaluation on the large number of options at the bottom of this decision tree, the analysis should be THEME Variations on Theme BASELINE TSM O O BUSWAY tLt / A A O o o o LIGHT RAIL HEAVY RAIL A î k e • o «0 Figure 8: Decision Tree Representation of Screening Process structured to maximize the chances that entire branches can be pruned early and with simple procedures. Of course, some major options, such as a carefully defined baseline alternative, and perhaps one version of the TSM alternative, may be retained to promote a sensible comparison at the end of the procedure. This choice probably should be a matter of policy, no matter what the measures of cost and effectiveness suggest. A pattern of "cuts" likely to represent an efficient CEAA process is shown in this figure. Note that it leads to the opportunity to prune out large numbers of alternatives early, so that remaining resources can be effectively applied to more detailed and complex choices. In this ex¬ ample, rather than performing a detailed evaluation of 17 alternatives, the first level of screening eliminates 9 implied alternatives, the second level of more detailed evaluation screens out two more, and the most com¬ plex evaluation is applied to only six alternatives. If sufficiently simplified demand estimation tools are available, abso¬ lute worthiness analysis should be a part of the earliest screening in the pro¬ cess. For example, pivot point analysis or some other sketch planning 67 technique may facilitate this (TCA, 1981). In the case where only a full-blown, system-wide demand forecasting procedure is used, the worthi¬ ness analysis may be delayed until somewhat later in the process, although the former approach would be preferred. Process Components Figure 9 depicts the key components in the evaluation process. Clearly this is an overview presentation of a process which is not only complex, but which (as suggested above) varies in detail depending on the point in the screening process at which it is applied. Furthermore, it must be recognized that a large amount of analysis is included only by implication, the major part of which is devoted to performance and demand estimation. This overview is presented with the presumption that the proposed approach begins with a clear description of the alternatives, specifications of physical facilities and rolling stock, operating plans and performance estimates, appropriately detailed demand forecasts, and predictions of all other relevant outcomes. Preprocessing activities will include development of the required measures of performance and (non-demand) impacts, estimation of the monetary benefits (especially and largely user travel benefits) and estimation of capital, operating, and maintenance costs. It is then necessary to manipulate the monetary time stream data to put all monetary costs and benefits on the same time basis. This could be present worth or equivalent uniform annualized worth. Procedures for accomplishing this are well established, although it is all too frequent to observe the inappropriate analysis of time streams. The software system ALTERNATIVES AND THEIR EXPECTED ATTRIBUTES PERFORMANCE IMPACT MEASURES BENEFIT ESTIMATION COST/EFFECTIVENESS ANALYSIS ABSOLUTE WORTHINESS ANALYSIS TIME STREAM ANALYSIS DISTRIBUTIONAL ANALYSIS FINANCIAL ANALYSIS c SENSITIVITY ANALYSES J J INTEGRATED PRESENTATION TABULAR PARTIAL C/E NET WORTH TRADE OFFS DISTRIBUTIONAL MEASURES FINANCIAL MEASURES OTHER FINANCIAL STUDIES, ACTIONS Figure 9 : Components of The Proposed Approach ch oo 69 developed in this research supports the efficient analysis of monetary time streams. Some of the analytic procedures used are reviewed in the next chapter. The principal evaluation activities are divided into four tasks, each of which has been described earlier: Cost/effectiveness analysis; absolute worthiness analysis; distributional analysis; and financial analysis. The cost-effectiveness analysis concentrates on reporting selected measures of costs and effectiveness in the form of tables, partial cost/effectiveness measures (e.g., cost per unit of effectiveness), and two-dimensional, graphical trade-off analyses. The major product of absolute worthiness analysis is net (present or annualized) worth. Where this parameter is negative, it can be viewed as the remaining or unoffset costs, after monetary costs have been subtracted from monetary benefits. The distributional analysis reports key effectiveness or partial cost/ effectiveness measures by subgroup or geographical area. The financial analysis estimates current dollar costs and revenues, some of which may be reported in the integrated presentation. Probably the most salient financial measure would be the typical annual operating deficit in current dollars. Ultimately, selected products of these for analysis components should be compiled into an integrated presentation comprising: o Cost/effectiveness tables (i.e., alternatives vs. selected measures of costs and effectiveness) o Selected partial cost/effectiveness measures o Net worth of alternatives 70 o Descriptions of key trade-offs among alternatives (i.e., selected effectiveness measure vs. cost measure such as total cost or capital costs) o Selected distributional measures, usually in graphical form o Selected financial cost measures This integrated presentation should be carefully designed to minimize the information presented to decision makers, while assuring that essential differentiating measures are included in an understandable format. It is important that the limitations of the absolute worthiness analysis be stated clearly and, in particular, that the decision maker not be per¬ mitted to focus on net worth of alternatives alone, without examining other important measures of effectiveness. We propose that this can be achieved by marrying the two kinds of information in the same presentation. Parsimony in the information set is essential if the products are to be used in the choice process. It was easy, in the early days of CEAA, to construct extremely long lists of measures of effectiveness and costs, to the point where hundreds of measures were suggested. Yet decision makers, and even technical analysts, have no hope of processing so many measures. A more constructive approach is to be highly selective in the choice of measures, so that an absolutely minimum set is presented. Redundancy should be eliminated. Measures which fail to differentiate between options should be dropped. Only those measures which can clearly show distinctions between options should be retained, especially at the earliest stages of the screening process. 71 Supplementary evaluation reports may be prepared to include extra, redundant measures if this is felt to be essential. Such reports might include non-differentiating measures which are important for showing that all alternatives meet, or fail to meet, key requirements. The integrated presentation should be designed to answer key questions before they are asked. For example, the presentation should: o highlight consistent and comprehensive measures of costs (e.g., total annualized or present worth costs) o highlight major measures of effectiveness (e.g., ridership) o clearly report one or two essential measures of partial cost-effectiveness (e.g., total annual cost per annual rider) o report net (present or annualized) worth of each alternative o provide a simple, concise verbal description of each alternative (without acronyms, buzzwords, locally unique phrases, etc.) o provide, early and prominently in written reports, in simple tables and/or graphical formats, short but highly informative summaries of the alterna¬ tives and their interrelationships Too many CEAAs (in the form of draft Environmental Impact Statements) are still written as mystery stories which unfold in a slow and complex process, such that it is not clear until the end which alternatives survive 72 the screening and why. It would be more helpful, if this product is to serve as the primary decision document, for the findings, and the essen¬ tial reasoning to come first, with supporting analysis reserved for later in the report. Role of The Computer Software Developed in This Research A major component of this research effort involved the development of an interactive graphics computer system designed to support the kind of evaluation analyses described above. This system should be viewed as illustrative of both the kinds of analyses that can be done and the bene¬ fits of strong, well-structured computer support in accomplishing them. Some agencies may find it convenient to make direct use of this software system, or components of it. Others may wish to pursue parallel develop¬ ments on their own. Still others may find that they can pursue the recom¬ mendations of this research in the absence of any committed computer routines. The intent of the computer work reported here was to make such options available to local agencies. This was accomplished by developing a flex¬ ible analysis capability designed to take the analysts, and his or her clients, through a structured investigation of the attributes of alter¬ natives. The basic framework can be described as a cost-effectiveness analysis; imbedded in this framework is an explicit treatment of absolute worthiness analysis. However, the structure of the system has been built to make it difficult to view one side of the picture (e.g., cost-effec¬ tiveness) without seeing the other (e.g., absolute worthiness). Thus, the software is hierarchical in design, providing a number of 73 branching options to the user, but assuring that a relatively comprehen¬ sive analysis of alternatives is presented. One advantage of this system is that it accomplishes some of the more tedious, but important, analytic tasks, such as treatment of monetary time streams, internally and at high speed. In addition, report formats were designed to provide the possi¬ bility of generating "hard copy" outputs which can be used directly in CEAA reports and briefings. Still, perhaps the most productive use of this tool, in the near term, is likely to be its application by the analyst to facilitate searching through many alternatives to screen out those which are less attractive in a simple and efficient manner. The computer software system does not deal explicitly with distri¬ butional measures, although selected measures of this type can be in¬ cluded at the user's option. It also does not accomplish a financial analysis. The most interesting recent contributions in the latter area is probably the work of Peat, Markwick, Mitchell and Company on the Houston alternatives analysis (Peskin, 1981). The current version of the software system, referred to as GRAFAA (for graphical Alternatives Analysis), is made up of five interlinked sub¬ routines or modules. Data Input is the module which reads externally prepared data and prepares it for use in other modules. It accomplishes much of the re¬ quired time stream manipulation, among other things. Display provides a detailed, verbal description of each alternative, which can be recalled periodically by the user. Time Stream displays the key time stream data in graphical and numerical form. Matrix has the capability to construct 74 and display a table of measures of effectiveness and costs for all alter¬ natives. Measures can be selected by the user, or the predefined measures built into the routine may be used. This module can also show some of the more interesting interrelationships between measures and alternatives in¬ cluded in the effectiveness matrix. Worth is the module which performs and reports the results of the absolute worthiness analysis. It has four components or submodules. Summary provides a simple matrix display of net worth and a small number of additional measures. Some of these additional measures are preselected in the program to help assure that the user clearly understands the absolute worthiness analysis. Trade prepares user-selected, two-dimen¬ sional trade-off graphs for all alternatives. In each case, the user must select a cost measure to trade-off against some other, non-cost measure. This sub-module is particularly helpful for identifying dominated alternatives, so that the efficient frontier of alternatives, the set of undominated alternatives, can be determined. Range plots traveler bene¬ fits versus total costs for each alternative in graphical form. It in¬ cludes a built-in sensitivity analysis of traveler benefits. Finally, Sensitivity performs and reports the results of a sensitivity analysis of net worths of alternatives to assumptions regarding user time values and vehicle operating costs. The software system is highly user-oriented and interactive. Its modules may be selected in any sequence and recycling is permitted. Operation is simple, and a self-teaching capability is included . The next chapter provides some additional detail on the system; a complete users' manual is included as appendix A of this report. Chapter Five THE GRAFAA COMPUTER SOFTWARE SYSTEM This chapter presents a more detailed overview of GRAFAA; a more technically-oriented users' Manual is included in this report as appendix A. The program is written entirely in FOTRAN IV, and is available in forms compatible with both CDC 6600 and IBM 370 Computer systems. It utilizes the PLOT 10 interactive graphics subroutines written by Tektronix, Inc., and it requires an appropriate graphics terminal. The current form is compatible with the Tektronix 4010-1 terminal; it is desirable to have a hard copy printer, e.g., Tektronix or equivalent, to make reproduceable copies of selected displays. The overall structure of the program is shown in Figure 10. Each of the principal modules will be discussed, in turn, below. Data Input The Data Input module reads a data file prepared offline, checks for selected logical errors, and prepares these data for use in the remainder of the program. The system in its present state accomodates 20 alterna¬ tives, although more could be analysed using sequential processing. It can manipulate up to 60 measures of performance, costs, and impact, 14 of which are "built in" (that is, expected to be provided), and the remaining 46 of which may be user-specified. This should be a more than sufficient measures set and, as suggested above, the analysis and choice process would probably benefit from the use of a much smaller collection of measures. 75 Figure 10: Structure of Program Modules. 78 The last five built-in measures are indicators of partial cost-effec¬ tiveness. That is, they are useful measures of the relationship of some (or all) of the costs to some aspects of the effectiveness of alternatives. This module calls for a number of basic data items used in the analysis but not displayed explicitly, including discount rate, analysis time period, vehicle operating costs, value of travel time, rates of change for time stream data, resident corridor population, guideway length, etc. It performs all of the computations necessary to prepare data in the form used by other modules. Perhaps most important and useful of the internal computations per¬ formed are those concerned with time stream manipulation. Data, e.g., ridership, may be entered and manipulated in several ways : o Estimates for selected years, which can provide the basis for linear interpolation/extrapolation to produce a complete time stream; o First year estimate of measures, which can be extra¬ polated based on a constant, compounded percentage growth rate ; o Estimates for selected years, which can be extra¬ polated at a constant level; These extrapolation options will estimate the entire time stream where this is appropriate. This module will forward lump sum capital cost estimates to the Time Stream module for direct display. In addition, it will find the discounted present worth of these and other costs using the equation: 79 PW = I n C. V 1 n j=l (l+i)j Where PW = Present Worth i = Discount rate per year j = Number of years n = Horizon or terminal year for analysis period The equivalent, uniform annualized worth of such cost items will also be computed, using : This module accounts for differences between the life of capital com¬ ponents and the length of the time horizon for the analysis. For example, if the component lasts longer than the time horizon, the present worth computation is as follows : let M = component life > n EUAW = PW i(l+i)n u+i)n-: M EUAW (over component life)= C i(l+i) M (l+i) -1 PW (over time horizon) = EUAW (component) (l+i)n-l • / .. n i (i+i) EUAW (time horizon) = PW(time horizon) i(l+i)n (l+i)n-l 80 This assures that only the component costs amortized during the analysis period are included in the absolute worthiness analysis. In the case where M (service life) < n (time horizon), the software does the following computations : EUAW (component) = C i(l+i)M M (1+i) -1 This cost is treated as the annual cost of the component for each of the n years. Thus, the present worth of the component's cost over the time horizon becomes : PW (time horizon) = EUAW (component) (1+i)"-l i(1+i)n This assumes that the component is replaced (at a constant, uninflated cost) when its service life expires. Display The Display module presents an extended description of each alterna¬ tive — up to 25 lines in length. This, of course, comes directly from the input data file. This description is included in order to inform the user of the detailed nominal attributes of each alternative. In subse¬ quent modules, extremely brief tags are used to describe alternatives. It was expected that some users may have a periodic need to be reminded of the detailed attributes of options. Thus, it is possible for users to return periodically to the Display module for this reminder. This capability also facilitates explicit reporting of qualitative characteristics of alternatives which are not included elsewhere in GRAFAA. An example of the format of this description is shown in Figure 11. NAME - BUSLANES ALTERNATIUE 2—EXCLUSIUE BUS LANES MODE AND CORRIDOR DESCRIPTION—BUS SERUICE OPERATED ON EXCLUSIUE BUS LANES CONSTRUCTED IN THE KENNEDY EXPRESSWAY MEDIAN SERUICE—DAYTIME HEADWAYS OF 15 MINUTES, EARLY MORNING AND EUENING HEADWAYS OF 30 MINUTES, AND NIGHTTIME HEADWAYS OF 60 MINUTES. IMPLEMENTATION—2 TO 3 YEARS MAJOR IMPACTS—LOWER TRAUELTIMES Figure 11: Example of Display Format 82 Time Stream This module displays graphically time streams of the following variables (selected by the user): o Capital costs o Annualized capital costs o Operating and maintenance costs o Total annual costs o Ridership and revenues o Operating deficit (operating and maintenance cost less revenues) o Total deficit The graphics terminal screen can display up to two time streams simultaneously. The user may select two measures for one alternative, or the same measure for any two alternatives. In addition to displaying the year-by-year time stream, the system reports the present value and the equivalent uniform annualized values for each time stream. This is illustrated in Figure 12. For alternatives other than the baseline option, the user may elect to delay implementation up to 5 years beyond the initial start date. In this case, values of the selected measures for the baseline alternative will be substituted in each of the "delayed" years. Time streams will be displayed as continuous line charts if inter¬ polation is permitted by the user (e.g., ridership data), or as bar charts if interpolation is not permitted (e.g., capital costs). Tests have demonstrated that this particular module offers an efficient way to accomplish otherwise time-assuming time stream manipulations. Per- M I L L I 0 N $ 200-n 0- 1975 OPER. 3< MAINT. COST FOR ALT. 4 AT DISCOUNT RATE . M I L L I 0 H 200' 0 — 1975 TOTAL COST FOR ALT. 4 ■ r-ffi f>J- T 1980 0 r 1985 YEAR 1 I I 1 T 1990 1995 AT DISCOUNT RATE .10 PRES UAL=» 539.3 EQUIU ANN UAL= 77.5 PAUSING. HIT 'RETURN' TO CONTINUE,OR ENTER TO TERMINATE TIME STREAM ANALYSIS. ? 00 LU Figure 12 : Example of Time Stream Format ~1 Ï 1935 1990 YEAR 421.5 EQUIU ANN UAL= I 1995 55.4 1980 10 PRES UAL= 84 haps more importantly, with a hard copy machine available, this module produces graphics directly useable in reports in a matter of seconds. Matrix This module displays a table or matrix in which rows are alternatives and columns are measures of costs and effectiveness. The user may see the matrix for some or all of the alternatives; if there are more alternatives than the display screen can accomodate, additional alternatives are shown on successive "pages". A similar choice is available for selecting measures to be displayed; the user may see all measures or some (selected) measures. The user may also choose two prespecified sets of selected measures: five key or primary measures and the five partial cost-effectiveness measures listed earlier. Only five measures can be shown at one time; additional measures appear on successive pages of the display. The pace of paging is controlled by the user, so that sufficient time is available for analysis. Three different analysis options are included in this module. The first simply presents raw measures for each alternative. The second reports the ratios of each measure for each alternative to the corresponding measure for a user-selected (e.g., baseline) alternative. This helps to show the relationships among the alternatives. In the third option, the percentage difference between each alternative and a user-selected option is computed for each measure. Then, only measures which are different from the base alternative by a fixed (user-chosen) percentage or more are reported. This allows the user to ignore small (as defined by the user) differences between alternatives in order to focus on measures which differentiate more clearly 85 between options. Figure 13 shows an example of the products of this module. Worth This module provides several different views of the results of the absolute worthiness analysis. It is made up of four separate submodules, from which the user may select those which are desired. Summary presents a matrix, similar to those available in Matrix, showing the alternatives as rows and from three to seven measures of cost and effectiveness. Three measures, net worth, monetary traveler benefits, and total costs (all annualized) are automatically presented. The user may select up to four additional measures. The choice of required measures assures that the user sees the absolute worthiness measure and its compon¬ ents, along with other, chosen measures. The resulting display is more dense (compact) than that produced in Matrix. As an option, the user may call for a profile plot of the measures chosen for this module across all alternatives. The profile reports each measure value as a percentage of the maximum value of that measure over all alternatives. A sample of this format is shown in Figure 14. Tradeoff displays two-dimensional trade-offs between a cost measure and an effectiveness measure. The cost measure may be either annual equivalent capital costs or annual equivalent total costs. All alterna¬ tives will be shown in this presentation, an example of which is shown in Figure 15. Range illustrates graphically the relationship between total annual costs and monetary user benefits. As shown in Figure 16, costs are on the horizontal axis and benefits on the vertical axis. The line "costs = benefit^' is also shown (in this case it does not have a 45 degree slope 10DE OF DISPLAY= 2 MEASURES ALTERNATIVES 6 7 8 9 10 ANN EQUV ANN EQUV ANN EQUV AUG ANN PRES UAL TOTAL TOTAL RELATIVE RIDRSHIP CAP COST COST DEFICIT NET $000 PER WORTH RT MILE 1 DO NOTHING .55 1.04 0.00 .18 0.00 2 BUSLANES 1 .00 1.00 1.00 1.00 1.00 3 C&NW EXT 12.33 26.51 32.35 1.94 9.03 4 CTA EXT 10.82 22.77 25.11 2.03 7.27 A UALUE OF 999999 INDICATES A UALUE OF 0 IN THE BASE ALTERNATIVE, SO THAT RATIO OR PERCENT DEVIATION CANNOT BE COMPUTED. PAUSING,HIT 'RETURN' TO CONTINUE, OR ENTER TO TERMINATE THIS DISPLAY. ? Figure 13: Example of Matrix Format oo BUSLANES C 2) C&NW EXT < 3> CTA EXT C 4 > PAUSING. HIT 'RETURN' TO CONTINUE. ? Figure 14: Example of Profile Plot Format A U E R A G E A N N U A L R I D E R S H I P 20—i 15 — 10 — 5 — 0- 0 T 20 1 Ir 40 ~r 60 1—r 80 ~l 100 ANNUAL EQUIU TOTAL COST PLEASE HIT RETURN TO CONTINUE. Figure 15 : Example of Trade Off Format COSTS AND POTENTIAL RANGE OF USER BENEFITS FOR EACH ALTERNATIUE T R A U E L E R B E N E F I T $ 0 0 0 25-n 20 — 15 — 10 — 5 — 0- O o 0 "1 T I ' 1 1 1 1 1 20 40 60 80 100 TOTAL COST < $000 > PLEASE HIT RETURN TO TERMINATE THIS DISPLAY ? CD TO TERMINATE THIS DISPLAY. ? Figure 17: Example of Sensitivity Format 10 H 92 Application of GRAFAA As discussed above, the software has many user options in its structure, perhaps the most important of which is the ability to see any or all of the presentations described briefly above. In addition, there is an ample choice of measures, with only a few key measures prespecified. The user may review all or only some of the alternatives. The worthiness indicators are closely integrated with the cost-effec¬ tiveness measures, making it easy to view them as a unified set, and difficult to see only one aspect of the alternatives. The Display modules provides a reasonable amount of space for reporting qualitative aspects of alternatives. It has proven feasible, once the input file has been prepared, to scan all of the displays for 6-10 alternatives and 20 measures in about 30 minutes. With a hard copy device attached, excellent graphics for report writing can be prepared with no additional time commitment. The user may, at a nominal additional cost in computer time, linger much longer over the displays to study and analyze them, cycle back to earlier presentations, and explore options in considerable detail. An interesting application of this system is to make the terminal available for scanning the alternatives to a wide variety of users, who may scan through the presentations at their own pace and on their own schedules. The system has been written for use by laymen, and, through its self-teaching capability and resistance to errors in commands, has been successfully used by high school students, engineering graduate students, and practicing planners, in each case with no apparent operational difficulties. 93 Prologue to Chapter Six CEAA is a difficult process, and no analysis tool will make it simple. In the course of this research, a number of additonal strategic issues were exposed, a discussion and resolution of which may serve to make future CEAAs better, if not easier. These issues are reviewed in the next chapter. Chapter Six DESIGNING USER-ORIENTED ALTERNATIVES ANALYSES Introduction This chapter identifies and discusses some of the shortcomings of alternatives analyses and, where possible, recommends potential solu¬ tions to these problems. A single objective - that of making alter¬ natives analysis more useful to the users of the information and products produced - provides the basis for identifying the shortcomings of traditional cost-effectiveness alternatives analysis (CEAA) and guides the development of potential solutions. A number of the characteristics of CEAA as a component of the trans¬ portation planning process bring about the need for its usefulness and contribution to public sector decision making. While these are by no means unique to cost-effectiveness alternatives analysis, it is particularly relevant to consider them in the development of CEAA procedures. First, as a planning procedure, CEAA is explicitly focused on supporting decision making, or at least on documenting a rational choice; whereas most recent actions to advance planning, through regulations and research, have been more concerned with technical advances or the implementation of certain policies (e.g., TSM). Second, the large effort applied toward advancing CEAA tools and methods, and the large expenditure or resources in preparing CEAAs and implementing major transit projects, increases the importance of assuring that the results are useful to decision makers. Third, the possible extension of the CEAA philosophy to other transpor¬ tation investments offers a special opportunity to increase the effective¬ ness of transportation planning as a whole. Fourth, CEAA often involves controversial, high-impact alternatives and accordingly has a large and diverse set of users. 94 95 Conflicting Objectives For Alternatives Analysis Alternatives analyses reports are responses to a number of different objectives, some of which may best be achieved by different and conflicting planning process activities and differing planning product emphases. Among these objectives are: 1. Provide evidence that at least one alternative is sufficiently attractive to merit discretionary funding; 2. Comply with federal rules concerning citizen participation, analysis methods, environmental impact assessment, and range of alternatives, etc.; 3. Provide information to local and state decision makers to enable them to decide which if any action alternatives, are preferred; 4. Provide justification for a preferred choice; 5. Preserve or enhance the reputation of the agency by providing professionally reputable information and services to the plan users and public. Several factors have contributed to the inability of CEAA to fulfill all of these objectives satisfactorily. The factors include: the demands of complex..regulatijans which may have lost sight of the decision makers' needs and the planners' role, the tendency of many planners to place dis¬ proportionate emphasis on the collection and manipulation of vast amounts of data rather than on clarifying the choice, the complexity brought on by the increased concern for and ability to deal with social, economic and environmental issues and, finally, the inherent difficulty of integrating these often contradictory objectives into a single transparent, decision- oriented CEAA process and product. Generally there is a need to make alternatives analysis more supportive of the needs of decision makers to assure that the process itself will have both meaning and appropriate impact. There are a number of possibilities for reallocating planning resources in order to integrate the competing 96 demands on the CEAA process and to satisfy the needs of different users of the CEAA products. One cannot, of course, make sure that all of the infor¬ mation provided in CEAA is used or interpreted in a manner that appeals to all planners, but efforts can be made to ensure this possibility. The remainder of this chapter is focused on three general issues: explicit problem definitions, opportunities for improving CEAA measures and methods, and, improving CEAA products. Collectively these provide an oppor¬ tunity to improve the responsiveness of planning and ultimately to increase the utilization of the results of analysis and planning in the decision making process. The Role of Explicit Problem Definitions One target for improving the usefulness of alternatives analysis in decision making is clarifying the definitions of the problems to which the alternatives are intended to respond. (Without clear understanding of the problem(s) to be solved or the goals to be met, it is difficult to identify and evaluate the relevant tradeoffs between alternatives. The issues surrounding the proposals to construct a Franklin Street subway in Chicago exemplify the range of interpretations of a problem studied in alternatives analysis. The initial problem, stated decades ago, was the alleviated by replacing it with a subway. The single-purpose agency created in 1968 to solve this problem has, over the past decade, attempted to por¬ tray the problem as. the need to improve central area transportation. This emphasis on quality of transportation is more compatable with the objectives of those agencies providing the majority of financial resources for the proposal actions, but it diverts attention from the initial problem to a new one, while the alternatives themselves were not changed. the Loop elevated structure in the CBD, which might be 97 More recently, the decision not to build a particular section of an Interstate highway created the opportunity to transfer resources to solve other transportation problems of the Chicago region. This increased bud¬ getary flexibility at the regional scale expanded the range of actors and problem definitions: replacement of the Loop elevated was a high priority for one agency in one part of the region. Other interests, with at least potential access to the diverted resources, held other priority problems. Thus,Vome perceived the problem as one of finding the best (regional) transportation investment package. Other actors were primarily concerned with maximizing the economic stimulus associated with spending large amounts of federal money in the area: They perceived the problem as that of finding projects that could be implemented quickly and would boost the local economy. Still others per¬ ceived the problem as one of getting as much money as possible for their respective jurisdictions. Some actors defined the problem as that of finding the best public transit investment in the city proper : maximize trans¬ portation and associated energy, accessibility and environmental benefits. The series of planning documents further narrowed the scope by apparently adopting as a primary problem definition how best to replace the existing downtown elevated rail structure with a subway, a definition that constrains the range of alternatives to those compatible with the narrowly defined agency role. While some of these varied concerns were apparent in the planning pro¬ cess and resultant analysis products, there has never been a clear, open, reasoned debate on alternative goals and problem definitions. As a result, considerable uncertainty remained regarding what was being evaluated on what terms. Thus, in the written reports, decision makers were presented 98 with less than a clear picture of the range of the alternatives, the.full set of choice options and the relationship of those options to the problem to be solved. Such information seems essential for making a strong, clear, defensible choice of a course of action. The Chicago example is not unique. The high cost of most actions evaluated in alternatives analyses implies that a number of long-term resource allocation issues must be resolved implicitly or explicitly prior to or during the CEAA process. A serious source of controversy in CEAA may be the lack of coordination with the long-range programming process. Programming itself becomes difficult because of the uncertainty associated with securing a specific amount of discretionary transit capital grant funds. The approval of a given project may well make federal approval of other discretionary fudn projects less likely in the future and, at the local level, may tie up local matching fund resources, thus implicitly prioritizing future projects. Thus, often the primary problem of selecting an alternative is combined with the problem of choosing which problem to study, or which project to advance first. The absence of an explicit programming activity will become increasingly important if any of the following trends or changes occur: 1) If discretionary funding allocation at the federal level is terminated in favor of a formula allocation of funds; in such an event the whole choice process, though con¬ strained in terms of available resources, would be focused at the local or state level. This may constructively force more serious consideration of priority programming. The same outcome might occur if political pressures for proportional allocation of resources diminish the actual discretionary power in federal funding. 2) If the continued de-emphasis of long-range planning forces issues dealing with system planning priorities and goals to 99 be addressed only during the CEAA process. 3) If the proposed requirements for the CEAAs for highway projects is reactiviated, providing a focal point for challenges to highway program priorities; and 4) If intermodal planning and funding are integrated to the point where CEAA and programming activities involve serious intermodal resource allocations. In general, as the decision making discretion at a given governmental level increases, an objective of many who argue for local control, the poten¬ tial range of problem definitions increases. The potential controversy and/or the complexity of the analysis and decision making activities necessary at the decision level increases accordingly. One approach for dealing with such complexity is to focus on clarifying problem definitions and establishing action priorities and decision sequences. Improving Problem Definitions Ill-defined problems have always plagued planners (Faludi, 1973) and there is considerable literature suggesting that a clear problem definition is a key to success in problem solving (E.S. Quade, 1966; C. Alexander, 1965). The challenge of defining problems has actually increased as we have learned more about transportation and its potential impacts. For example, transpor¬ tation changes are commonly implemented not only to meet mobility needs, but to improve air quality, guide land-use development, increase employment, or redistribute resources among social and economic groups. In many cases, transportation performance goals are of only secondary interest to decision makers; yet, without clear statements of true or ultimate problems and goals, the content of alternatives analysis reports may not correspond well with decision making needs. The implied problem definitions in alternatives analysis may not be acceptable to all of the potential plan users. This suggests the merit in articulating several views of the problem, if such 100 multiple perspectives truly exist in the community. The alternatives analysis itself might then be structured to assess alternatives in terms of the various explicit problem concepts, instead of working with a single, implicit problem definition. Of course there is a clear risk in this approach in that federal decision makers may find it easier to recognize cases where transit moneys are intended for solving essentially non-transportation problems. The federal government may ultimately find it logical to adopt one of two policies. First, it might admid that transit resources are commonly used to solve other social and economic problems, and that these projects are appropriate. The second policy option calls for the Department of Transportation to clairfy its commitment to transport goals only, which will eventually shift the focus and nature of local transit projects or force them to be financed by a variety of sources reflecting the variety of goals they strive to accomplish. This more explicit, problem-oriented approach to alternatives analysis might serve to point out that, in many cases, the choice between alternative solutions is in fact a choice between solving alternative problems. An open debate on the problems themselves rarely occurs because of the limited infor¬ mation exchange and public involvement early in the alternatives analysis process, when problem definitions are formulated. The serious interaction occurs once the alternatives are on the table. Clarifying the link between problems and alternatives should not only make the planner's job clearer, but should also highlight what the real decision is all about. Defining and selecting problems to be solved and the goals to be met is a political process as well as a technical one. These efforts must inte¬ grate the priorities and values of the elected officials and public with the abilities of the planners to inventory, analyze and solve the problems. 101 This requires a willingness on the part of planners to acknowledge the realities of political decision making, including such things as partisan¬ ship and compromise as well as the possibility that the transportation decisions may not be rational from the perspective of the planners because they were parts of packages of decisions that involve such things as education, health care, or taxation. This understanding can lead to improved relation¬ ships between decision makers and planners that are necessary in order to ensure timely access to the true motivations and priorities of the decision makers as they relate to the identification and definition of problems. Both planning theorists and practitioners argue against devoting significant amounts of resources to establishing and prioritizing goals (Cohen, 1978). The reasons cited are the difficulty in obtaining this information and the inability to translate these goals into acceptable alter¬ natives; yet, these differences in values and goals shape the differing perceptions of the problem. While it is seldom possible to establish a consensus of values (or even to assess the implications of a given set of values on the design and evaluation of specific alternatives)Y it is highly desirable to make explicit the differing or competing values and goals and the resultant problem definitions. This provides a sound basis for shaping the planning process, defining the scope of the alternatives to be considered, and conducting the evaluation itself (Wachs, Hudson, and Schofer, 1974). While a general method for establishing consensus on the problem definition and underlying goals will not be proposed here, this situation could be improved by providing, as part of the alternative analysis process and products, both explicit clarification of the problem definitions assumed and the constraints on the set of solutions considered. Choosing problem definitions, deriving design and selection criteria and then evaluating the 102 alternatives are tasks that should involve as much debate and analysis as the final selection of an alternative. It may be helpful to the analyst to identify which decision makers have, or are likely to assume, the responsi¬ bility for these critical choices. It should be noted that focusing on the specification of problem definitions and problem priorities may well be easier, and is likely to be as productive in terms of structuring the analysis and choice processes, as goal formulation. For example, despite the strong arguments for founding transportation planning and CEAA on well-specified goals, even when this is done, the stated goals are often too general to be of much use. It seems likely that clear problem statements will be easier to prepare, will high¬ light priority conflicts appropriately, and will provide a stronger basis for assessment of opportunity costs, for evaluation and for decision. Opportunities For Improving CEAA Methods and Measures The organizational and analytical methods used for alternative, analyses and the set of available and recommended measures are always undergoing refinement and adaption. This discussion will only center on a few of these issues that are believed to be particularly important from the perspective of the plan users. The types of alternatives : CEAA as it is presently structured fails to deal adequately with certain types of alternatives. The real costs and benefits of rehabilitation or renovation alternatives are difficult to quantify and express consistently with information on other alternatives (service additions). There is only poor data available to estimate such factors as remaining life expectancy, relative safety, comfort, maintenance costs and other impediments such as reduced reliability, reduced accessi¬ bility or other factors that may affect ridership during rehabilitation. 103 Additional experience and an acknowledgement of uncertainties appears to be the best means of dealing with this difficulty. System Performance: In instances where the CEAA deals with additions to an existing system, analysis often fails to deal with the system-wide impacts of the proposed alternatives. The requirement that a given invest¬ ment be evaluated based on its own merits neglects the reality that is, or may in the future be, part of a larger system and may affect the existing or potential performance of the whole system. Addressing this requires that these actual or potential impacts be dealt with by the incorporation of some system performance measures in the analysis. It may also bring about the need to pursue comprehensive tests of the full system implications of a wide range of long-term programming options. Flexibility of alternatives: A decision maker may very likely be con¬ cerned about the extent to which a given choice constrains the ability to select other alternatives or adapt to different conditions or values at a later date. Existing measurements often fail to incorporate this important decision criterion into the analysis. Two approaches to dealing with this issue merit consideration. The first involves a more accurate portrayal of the characteristics of alternatives; the second involves the use of a variety of techniques for describing sensitivity, uncertainty, and flexibility. Improvements in the descriptions of alternatives should be designed to enable the users to understand the nature of the assumptions about those alternatives which are likely to affect the evaluation. One important example deals with assumed operating characteristics. In many cases the performance, impacts and costs of a given network configuration can vary widely, depending on which of many alternative operating plans are considered. Manipulation of the cost and availability of parking, fares, headways. 104 security precautions, feeder service, marketing and other parameters may significantly affect the resultant use, costs and benefits of a facility; yet these options are not fully constrained by the choice of a given capital investment (e.g., fixed guideway system). Investing in a given network configuration opens up, at least conceptually, an entire range of outcomes: the community implementing a chosen configuration might be faced with per¬ formance, impacts and costs anywhere within a broad range in some future year, yet in CEAA, this range of measures is seldom reflected. This problem suggests the need for a change in alternatives analysis. The operating characteristics used to test an alternative should be clearly specified, both in the description of each alternative and through the measures themselves. The relationship between the chosen operating charac¬ teristics and the feasible range of those characteristics should be indicated; this might be accomplished by reporting the potential capabilities and capacities of the system components as well as their anticipated use levels. A better understanding of the flexibility or range of potential per¬ formance of a given alternative can also be communicated by using future scenarios or other schemes aimed at disclosing the sensitivity of performance measures to changes in the operating (and other) parameters. If parameters that are important to the decision makers are subject to significant uncer¬ tainty, efforts should be directed toward documenting their sensitivities and uncertainties and, subject to cost and state-of-the-art constraints, reducing those undertainties. Information documenting this flexibility typically is provided, if at all, in the appendices of reports; little effort is made to portray the flexibility of alternatives directly. Many transit facilities presently being contemplated or planned are designed to be complemented by future service extensions, or are intended to 105 replace existing facilities that may otherwise need major rehabilitation or face discontinuation. Evaluating these contingent alternatives, with impacts largely dependent upon assumptions about future decisions, virtually requires that the flexibility of alternatives be dealt with explicitly. Presenting information on contingencies is not likely to be easy. Per¬ haps the principle issues to keep in mind are 1) the need to make important hidden assumptions explicit and 2) the need to portray the feasible, future development paths which lie beyond the alternative, intermediate outcome states. For example, when considering facility rehabilitation options, information of the type shown in Figure 18 should be available. in Year X Do Rehabilitate Outcome Measures Figure 18: Representing the Assumptions Behind Contingent Alternatives The comparison of options should focus on the three (or more) sets of out¬ come measures so that the relative merits of rehabilitation become apparent. When assessing a time-sequence of inter-linked alternatives, a similar, simple decision tree structure may be helpful in assuring that both planners and decision makers understand the options; this is shown in Figure 19. Again, the relevant comparison should focus on a small number of key out¬ come measures across all option paths. 106 Option A Subsequent Option A1 No Subsequent^ Option Outcome Measures -»• Outcome Measures Option B Subsequent Option B1 (sames as A^) Subsequent Option B2 No Subsequent^ Option Outcome Measures Outcome Measures -** Outcome Measures Figure 19: Representing Staged Options Such a format may help to isolate the most important interlinking attributes of alternatives. For example, it could show the implications not only of building project A followed by A , but also of building A and finding this to be the only major transit investment which is feasible in the foreseeable future. The serious risk lies in the tendency to wish to implement A because of its performance with A^, ignoring the possibility that A^ may never be built. It is also important to recognize the flexibility inherent in the timing of the choice of an alternative. CEAA assumes implicitly that now is the time to make a decision - either build one of the alternatives or implement the no-action decision - which is typically evaluated assuming little change over the time span of analysis. In reality, there are commonly important time staging decisions to be made. Deferring an action by adopting 107 the no-action alternative now allows for the possibility that a future CEAA may result in the discovery of an attractive alternative or the values, conditions or technology may have changed such that one of the marginal alternatives is now worthy of investment. This time flexibility is not often portrayed to the decision maker. Deferring various actions represents an additional alternative that is sometimes worthy of consideration. Expecting the decision maker to choose between the normally low (or zero) benefit no-action alternative and a list of new facilities or services based on the limited information provided may only encourage them to look beyond the planning documents for information on which to base decisions, or it may diffuse and delay the choice itself. Process and product credibility: The design of alternatives analysis reports should enable the plan user to assess the uncertainty, and hence the credibility of the information collected and processed in the planning process. Uncertainty, in the broadest sense, means the level of doubt associated with predicted (expected) consequences of implementing parti¬ cular alternatives. Credibility refers to the degree to which a user of alternatives analysis products believes in the validity of those products. The latter clearly depends on the former, but in a rather complex way. For example, eliminating uncertainty in forecasts will not assure product credibility, which also depends on other attributes of the product, as well as the perceptions of the analyst on the part of plan users. Similarly, an analysis product with considerable uncertainty may be quite credible if the nature of the uncertainty is spelled out. The need for a better treatment of uncertainty and credibility can be identified in many planning efforts. For example, rail-transit capital and operating costs, and anticipated patronage - two of the more important 108 evaluation measures - have not been accurately predicted in some recent experiences (Wachs and Ortner, 1979); this suggests room for improvement, through either sensitivity analysis or enhanced prediction methods. Twenty years ago, inflation and procedural delays were not common sources of uncertainty in large public-sector projects; today, the uncertainty associated with predicting implementation dates and accurately estimating capital costs is virtually always a problem that must be addressed. Pre¬ diction techniques may never be able to deal with these issues satisfac¬ torily;! yet, it is possible to accomodate such sources of uncertainty explicitly, using scenario-based methods. Without improvement in the ability of the planners to estimate accurately, or at least to expose and discuss the uncertainties involved and their implications, the credibility of both planners and their products is likely to diminish. Practical and effective ways for communicating uncertainty to decision makers are not obvious. Not the least of the reasons for this is the no¬ tion that many transportation planners see clear advantages in burying uncertainty to avoid confusing already complex choices - treatment of uncertainty cannot fail to increase the dimensionality of evaluation results - and to protect their image of confidence and ojbectivity. One might also expect that decision makers themselves feel more comfortable ignoring uncertainty, the knowledge of which may appear to deflate their power and weaken the justification for actions, present and past (Slovic, et al., 1974). For the decision maker, there are two interrelated risks of ignoring uncertainty. First, there is the risk of making what turns out to be a poor choise; second, there is the risk of being forced to account for that choice. The planner, too, shares these risks, and also faces the possible 109 immediate loss of credibility by admitting the existence of uncertainty. This short-term loss may be small, however, especially when facing experienced decision makers and constituents who have seen evidence of earlier predictive failures. For both the long term credibility of analysis, as well as the good of society and its decision makers, it appears important to recognize uncertainty at least in the key parameters of choice, such as cost, level of patronage, and developmental impacts. Among the tactics which might be considered for illustrating this kind of uncertainty are the following, listed (approximately) from simplest to most complex: (1) Produce percentile ranges of key estimates in the form of a base value, and values (for example) 20% higher and 20% lower; (2) Produce base (point) estimates and judgmentally developed "worst expected outcome" values. The latter could come from analytic studies with varied assumptions or by applying percentage multipliers to base outcome values; (3) Develop a small number (e.g., 3) of planning scenarios and produce separate estimates of key parameters for each scenario; (4) Use analytic techniques with varied assumptions to produce high, medium, and low estimates of key parameters; (5) Apply approach (4) above, but develop subjective probability estimates to describe each outcome. It will be important to limit the treatment of uncertainty to a small number of key parameters to ensure feasibility of the choice process. The latter might be directed toward examining base-worst-best case comparisons on the few, salient dimensions. For some measures, credibility and uncer¬ tainty can be dealt with by disclosing the assumptions underlying methods and estimation procedures. Diagramatic presentations which show not only 110 point estimates of parameter values but also the range of such measure values as defined by changes in assumptions and sensitivity analyses may help to illustrate the implications of both alternative flexibility and uncertainty. This approach may be especially useful for showing situations where, given the level of uncertainty, two seemingly different alternatives are in fact not susceptable to differentiation. It may also suggest ways to combine the best attributes of several alternatives to produce a pre¬ ferred action. A review of recent CEAA reports shows that it is not uncommon to find that on many (sometimes most) evaluative measures several (sometimes all) alternatives are essentially equivalent; that is, differences in measures between alternatives are less than the likely errors in fore¬ casting. It may well be deceptive to pour through such predictions in search of the best alternative, when in reality (within the limits of forecasting capabilities) many (or all) alternatives are equivalent. It may be desirable in such cases to report that a selection among such alternatives is indeterminant. This stylistic choice offers several interesting possibilities. First, it may increase the credibility of the analysis by bringing uncertainty to the surface. Second, this approach may be used to give local decision makers more freedom of choice by pointing out the apparent technical indifference between alternatives. Finally, it may be reasonable for technical analysts to take a subjective position on a preferred alternative if a logical, but non-analytic, basis for such an action can be found and reported. One of the display formats offered in the graphics routine developed in this research, for example, highlights those alternatives which are, Ill and are not, significantly different from a selected alternative in terms of measures of costs and effectiveness. Here "significantly different" is defined in terms of a minimum fixed percentage difference in the measures, but more complex definitions are possible. It may be possible to increase product credibility by documenting the historical accuracy of the data collection and analysis techniques used; further support may come from "reasonableness" checks, comparing selected results with those produced in other cases and other cities. The planners should strive to document the accuracy of the information they provide in at least three general areas: 1) credibility of the theories used to explain the behavior of the phenomenon under consideration (e.g., travel behavior, transportation/land-use interaction, etc.); 2) validity of the transformation of those theories into models or rules used for prediction and evaluation; and 3) statistical accuracy of the data collec¬ tion and processing techniques used in analysis and evaluation. In some cases it may also be important to provide credibility enhancement for local policy assumptions which enter prominently into the analysis. For example, to the extent that the viability of an alternative is dependent on the evolution of a preferred land-use pattern, actions required to ensure that pattern should be revealed, their likelihood assessed, and the potential influence of mitigating factors explored. A related example is the serious problem of assuming that local govern¬ ment will be able to support financially the continued operation of proposal alternatives and existing services. A number of major transit properties in the U.S. exist on the edge of financial disaster. Experience suggests that major new capital investments are likely to degrade the cash flow positions of transit operators. It seems reasonable to demand that such 112 local agencies either document their future ability to meet operating and maintenance costs or indicate the nature of these fiscal uncertainties. This may not only increase the likelihood that affordable systems will evolve, but it may also provide local governments with the incentive to get their financial houses in order prior to seeking funds for new capital investments. Plan users' confidence is also influenced by how well they understand the information, the reputation of the planner(s), perceptions of the plan¬ ners' motivations, and the appropriateness of the planning process. The level of understanding can sometimes be increased by eliminating information not directly salient to the decision (or relegating such items to appendices) and focusing on the more important issues (see next section). It is also important to assure that reports are readable and unambiguous. In addition to the care needed in writing documents, there are likely to be benefits derived from having non-technical reviewers preview and critique draft documents. One approach for dealing with the overall plan credibility - and for focusing and increasing the understanding of the plan - would be to provide independent assessments of the procedures and findings. This could be designed as both a review of the planners' work and a qualitative analysis of the highly technical information that cannot easily be assessed by most plan users. Many plan users already rely on the assessments of others in whom they have confidence as a basis for determining how they will respond to a plan; yet, not all actors have access to the resources required to hire experts and few such experts are truly disinterested parties. Providing an "unbiased" (or at least independent) qualitative assessment of the planning process and findings provides a realistic means of compressing the large 113 volumes of information and assessing the credibility of this information. The individuals charged with this responsibility would have to be able to deal with technical information, assessing it in terms that could be communi¬ cated to plan users. The reviewers themselves would need to have a high level of credibility and independence to make such an approach effective. The products of this review process might be in the form of a brief written report or presentation summarizing major findings of the planning effort and identifying critical areas of strengths and weaknesses. The visitation teams organized by the American Institute of Architects provided on a cost-of-travel basis suggest a way for professional socie¬ ties to contribute to this review process. The value of bringing a multi- disciplinary team of people from outside the region of concern should not be overlooked; it is common to find that a large fraction of the key profes¬ sionals within a region undertaking alternatives analysis have a vested interest either in the process itself or its outcome. To get a better understanding of plan users' views on product credibility, it would be useful for local agencies with federal encouragement, to seek feedback from various decision-makers during and after alternatives analysis processes. That is, specific efforts should be made to assess decision makers' evaluation of alternatives analysis information support, particularly in terms of the usefulness of, and their confidence in, the results of plan¬ ning and analysis efforts. This might be accomplished through informal interviews or workshop sessions, and these, too, might best be conducted by disinterested parties. Improving Alternatives Analysis Products Alternatives analysis .products could be improved by a stronger focus on those parameters most relevant to decision making. This focusing would 114 be aided considerably by a refined problem definition and a knowledge of the background issues and values. Focusing would not necessarily require that certain impacts be neglected, but only that the significance of the impact be reflected in the presentation. Significance is a function of the importance of an issue or impact to the community and its decision makers; it also reflects the degree to which an impact measure can actually help to differentiate between alternatives. The latter is affected not only by the sensitivity of measures to attributes of alternatives, but also by the ability of forecasting techniques to provide accurate future estimates of the measures. Too often, detailed analyses of easily identified but relatively unimportant parameters are carried out while more significant issues remain superficially treated. For instance, voluminous, readily available socio¬ economic statistics are often reported in transportation plans. Although some of this information is relevant, methods for compressing it while still reflecting the geographic, racial, and economic distributions of impacts may increase clarity and jreduce decision maker work load. Insig¬ nificant information should be omitted. Recently-issues guidelines for preparing environmental impact state¬ ments (CEQ, 1977) clearly demand sharply focused documents. A number of the recommendations contained in these guidelines are relevant to alter¬ natives analysis: a. Erepare analytic rather than encyclopedic environmental impact statements ; b. Discuss non-significant issues only briefly; c. Write environmental impact statements in plain language ; d. Emphasize the portions of the environmental impact statement that are useful to decision makers and the public; reduce empha¬ sis on background material. 115 Still the temptation to provide large volumes of unstructured information seems to remain strong. Another way to provide focus in alternatives analysis is to exclude those parameters that do not discriminate between alternatives. For example, some aggregate statistics (e.g., vehicle-miles of travel, total person trips) may vary little across radically different alternatives. Whether this lack of variation is due to limitations of forecasting procedures, the use of an overly large study area that reduces the apparent signifi¬ cance of the impacts of an individual facility improvement, or true inertia in the overall urban system, inclusion of such measures in planning products may cause distraction and confusion among plan users. Of course, in some cases it may be useful to reassure decision makers that a particular measure does not vary across alternatives, or that while an insentitive measure, it meets some minimum or maximum standard; here, however, the decision to include such a measure should be a conscious choice based on need, rather than a reflex response based on past practice. Focusing could be furthered by producing multiple alternatives analysis reports tailored to the specific needs of varying audiences. For instance, a summary report in non-technical language, focusing on the major issues, might include qualitative summaries of the information without much detail and be far more widely distributed than the traditional CEAA reports. Recent publications outline what such a report might include (Cohen, 1978; CEQ, 1977). Reports could also be tailored to specific geographical areas or jurisdictions. For example, a state official might be primarily interested in the state resources required, integration with other state facilities, and aggregate impacts; representatives of individual communities, on the 116 other hand, may be primarily interested in how a facility or service may affect them at the disaggregate level. Presentations formats uniquely suited for displaying community-level impacts of alternatives have been proposed (e.g., Schofer and Stuart, 1974). Indeed, it may be possible to design reports in an efficient "nested" format, so that components of a single large document could be selected from the table of contents and reassembled into smaller, more specialized documents for particular audiences. A clear understanding of the results of alternatives analysis depends on conveying realistic descriptions of the alternatives themselves, as well as on presenting a useful set of measures. At the simplest level, it is desirable to develop concise and meaningful verbal labels for alter¬ natives; these are likely to be more instructive for plan users than the numerical or alphabetic tags developed in the technical planning process. The interrelationships between alternatives need to be clearly indi¬ cated, so that the plan users appreciate the degree to which some alter¬ natives are derived from others. For example, that one alternative may encompass all attributes of another, plus additional features, should be indicated in the labels, to facilitate understanding the tradeoffs illus¬ trated in the measures set. Likewise the performance measures need to be more clearly explained so that the user understands the definition and significance of the measure, the meaning of the scale or units used, and the geographical area or level of aggregation to which the measure refers. In approaching the CEAA process, as well as in preparing CEAA products, the analyst may find it helpful to consider - and respond to - a structured series of questions based on the ideas in this chapter. Such question sets may begin with the list presented below, although the essence of this 117 chapter is that the questions, their answers, and the CEAA report itself, should be uniquely defined to meet the needs of a particular context. 1. To what audiences is this effort addressed? What information will be most useful to each audience? 2. What are the problems which are to be solved by the chosen alter¬ native? What alternative views exist of these problems? 3. What are the alternatives? How are they related to each other in terms of physical layout and service structure? 4. How do the alternatives contribute to solving the problems? A. Where does each alternative fit into the relevant performance/ cost/impact space? What tradeoffs are involved when moving between alternatives in this space? B. How is the uncertainty of the relative location of alternatives in this space affected by uncertainties in data, forecasting procedures, and the potential range of measures which results from the inherent flexibility of alternatives? Taken together, what reduced set of meaningfully different alternatives remains? C. What is the apparent feasibility of alternatives? o In a cost sense: how do the capital and ongoing costs of alternatives relate to the ability to pay at all levels of government involved? o In a technological sense: how realistic are the cost, per¬ formance, and impact estimates given the current state-of- the-art? o In a political sense: how marketable are the alternatives in the context of the current political environment? D. what feasible, meaningfully different alternatives remain? 5. What staging (future system development) options are offered by the alternatives? What are the opportunity costs, precluding or including other, future alternatives, associated with the alter¬ natives in this study? What would be the result if the preferred alternative in this study were the last major system investment for the foreseeable future? 6. What is the apparent efficacy of the feasible, meaningfully different alternatives? Which ones may be worth the costs to all parties in¬ volved? To each of the identifiable interest groups and governments? 118 7. What feasible, meaningfully different alternatives remain in the choice set? which ones appear most attractive? why, in terms of the original problem definition? 8. In the answers to the questions stated above, what information, in the end, is of little value to the audience for this report/product? Will the audience be likely to find that all of its major questions are at least raised? Will most of these answers will be provided? Will both the questions and the answers be understandable in terms familiar to the audience? Will this product lead clearly to the next steps in the process? Conclusions It appears that a shift in emphasis of the alternatives analysis process and products is in order to increase the responsiveness to the needs of users. Recommended steps in this direction include a generally greater sensitivity to the importance of this issue as well as consider¬ ation of the following specific changes 1) a more explicit problem defi¬ nition and/or an evaluation that is more sensitive to the different perceptions of the problem, 2) efforts to improve CEAA methods and measures by dealing with such issues as the appropriateness of alternatives, system performance, flexibility, uncertainty, and product credibility; and, 3) efforts to improve CEAA products by sharpening the focus and adequately communicating the nature of the alternatives and meanings of their measures. Chapter Seven CONCLUSION This report has tried to make the clear case that an explicit considera¬ tion of the absolute worthiness of major transit investments is important in the evaluation and decision processes of CEAA. The importance of this issue, we submit, is independent of the specific objectives of a particular investment. That is, even if the basic intent of the proposed investment focuses on urban development or employment, the efficient use of public re¬ sources demands that serious efforts be made to assess whether or not the any or all of the options are truly worth their costs. Cost-effectiveness analysis, alone, does not do this job well, for it does not concentrate sufficiently strongly on the worthiness issue. The judgment of worthiness in cost-effectiveness is largely implicit, and it appears quite likely that decision makers, particularly at the local level, are too easily caught up in the relative worthiness question — which of the alternatives seems best — leaving open a strong possibility that the question of absolute worth may be overlooked. This argument is strengthened today because of ever-tightening budget constraints. In an era of fiscal conservatism, and in the face of a policy trend away from major capital investments in transit, ways must be found to identify those options which are most promising in an absolute sense, that is, those actions which have a high probability of returning to society an aggregate set of benefits exceeding their costs. Absolute worthiness analysis is limited in its scope because of its 119 120 necessary focus on monetary benefit measures, (largely) measures of traveler benefits. While this narrowness must be recognized, both by analyst and decision maker alike, pursuing this analysis still seems worth¬ while, particularly when one examines past CEAA efforts. Indeed, in the context of the latter, the suggestion here is to broaden the typical analysis by including an explicit consideration of the relationship between monetary (user) benefits and total costs. This suggestion would reflect a narrowing of concern if it were made in a vacuum; the approach proposed here calls for integrating cost-effectiveness and absolute worthiness so that a more com¬ prehensive picture may be presented in the decision process. The measure of absolute worthiness suggested in this research is, as argued above, at best a partial measure. It shows the relationship of some of the benefits to (perhaps) all of the monetary costs. There will certainly be other, important consequences associated with each option, some of which may be classified as benefits, and others as costs. Measures of these con¬ sequences should be clearly specified in the cost-effectiveness portion of the analysis and presentation. Reporting partial worthiness in terms of net worth of the alternatives provides a logical link between this perspec¬ tive on the options and that provided by cost-effectiveness analysis. It is the two views together which have the potential for improving the basis for action decision making. The general approach presented here is by no means limited to major, urban, fixed guideway transit options. It is applicable to other large scale transit options (e.g., major changes in a bus service); to other urban transportation programs, and even to significant non-urban passenger trans- 121 portation planning contexts. While the strategic framework is appropriate for freight service planning as well, the tactics for benefit measurement are less-well developed and less clear at this time. The approach is sug¬ gested for use on relatively large scale investments under the presumption that some of the key measures of effectiveness and benefits may not routinely be predicted for less significant actions. This is particularly true for smaller scale, highly flexible actions (e.g., bus service improvement), where it may be more sensible to forego the costs of impact forecasting be¬ cause it is relatively easy to make adjustments to the resulting in-service systems should they not meet local needs and expectations. It should be noted that the software system developed in this research need not be applied to implement some or all of the major concepts presented here. Calculations and presentations, adapted from the format suggested in the preceding chapters, may be developed by hand or using in-place com¬ puter capabilities. However, experiments with this system suggest that it is particularly helpful — and highly efficient — for developing and pre¬ senting an integrated view of the cost-effectiveness and absolute worthi¬ ness of alternative investments. It offers a rapid approach to screening may alternatives over many measures of effectiveness and cost, which may be easily applied by individuals with or without prior technical training. The sequence of presentations has been structured to allow the user, in only a matter of minutes, to get a strong overview of the characteristics of alternatives and their expected outcomes from several different per¬ spectives. The capability to cycle back through displays, and the high degree of user control over both the alternatives and measures presented, 122 promotes the review of options at several levels of detail and the screening of many alternatives, on a few measurement dimensions, down to a few alter¬ natives on many dimensions. The software system has also been found to be a highly promising way to generate elements for written reports and oral briefings. Indeed, the experience of the research team suggest that it may be quite timely to begin routine use of interactive, computer-supported presentation and evaluation tools of this type. This parallels the long history of development of ef¬ ficient and responsive computer tools for impact forecasting, which have been highly successful. Indeed, the software developed here may be viewed as an efficient, interactive, post-processing capability which can convert the results of other computer-based analyses into formats which are directly useful in technical and political decision making. In the reviews of planning documents and processes conducted in this effort, covering alternatives analysis and other aspects of public sector planning, it became increasingly clear that the degree of user-orientation of planning may have an important effect on the impact of that task. User orientation, here, means providing users — analysts and decision makers — with clear, concise, and meaningful informational products which have a high probability of supporting, and influencing constructively, the out¬ come of screening and choice of alternatives. User orientation may be achieved by structuring both the planning pro¬ cess and the evaluation task in appropriate ways. Some of the key issues which need to be addressed and resolved in designing the CEAA process include the following: Clarifying objectives set for the alternative investments and dealing with conflicting objectives early and explicitly Establishing a clear problem definition, to which alterna¬ tives are expected to respond, so that assessment of effec¬ tiveness becomes more straightforward, and can be accom¬ plished by all of the actors from a relatively unified perspective; Specifying the characteristics of the alternatives, in¬ cluding their associated operating plans, in understand¬ able ways; as well as finding simple ways to show the hierarchical interrelationships between alternatives; Designing an hierarchical screening process, to assure that the evaluation and decision effort is appropriately allocated across the competing options; Selecting a parsimonious set of clearr non-redundant measures of cost and effectiveness which show high promise for their ability to differentiate between truly different alternatives; Dealing explicitly and constructively with the principal sources of uncertainty, through sensitivity analyses and the careful use of parallel scenarios; and finally, Designing clearer, more understandable planning products which highlight the issues stated above to facilitate the use, and usefulness, of the results of the analytic planning effort. 124 This report offers some guidelines for meeting these challenges. Ulti¬ mately, the responsibility for good analysis and presentation falls on the practicing planner, and the degree of commitment and creativity which he or she can bring to the task at hand. Doing this job well, that is, putting sufficient energy into the evaluation and presentation aspects of CEAA, will be important not only for the future of our transportation systems, but also for the future of good analysis and those who perform it. REFERENCES Alexander, C. (1975), Notes on the Synthesis of Form, Harvard University Press. Beimborn, E.A. (1976), "Structured Approach to the Evaluation and Comparison of Alternative Transportation Plans," Transportation Research Record 619, Transportation- Research Board. Brown, M.H. (March 10, 1980), "The Price of Life," New York Magazine. CEQ (Council on Environmental Policy), (December, 1977), "Draft Regula¬ tions to Implement the National Environmental Policy Act. Cohen, H.S., J.R. Stowers and M.P. Petersilia (1978), "Evaluating Urban Transportation System Alternatives," Prepared for the Office of the Secretary, U.S. Department of Transportation by System Design Concepts, Inc. CTLA (Commission on the Third London Airport) (1971), Report series, Her Majesty's Stationery Office. Downes, J.D. (1980), "Variations of Household and Person Travel Time Budgets in Reading," Transportation and Road Research Laboratory, TRRL Report 166. Faigin, B. (1976), "1975 Societal Costs of Motor Vehicle Accidents," U.S. Department of Transportation, National Highway Traffic Safety Administration. Faludi, A. (1973), Planning Theory, Urban and Regional Planning Series (Vol. 7), Pergamon Press. Federal Register (December 7, 1978), "FHWA/UMTA Major Urban Transpor¬ tation Investment," pp. 57478ff. Federal Register (October 30, 1980), "Urban Transportation Planning," pp. 71990ff. 125 126 Federal Register (March 30, 1981) , "Urban Transportation Planning; Deferral of Effective Date," p. 19233. Foster, C.D., and M.E. Beesley (1963), "Estimating the Social Benefits of Constructing an Underground Railway in London," Journal of the Royal Statistical Society, v. 126, Part I. Jones, B.D. (Fall, 1979), "Distributional Standards and Service Delivery Decisions," The Urban Interest. Knight, R.L. and L.L. Trygg (1977), "Land Use Impacts of Rapid Transit: Implications of Recent Experience," U.S. Department of Transpor¬ tation, Office of the Secretary. ijKoppelman, F.S. (1981), "Uncertainty in Travel Behavior Forecasting," New Horizons in Travel Behavior, W. Brog, et al., (eds.), Lexington Books. Lichfield, N. (1971), "Cost-Benefit Analysis in Planning: A Critique of the Roskill Commission," Regional Studies, p. 157ff. Mao, J.C.T. (March, 1966), "Efficiency in Public Urban Renewal Expen¬ ditures Through Benefit Costs Analysis," Journal of the American Institute of Planners. Mendeloff, J. (May, 1980), "Reducing Occupational Health Risks: Uncertain Effects and Unstated Benefits," Technology Review. Peat, Marwick, Mitchell and Co., and Barton-Aschman Associates (1980), "Summary of the Results of the Operating Cost and Financing Require¬ ments Analysis for the Phase I Analysis," prepared for the Metro¬ politan Transit Authority of Harris County, Houston, Texas. Quade, E.S., Ed. (1964), Analysis for Military Decisions, the RAND Corporation. Savas, E.S. (April, 1978), "On Equity in Providing Public Services," Management Science. 127 Schnaiberg, A. (1980), The Environment from Surplus to Scarcity, Oxford University Press. Schofer, J.L. (1978), "Evaluating Transportation Alternatives," Proceedings, Seminar on Emerging Transportation Planning Methods, U.S. Department of Transportation, Research and Sponsored Programs Administration. Schofer, J.L. and P.R. Stopher (1979), "Specifications for a New Long- Range Urban Transportation Planning Process," Transportation, v. 8, No. Schofer, J.L. and D.G. Stuart (1974), "Evaluating Regional Plans and Community Impacts," Journal of the Urban Planning and Development Division, Proceedings, American Society of Civil Engineers, v. 100, No. UP1. Slovic, P., H. Kunreuther, and G.F. White (1974), "Decision Processes, Rationality, and Adjustment to Natural Hazards," in G.F. White (ed.), Natural Hazards, Local, National and Global, Oxford University Press. Stopher, P.R. and A.H. Meyburg (1976), Transportation Systems Evaluation, Lexington Books. TCA (Tottenham Court Associates) (1981), "Demand Estimation Techniques for the Proposed Dan Ryan Rapid Transit Extension," prepared for the City of Chicago, Department of Public Works, Bureau of Transportation Planning and Programming. Thomas, E.N., and J.L. Schofer (1970), Strategies for the Evaluation of Alternative Transportation Plans, National Cooperative Highway Research Program Report 96, Highway Research Record. Wachs, M., B. Hudson and J.L. Schofer (April, 1974), "Integrating Localized and System-wide Objectives in Transportation Planning," Traffic Quarterly. 128 Wachs, M., and J. Ortner (1979), "Capital Grants and Recurrent Subsidies: A Dilemma in American Transportation Policy," Transportation, v. 8, No. 3. Zahavi, Y., and A. Talvitie (1980), "Regularities in Travel Time and Money Expenditures," Transportation Research Record 750, Transporta¬ tion Research Board. APPENDIX A INTEGRATING PROJECT WORTHINESS INTO A COST-EFFECTIVENESS FRAMEWORK FOR ALTERNATIVES ANALYSIS INTERACTIVE COMPUTER GRAPHICS PROGRAM USER'S MANUAL Mark A. Turnquist School of Civil and Environmental Engineering Cornell University Ithaca, NY 14853 prepared for Urban Mass Transportation Administration U.S. Department of Transportation under Grant IL-11-0026 to Northwestern University Evanston, Illinois April, 1980 129 130 1. Introduction Program GRAFAA (graphie analysis of alternatives) is largely an input- output processor, which does some limited computations for analysis of a data set describing alternative transit systems to be evaluated. Its princi¬ pal purpose is to manipulate an input data set into a form suitable for graphic display, and then to provide such displays in an interactive mode. The user interacts with the program by providing answers to a series of questions. This allows him/her to specify which subsets of data are to be displayed next, and the type of display desired. The program then provides the selected display on the interactive graphics terminal, and waits for further instructions. Iterations of this process continue until the user has examined all desired subsets of the data, at which point the program can be terminated. The program has been designed specifically to allow analysis of invest¬ ment worthiness questions within a context of more general cost-effectiveness analysis. It thus allows the user to examine subsets of cost measures alone, effectiveness measures alone, or selected combinations of cost and effective¬ ness measures together. The program includes several different graphic output formats, such as bar charts, line charts, and two-dimensional plots of trade-offs and sensitivity, as well as tabular summaries of data. This manual is intended to provide the information required to use the program. Part 2 of the manual is a brief non-technical overview of the program's features and capabilities. This is designed primarily for managers 131 and decision-makers who may use the information provided by the program, but who will not be involved in actual data set preparation or execution of the program. Part 3 of the manual is directed at technical analysts who will be responsible for preparing inputs to the program and working directly with the analysis. This section includes a more detailed description of program capabilities, options and limitations. It also includes details on preparing data sets for the program, with an example of a complete data set which can can be used for test purposes. An Appendix to this manual describes error messages generated by the program, with an indication of the probable cause and suggestions for corrective action. 2. Non-Technical Overview of the Software The computer software system developed to assist in the analysis of major urban transportation investment alternatives comprises five principal modules. 1) DATA INPUT - This module reads a data file which has been prepared offline, checks for errors, and processes the input data so that it is ready to be displayed in various forms by the other modules. The system is designed to accommodate up to 20 alternatives with as many as 60 measures of performance for each, up to 46 of which may be user-specified. There are 14 measures built into the system, reflecting the most common and generally applicable measures for alternatives. 2) DISPLAY - This module provides a twenty-five line (one page) descrip¬ tion of each alternative under consideration. 3) TIME STREAM - This module prepares a graphic display of the time streams of lump-sum capital costs, annualized capital costs, operating and maintenance costs, total costs, ridership and revenues, operating deficit, and total deficit over the pre-specified time 132 horizon (up to 25 years) for analysis. The module can present one measure for a single alternative, two measures for a single alter¬ native, or one measure for two alternatives at any one time. For each measure, it reports the present and equivalent annualized value of the time stream. The module appropriately accounts for costs occurring after the fixed analysis period. It also allows the user to test the effect of delaying the implementation of any alternative (other than the "do-nothing" or "no build") for to five years. 4) MATRIX - This module displays a matrix of measures for the various alternatives. It can show some measures for some alternatives, some measures for all alternatives, all measures for some alternatives, or all measures for all alternatives. No more than 11 alternatives or 5 measures can be seen on the display screen at one time. The module can also display five primary measures pre-selected within the software to provide a general overview of the characteristics of the alternatives. Most of the measures included in this module are internally specified; some may be defined by the user. The module offers three modes of display. It can show the simple measures for the selected alternatives; the ratio of the measure for each alter- nativè to the same measure for a baseline alternative; or the per¬ centage deviation of each measure for the selected alternatives relative to some baseline alternative. 5) WORTH - This module provides various perspectives on the economic worthiness of the alternatives in terms of the net present worth. It comprises four separate sub-modules. Sub-module SUMMARY provides a simple matrix display of the relative net worth of each alternative, along with five other 133 measures of the alternatives, three of which may be user selected. This sub-module also prepares a graphic, profile plot of the alter¬ natives in these six dimensions. Sub-module TRADE prepares two-dimensional trade-off plots relating a cost measure to one of several impact measures. The available cost measures are annual equivalent capital costs and annual equivalent total costs. Some of the impact measures are pre-specified, while others are user-selected. RANGE is a sub-module which displays a graphical plot of traveller benefits versus total costs of each alternative. Built into this sub-module is a sensitivity analysis capability which computes traveler benefits for high and low values of user travel time. SENSIT is a sub-module which computes and displays in matrix form the relative net worths of the competing alternatives as a function of both the value of travel time and the value of auto¬ mobile operating costs. The system has been designed from its inception to be user-oriented and interactive. Operation of the system is quite easy, and it is designed to be self-teaching and tolerant of mistakes on the part of the user. It has been operated by several graduate students and practicing transportation planners without difficulty. The remaining parts of this users' manual describe the system in much more detail, with full technical documentation and instructions for preparing input and tracing errors. These sections are intended primarily for the technical analysts who will use the software directly. 134 3. Technical Description of the System This section of the User' Manual is written primarily for the technical analyst who will be the principal direct user of the program. It describes the overall structure of the program, options available within the program, preparation of input data, and examples of output results. The program is written entirely in FORTRAN IV, and has been implemented on the CDC 6600 computer at Northwestern University and the IBM 370/168 at Cornell University. In addition to the normal FORTRAN library subprograms, the program utilizes the PLOT 10 package of subroutines for interactive graphics. This software package has been developed by Tektronix, Inc., primarily for use on their graphics terminals. The program has been tested on a Tektronix 4010-1, 4013, and 4014 graphics terminal. Minor modifications have been followed as much as possible in coding the program so that problems in transferring to other computers could be minimized. However, it is to be expected that some minor program modifications would be required to implement the program on other computer systems. The program comprises approximately 4300 statements, of which 20-25% are comments. It requires approximately 115K (octal) words of core memory to load and execute, about half of which is for the PLOT 10 software. 3.1 Structure of the Program The basic structure of the program is indicated in Figure 3-1. A main sub-program is linked to several major modules which perform various specific functions. These modules include data input; display of alternative descriptions; analysis of time stream data; matrix display of various measures for selected alternatives in a tabular form; and investment worthiness analysis, including Figure 3-1. Structure of Program Modules. CO CJ1 136 several different types of displays. Each of these modules is in turn linked to various utility subprograms, including the PLOT 10 routines. The following subsections describe each of the individual modules. 3.1.1 Data Input-Subroutine READIN Subroutine READIN reads the data file (attached to the program as TAPE3) and then processes the information and stores it in a form that is used or displayed in other routines of the program. The input file consists of system variables (constant for all alternatives) and alternative-specific information (costs, ridership, VMT, etc.). The system variables are read first, followed by blocks of information pertaining to each alternative. The structure of the input file is discussed in detail in section 3.2. 3.1.1.1 Capital Cost Inputs Capital cost items for each alternative are input as lump-sum pay¬ ments in particular years. The program then computes equivalent annual values for each of these costs over their useful life. Each lump-sum cost is amortized separately, and then the equivalent annual values are summed to yield a stream of equivalent annual capital costs over the period of the analysis. A major reason for doing the analysis this way is to account correctly for the salvage value of capital items at the end of the planning horizon. A simple example will help to illustrate this. Suppose one alternative calls for construction of a busway in year 1, at a cost of $30 million. Also in year 1, 25 buses are purchased, at $40,000 each, to begin service. In 137 year 6 we anticipate having to add 20 additional buses also at $40,000 each, to handle increased demand. Suppose the guideway has a 25 year useful life, buses have 10 year useful lives, and our time span of analysis is 15 years. Because the initial buses will wear out before the end of analysis period, they must be replaced in year 11. Let us assume a discount rate of 10%; then the capital recovery factors for buses and guideway are .163 and .110 respectively. The stream of annual equivalent capital costs is as shown in Table 3-1. Table 3-1. Annual equivalent capital costs ($ million) for example. 1 2 3 4 5 Busway 3.3 3.3 3.3 3.3 3.3 3.3 3.3 . 3.3 Buses (A) .163 .163 .163 .163 .163 CO KD « .163 . .163 (B) - - - - - .163 .163 . .163 TOTAL 3.463 3.463 3.463 3.463 3.463 3.626 3.626 . . 3.626 Note that the costs of the busway have been amortized over 25 years, even though the analysis horizon is 15 years. This reflects the fact that there is a salvage value equal to 10 years' use left in the busway at the end of the analysis period. Hence, the costs used in the analysis reflect only the actual use of resources over the planning period. The same is true for the buses purchased in year 11 to replace those purchased in year 1. Thus, the stream of equivalent costs for the (A) set of buses is constant over the entire period of analysis. Because different capital items will be acquired in various years (e.g. buses (A) and (B)), the stream of total equivalent annual costs will not be constant over time. However, this stream itself can be converted to a uniform 138 annual equivalent capital cost value if desired. This computation is also done by subroutine READIN. 3.1.1.2 Other Time Stream Data In addition to capital costs, the data for each alternative includes estimates of operating and maintenance costs, transit system ridership, person hours of travel (all modes) and vehicle-miles of auto travel (VMT) for selected years in che analysis period. The user has several options for interpolating be¬ tween specified points, or extrapolating from the last specified point to the end of the analysis period. Interpolation options are: . linear interpolation; or . no interpolation (intermediate values are assumed to be zero). For extrapolating to the end of the period, the user has three options: . constant value (the last value input is projected to the end of the period); . linear extrapolation of last two data points; or . percentage rate of change. Figure 3-2 shows an example of these three extrapolation options. This set of interpolation and extrapolation options provides the user with substantial flexibility for easily inputting various types of data. For example, if ridership predictions are made for 1, 5, 10 and 15 years after implementation, it may be desirable to input these four values and allow the program to interpolate linearly between them for intermediate years. As another example, if a current estimate of total VMT is available, and it is projected to grow at a rate of 3% per year, the user need only input the current value and the rate of change for extrapolation, and the program will fill in values for the remaining years. 139 measure value time Figure 3-2. Illustration of Extrapolation Options 140 3.1.1.3 Data for User-Specified Measures The program will accept and prepare for display, data for up to 46 user-specified measures of effectiveness for each alternative. This gives the user a great deal of flexibility in specifying measure sets for any application. Because the program has no advance knowledge of these measures, this data will not be processed or manipulated. It will simply be stored, and displayed with other measures in the appropriate output routines. 3.1.1.4 Input Processing In addition to computing net present value and equivalent annual value for the time stream data discussed in sections 3.1.1.1 and 3.1.1.2 above, READIN also computes other measure values for each alternative from the data input. These include: 1) transit ridership 2) passenger revenues 3) traveler benefits (savings in travel time and vehicle operating costs, relative to "no build") 4) operating deficit (operating and maintenance costs, less passenger revenue) 5) total cost (capital plus operating and maintenance) 6) total deficit (total costs, less passenger revenue) 7) net worth (traveler benefits, less additional total costs over and above "no build" option). Values of traveler benefits are computed for three different values of vehicle operating cost and the value of time saved, for use in sensitivity analyses. Five partial cost-effectiveness measures are also computed, for use in the investment worthiness analysis. They are: 141 1) annual capital cost/route mile 2) annual operating deficit/passenger trip 3) annual total cost/capita 4) annual operating deficit/capita 5) annual total cost/passenger trip. All of the basic measure value computations are done for each of three different values of the discount rate. This is done so as to provide the ability to examine sensitivity of the evaluation to this important parameter. However, the output facility for providing this information is not yet operational. 3.1.1.5 User Interaction and Error Checking There are no online inputs to the READ IN subroutine and normally no online outputs. However, there are several data checks built into READIN. If an error is found in the input file, READIN will print an error message, identify the data record number, and print that data record at the inter¬ active terminal. This will aid the user in locating and correcting the error. 3.1.2 Display Descriptions of Alternatives-Subroutine DISP Subroutine DISP prints to the terminal a description of each alternative which may be up to 25 lines in length. This fills the screen of the Tektronix terminal (one "page"). The description must be input as part of the data file (see section 3.2). It is stored, and then printed verbatim when desired by the user. Normally, this description should include the number and abbreviated name to be assigned to each alternative for purposes of the analysis, and information regarding the location, service area, service characteristics, impacts, implementation schedule, etc. An example of such a description 142 is shown in Figure 3-3. However, it should be emphasized that the format and content of the description are controlled entirely by the user through preparation of the input file. The program simply prints what it has read. 3.1.3 Plotting of Time Stream Data - Subroutine TIMSTR Subroutine TIMSTR prepares and presents a graphic display of time streams of lump-sum capital expenditures, annualized capital costs, operating and maintenance costs, total annual costs, ridership and revenues, operating deficit and total deficit over the length of the planning horizon (up to 25 years). The user interacts with TIMSTR by selecting combinations of measures and alternatives to be displayed. After each display is drawn on the screen, the computer waits for further instructions. When TIMSTR is entered, an initial message is written explaining the basic options available. The computer then waits for the user to type the word "GO" followed by a carriage return, before proceeding. The first request is for the type of measure to be displayed in the first plot. The program writes a list of candidate measures and keywords, and then waits for the user to enter the keyword for the desired measure. This keyword should be entered in the first three columns of the line, with no leading or embedded blanks. If the user's entry does not match one of the available keywords, an error message is written on the screen, and the user is asked to reenter the keyword. DESCRIPTION OF ALTERNATIMES NAME - ADUANCED BUS ALT 1. ADUANCED BUS SYSTEM MODE iU CORRIDOR DESCRIPTION— ALL BUS. MAJOR SERUICE ON STATE STREET, FEEDER DISTRIBUTOR SERUICE IN N.CORRIDOR TO THE UN- IUERSITY TO RETURN TO MAIN PROGRAM. ? Figure 3--3. Example of Alternative Description. 144 After the first keyword has been entered, the user 1s asked to specify a second keyword, if desired, TIMSTR divides the screen into an upper half and a lower half, allowing two plots to be displayed simultaneously. This is often useful for visual comparison of two plots. The user has the option of seeing a single measure for two alternatives; two measures for a single alternative; or, of course, a single measure for one alternative (which will produce a single plot in the top half of the screen). If the user enters "ALT" when asked for a second measure, the program will assume a single measure is desired, and will move on to the next step, the selection of the desired alternative(s). After the user has selected one or two measures, the program will ask whether or not a list of alternatives should be written on the screen. Input of desired alternatives to be displayed is by number. If the user does not remember which numbers correspond to which alternatives, he/she should enter "YES" in order to see the list of names and numbers. After the listing (or if the user entered "NO" when asked if the listing should be written) the computer will ask the user to enter the number of the first alternative to be plotted. This entry is read with a FORTRAN 12 for¬ mat, so the number must be right-justified in the first two columns of the line or it will not be interpreted correctly. If an alternative other than the "no build" is specified, a message will be written on the screen regarding the current schedule for implementation of that alternative, and the user is given an opportunity to accept the current schedule, or to delay it by a period of up to five years. If an alternative is delayed, values for the selected measure(s) from the "no build" alternative will be inserted into the stream(s) for the selected 145 alternative over the period of the delay, to reflect the operational conse¬ quences of delaying implementation. If the user has specified two measures and one alternative, the two plots to be presented are completely defined and will appear on the screen without further user action. However, if only one measure has been specified, the user will be asked to enter a second alternative number, if desired. If only one alternative is desired, entering "-1" as an alternative number will cause the program to proceed with the plot. An option which can be useful is to specify one measure and then the same alternative number twice. In one case, implementation can be delayed and in the other case not. This provides a convenient visual comparison of consequences of delays in terms of streams of costs, revenues, etc., as well as changes in summary statistics of present value, etc. The plots which appear will be either line charts or bar charts. Examples are shown in Figure 3-4. Line charts are produced if the user has specified that interpolation between data points should be performed. Bar charts will be drawn if there is to be no interpolation. The interpolation indicator for each data type is coded into the input data file, as discussed in section 3.1.1.2. Details of file preparation are discussed in section 3.2. Under each plot, the values for present value and equivalent annual value are written. Then the program pauses, and waits for further instruc¬ tions from the user. This sequence of questions and user answers to define desired plots continues until the user enters "EXIT" after viewing a plot. This causes control to return to the main program. The menu of analysis modules will be written on the screen, and the program will wait for further instructions. $ 40—1 M I L L I 20 — 0 N RIDERSHIP< > tu REUENUE FOR ALT. 10 0—G-—©—©—-O-—©—G --^1—e- —-cr- 1—60 — 40 0~ti^ T 1980 1985 AT DISCOUNT RATE .10 PRES UAL» 1 n 1990 1995 YEAR 122.3 EQUIU ANN UAL» 16,1 •20 M I L L I 0 N $ TOTAL DEFICIT FOR ALT. 10 50 — N 0" 1 1 1 1980 * 1 1 1985 1 YEAR 1990 n 1995 AT DISCOUNT RATE .10 PRES UAL» 283.8 EQUIU ANN UAL» 37.3 PAUSING. HIT 'RETURN' TO CONTINUE,OR ENTER TO TERMINATE TIME STREAM ANALYSIS. ? Figure 3-4. Example Output from TIMSTR. 147 3.1.4 Tabular Summary of Data - Subroutine MATRIX Subroutine MATRIX provides a display of a two dimensional matrix in which the rows are alternatives and the columns are measures. Thus each entry of the displayed matrix contains the level of a given measure achieved by a given alternative. When MATRIX is entered, a complete list of alternatives is presented, and the user is given two options in selecting the alternatives to be included in the display. If "ALL" is specified, all the alternatives will be included; if the user is interested only in looking at a subset of the alternatives, "SOME" may be specified. In this case, the program will ask the user to specify the number of alternatives to be included in the display, and subse¬ quently to enter the alternative numbers one at a time. Alternative numbers must be entered right-justified in the first two columns of the line. Similarly, a complete list of measures is presented to the user, who has four options in selecting the measures to be included in the display. In addition to "ALL" or "SOME" the user can also ask to see a set of five prespecified primary measures by entering "PRIM," or a set of five prespecified partial cost-effectiveness measures by specifying "PCES." Once the alternatives and the measures are specified, the user is asked to specify one of the three display modes available. Display mode 1 simply provides the values of the selected measures for the selected alter¬ natives. Display mode 2 provides the ratio of those measures to the values of the selected measures associated with some user-specified alternative termed the "base alternative." Display mode 3 replaces the ratio in display mode 2 by the percent deviation from the base alternative. In display mode 3, 148 the percent deviations are displayed only for those cases where they exceed a certain threshold value, specified by the user. Up to 11 measures and 5 alternatives can be seen on the display screen at one time. If more measures and/or alternatives have been selected, the matrix will be shown in multiple "pages." After each page of the matrix is displayed, the user can continue to the next page or terminate the current display. If the display is terminated, four additional options are available: (1) to see more measures for the same subset of alternatives, (2) to specify new subsets of alternatives and measures, (3) to see the same alternatives and measures but with a different display mode, or (4) to return to the main program. An example of output from Subroutine MATRIX is shown in Figure 3-5. 3.1.5 Project Worthiness Analysis This module provides various perspectives on the economic worthiness of the alternatives. It comprises a main routine and four sub-modules to perform specific tasks, as follows: 1) SUMMARY - tabular and graphic summary of major measures for selected alternatives 2) TRADE - plot of tradeoffs between two selected measures for all alternatives 3) RANGE - plot of sensitivity of monetary traveler benefits to changes in value of tiem and vehicle operating cost 4) SENSIT - tabular summary of net worth sensitivity to value of time and vehicle operating cost These four sub-modules are described individually in the following sections. MODE OF DISPLAY- 1 MEASURES ALTERNATIUES 1 4 6 12 21 ANN EQUU ANN EQUU ANN EQUU ANN EQUU DOOR TO CAPITAL TRAUELER TOTAL TOTAL * DOOR SPD COST SAVINGS COST COST PER MPH CAPITA 1\ADVANCED BUS 4.61 96.49 24.61 18.54 11.70 2 BASE HR 28.16 91.74 51.79 39.02 17.00 3 INTERMEDIATE LR 36.13 105.99 58.70 44.22 15.20 4 NO BUILD TSM BUS .85 0.00 18.20 13.71 13.00 5 MINIMUM HR 23.28 82.72 49.86 37.57 15.00 6 BASE HR+ N. EXT 30.89 95.99 55.39 41.73 17.00 7 MIN HR+.5NE. EXT 21.95 86.48 49.63 37.40 16.00 8 BASE HR+ NE. EXT 33.77 96.98 58.12 43.79 18.00 9 MAXIMUM HR 37.59 103.99 62.02 46.73 18.50 10 MINIMUM LR 3Q.30 99.55 53.38 40.22 15.00 11 MAXIMUM LR 72.73 111.00 97.69 73.61 1 17.20 PAUSING/HIT 'RETURN' TO CONTINUE/ OR ENTER TO TERMINATE THIS DISPLAY. ? Figure 3-5. Example of Output from MATRIX. 150 3.1.5.1 Project Worthiness Summary - Subroutines IWSUMM and PROFIL IWSUMM produces a matrix display of 3 to 7 major measures for a set of alternatives selected by the user. Three measures are automatically provided by the program. They are: 1) net worth; 2) monetary value of traveler benefits; and 3) total costs. Traveler benefits are time savings and vehicle operating cost savings relative to the "no build" option, with time savings converted to monetary value. They are expressed on an annual basis. Net worth is computed by subtracting the difference in total annual cost between the selected alternative and the "no build" alternative, from the traveler benefits. Thus, the net worth of the "no build" alternative is zero by definition, and all other alternatives are evaluated relative to the "no build." In addition to these three measures, up to four more measures can be selected by the user. The program will print a list of candidate measures from which the user may choose additional ones for display. The list of candidate measures itself is input as part of the initial input file, and may contain up to 10 candidate measures. However, once the program is executing this list is fixed; the user may select from the list, but may not select a measure not on the list or modify the list. In the summary table the measures are presented as columns and the alternatives are presented as rows. The information pertinent to a given alternative is given in two successive rows; the first row gives the alter¬ native number and alternative name; the second gives the measure values for that alternative. Figure 3-5 shows an example of output from IWSUMM. INVESTMENT WORTHINESS SUMMARY ANN EQUU ANN EQUU ANN EQUU ANN EQUU TOTAL TRAVELER RELATIVE TOTAL $ COST SAVINGS NET WORTH 1 ADVANCED BUS 24.61 96.49 2 BASE HR 51.79 91.74 3 INTERMEDIATE LR 58.78 105.99 4 NO BUILD TSM BUS 18.20 0.00 5 MINIMUM HR 49.86 82.72 90.08 58.15 65.50 0.00 51.06 COST PER TRIP .71 s .87 : 1.26 : .56 i ,91 CONSTRCT RESIDENT AIR POLL EMPLOY RELOC TONS/YR JOBS ECUIV CO 150.00 0.00 95700.00 1000.00 79.00 90500.09 425.00 35.00 90600.00 90.00 0.00 95700.00 550.00 52.00 91500.00 PAUSING,HIT "RETURN" TO CONTINUE OR ENTER TO TERMINATE THIS DISPLAY. ■7 Figure 3-6. Example of Output from IWSUMM, 152 In selecting the alternatives, the user is given two options -- "ALL" and "SOME." The user should specify "ALL" to see the measures for all the alternatives. He/she should specify "SOME" to select a subset of the alter¬ natives. The user is then asked to enter the alternatives sequentially by their numbers. Alternative numbers must be right-justified in the first two columns of the line. Nc more than 5 alternatives can be displayed at one time. If the set of selected alternatives includes more than 5 alternatives, the summary table is given on more than one page. After each page is displayed, the user can continue to the next page by pressing the RETURN key, or specify "END" which terminates the current display. If "END" is entered, the user is given the option to: (1) select a new set of alternatives; (2) obtain a profile plot for the alternatives and measures just presented; or (3) return to the invest¬ ment worthiness main routine for selection of additional displays. If the profile plot is requested, a call is made to subroutine PROFIL to draw the plot for the alternatives and measures just presented by IWSUMM. Values plotted are percentages of the maximum value attained for a given measure across all alternatives (not just those selected for plotting). Subroutine PROFIL is not called directly by the user. It is accessed only through subroutine IWSUMM, and does not itself have any interactive elements. It simply draws a plot. An example of the output from PROFIL is shown in Figure 3-7. This output corresponds to the data shown in tabular form in Figure 3-6. PROFILE PLOT THIS PLOT SHOWS THE RELATIUE PERFORMANCE OF SELECTED ALTERNATIUES FOR SEUERAL MEASURES OF EFFECTIUENESS. UALUES FOR EACH MEASURE ARE REPRESENTED AS PERCENTAGES OF THE LARGEST UALUE OF THAT MEASURE ATTAINED BY ANY ALTERNATIUE. ANNUAL EQUIU TOTAL COST MAN- 97.69 ALT= 11 ANNUAL EQUIU TRAUELER SAUINGS MAX= 111.00 ALT= 11 ANNUAL EQUIU RELATIUE NET WORTH MAX= 90.08 ALT= 1 ANN EQUIU TOTAL COST PER TRIP MAX= 1 78 ALT= 11 CONSTRUCTION EMPLOYMENT- # JOBS MAX- 1150.00 ALT= 9 RESIDENTS RELOCATED MAX= 88.00 ALT= 9 AIR POLL TONS/YR EQUIU CO MAX- 95700.00 ALT= 1 -100 -50 o ADUANCED BUS < 1> -100 X -50 BASE HR < 2) a INTERMEDIATE LR < 3) + MINIMUM HR < 5> NO BUILD TSM BUSC 4) PAUSING. HIT 'RETURN' TO CONTINUE. ? Figure 3-7. Example of Output from PROFILE. cn CO 154 3.1.5.2 Measure Tradeoffs - Subroutine TRDOFF Subroutine TRDOFF provides a display of two-dimensional tradeoffs between a cost measure and an impact measure. Each alternative is represented by a point in a two-dimensional space. The point position corresponds to the values of the selected measures for that alternative. Each point is labeled with the alternative number next to it on the right. Figure 3-8 illustrates output from subroutine TRDOFF. When TRDOFF is entered, the user is asked whether or not the list of tradeoff measures should be displayed. If the answer is "YES," cost measures are listed first. The selected cost measure will appear on the x-axis of the plot. Currently available cost measures are: 1) annual equivalent capital cost 2) annual total costs. The user is then asked to select one of these two cost measures. If the user enters "NO" when asked whether or not to list cost measures, the program will ask for selection of a measure number immediately. The measure number is read with a FORTRAN 12 format, so it must be entered right-justified in the first two columns of the line. If the user requested listing of measures, the list of candidate impact measures will then be written on the screen. This list will include two measures which are built into the program: 1) annual equivalent operating deficit 2) average annual ridership. Other candidate impact measures will be any additional measures specified by the user as part of the input data. This input is described in more detail in section 3.2. T R A N S E N E R G Y 0 0 0 G A L L G A S 75000—1 74000— 73000 — 72000- 0 T 20 °ii0 O? 03 "T" 40 60 80 on 100 ANNUAL EQUIU TOTAL COST PLEASE HIT RETURN TO CONTINUE. Figure 3-8. Example of Output from TRDOFF. 156 The user will then be asked to select an impact measure from the candi¬ date set. This measure will appear on the y-axis of the plot. As with the cost measure, this number must be entered right-justified in the first two columns • the line. When the plot is completed, the computer will wait for further instructions. The user may repeat the plot for two other measures, or may return to the main investment worthiness routine. 3.1.5.3 Cost-Benefit Range - Subroutine CBRNGE Subroutine CBRNGE provides a simple display of the sensitivity of monetary benefits to the assumed value of travel time savings measured in dollars per hour and to the estimated vehicle operating costs measured in cents per mile. The present worth of total costs is shown on the horizontal axis and the value of traveler benefits is shown on the vertical axis. Since the purpose of the plot is to illustrate the sensitivity of traveler benefits to changes in value of travel time savings and vehicle operating costs, a vertical line between the lowest and highest values of traveler benefits is drawn to represent the sensitivity range for each alternative. Currently, three points are plotted on each line, the highest and lowest values of traveler benefits and the "base values." Upper and lower values are obtained by increasing or decreasing the assumed values of travel time savings and vehicle operating costs by 50%. The "base values" are those input as part of the data file. The line for each alternative is labeled by the alternative number above the upper-most point. A dashed line is drawn to designate the worthi¬ ness boundary where traveler benefits equal total costs. 157 When the display is completed, the program waits for the user to give permission to erase the display and return to the main worthiness routine. No other user interaction is included in CBRNGE. Figure 3-9 illustrates the output from CBRNGE. 3.1.5.4 Sensitivity Analysis of Net Worth - Subroutine SENSIT The table provided by subroutine SENSIT shows how benefits change with different assumptions regarding the value of travel time savings and vehicle operating cost. It is a tabular (and more detailed) version of the information presented graphically in subroutine CBRNGE. Base values for value of time savings and vehicle operating cost are contained in the input data file (see section 3.2). High and low values for these two parameters are computed as 150% and 50% of the input values. The sensitivity analysis table has five columns: 1) value of travel time savings ($/hour); 2) vehicle operating cost ($/mile); 3) annual equivalent total costs; 4) annual equivalent value of traveler benefits; and 5) annual equivalent net worth. The information for each alternative is contained in four rows, corresponding to 1) low value of time, low operating cost; 2) low value of time, high operating cost; 3) high value of time, low operating cost; and 4) high value of time, high operating cost. COSTS AND POTENTIAL RANGE OF USER BENEFITS FOR EACH ALTERNATIUE 200- T R A U E L E R B E M E F I T $ 0 0 0 150- 100- 50 — 0 3g o ©6 o © 0 ' 1—0j 1 1 j j 20 40 -1 1 1 1 60 80 100 TOTAL COST <*000> PLEASE HIT RETURN TO TERMINATE THIS DISPLAY ? Figure 3-9. Example Output from CBRNGE tn CO 159 At the beginning of the first row for each alternative, the alternative name and number are written. Figure 3-10 shows an example of output from SENSIT. When SENSIT is entered, the user is given the option of seeing the list of alternatives. The program then asks whether the user wishes to see "ALL" of the alternatives or only "SOME" of them. If "SOME" is entered, the user is then asked to input the alternative numbers sequentially. These numbers must be entered right-justified in the first two columns of the line. Up to 5 alternatives can be displayed at once. If the set of selected alternatives is larger than 5, the sensitivity analysis table is shown on more than one page. After each page is displayed the user can continue to the next page by pressing RETURN, or enter "END" to terminate the current display. If "END" is entered, the user has the option to select a new sub¬ set of alternatives, or to return to the main worthiness routine. 3.2 Structure of Data Input File The basic data file read by subroutine READIN is organized into blocks. The first block is general information, and contains from 4 to 96 records. Subsequent blocks contain alternative-specific information for each alter¬ native to be included in the analysis. Thus, if there are NALT alternatives to be considered, the data file will contain NALT + 1 blocks of information. All of the information for one alternative is read before proceeding to the next alternative. Table 3-2 gives details on the structure of the block of general infor¬ mation. Table 3-3 provides similar detail for the blocks of alternative- specific data. TRAUELER BENEFITS SENSITIUITY UALUE UALUE ANNU ANNUAL ANNUAL E OF TIME OF COST AL EGUIU EGUIU TR GUIU REL */HR SVMI TOTAL C AUELER S ATIUE HE OST AUINGS T WORTH ADUANCED BUS 1.40 .13 24.61 48.25 41 .84 1.40 .33 24.61 120.76 114.35 4.20 .13 24.61 72.22 65.82 4.20 .38 24.61 144.74 138.33 BASE HR 1.40 .13 51.79 45.87 12.28 1.40 . 33 51.79 114.81 81.22 4.20 .13 51.79 68.67 35.08 4.20 .38 51.79 137.61 104.02 INTERMEDIATE LR 1.40 .13 53.70 53.00 12.50 1.40 .33 58.70 132.65 92.15 4.20 .13 58.70 79.34 38.84 4.20 - 1 .38 58.70 158.99 118.49 NO BUILD TSM BUS 1.40 .13 18.20 0.00 0. 00 1.40 .38 18.20 0.00 0.00 4.20 .13 18.20 0.00 0. 00 4.20 .38 18.20 0. 00 0.00 MINIMUM HR 1.40 .13 49.86 41.36 9.70 1.40 .38 49.86 103.53 71.86 4.29 . 13 49.86 61.92 30.25 4.20 .38 49.86 124.08 92.42 PAUSING,HIT "RETURN" TO CONTINUE OR ENTER TO TERMINATE THIS DISPLAY. 7 Figure 3-10. Example Output from SENSIT. o Table 3-2. Data structure - general information Record No. Format Variable Name Variable Description 1 5 F10.2 DIS (1 ) ] Discount rates - 3 rates (low, medium and high) are DIS(2) \ the annual discount rates in decimal form. DIS(3) j VOC(2) Vehicle operating cost (auto) in dollars per mile for average study area auto and speed in base year dollars. VOC(1) and VOC(3) are computed by the program at 50% lower and 50% higher respectively than the input value for VOC(2). VTT(2) Value of travel time - 1n constant base year dollars per hour averaged for the study area. VTT( 1) and VTT(3) are computed by the program as 50% lower and 50% higher respectively than the input value for VTT (2). 2 F10.0,IX,14, POP Population of study area 1n thousands for design year. 4(3X,12 ) TNOW Time Now (integer) - the base year date (e.g., 1980); the year to which cost and other noted data pertains. TSPAN Time span (integer) of analysis (1-25 years); the time streams of cost Items will be projected to this year. NALT Number of alternatives (integer, 1-20). NOBLD Number of "no-build" alternative (Integer). NMEAS Number of evaluation measures (integers) whose values are input in addition to those calculated within the model). Up to 46 additional measures can be input. Table 3-2. Continued Record No. Format Variable Name Variable Description 3+NMEAS+2 4A8 NAMSUR(I.J) Name of measure - alphanumeris string of up to 32 characters that will be the axis labels for the measure displays. The four 8-character words of each data record fill one column of the matrix. NMEAS+3 -»-2*NMEAS+2 4A8 HEADER(I,J) Column header measure name - Alphanumeric string of four 8-character words that will be the four-line column heading for the measure displays. The four 8-character words of each data record fill one column of the matrix. 2*NMEAS+3 5(1X,12) IPM(I) Primary measure numbers (integer). Up to five numbers corresponding to the measures selected as primary measures (e.g., 2, 8, 12, 21, 36). 2*NMEAS+4 10(1X,I2) IWCAN(I) Investment worthiness candidate measure numbers (integer). Up to ten non-cost measures can be speci¬ fied for future use in the investment worthiness matrix display. Table 3-3. Data structure: alternative-Aspecific information Data Types Record No. Format Variable Name Variable Description 1 2A8 NAMALT(I.J) Name of alternative (alphanumeric) - the 16- character name of the alternative is stored in the two-word column of the matrix NAMALT. (Note: the alternatives are given numbers corresponding to the order in which they are read.) general 2+N+2 (where N is the number of lines of description, 1