Comments on Advance Notice of Proposed Rulemaking


The Federal Transit Administration should revise itscost-effectiveness rule as follows:

1. Require transit agencies to consider a full range ofalternatives.

Consideration of a full range of alternatives is vital to anhonest cost-effectiveness analysis. At the very least, alternativesto New Starts light- and heavy-rail projects should include avariety of bus-rapid transit lines and a system-wide reduction infares. Alternatives to commuter-rail projects should includelong-haul commuter bus services. Alternatives to streetcar andSmall Starts projects should include local circulator buses.Alternatives to any projects should also include no build,baseline, and significant alternatives submitted by members of thepublic.

2. Allow transit agencies to consider the cost-effectiveness oftransit projects in meeting any of several key non-monetarybenefits.

The FTA may continue to require that agencies calculate thecost-effectiveness of projects per hour of transportation systemuser benefit. But transit agencies should be allowed to considerother cost-effectiveness measures as well, such as thecost-effectiveness of reducing air pollution. However, only trueoutputs, not means to ends, should be considered. For example,reducing vehicle miles traveled is a means to an end, not anoutput, and so should not be rated as a cost-effectivenessmeasure.

3. Require transit agencies to accurately project the costs,monetary benefits, and non-monetary benefits of proposed transitprojects and their alternatives.

Cost estimates should use reference-class forecasting to accountfor historic forecasting errors for similar projects. Costs shouldalso look ahead at least 50 years to account for long-term capitalrenewal and replacement costs. Finally, the FTA should requiretransit agencies to do a "stress-test" to insure that the costs ofnew projects do not force agencies to reduce existing service inthe event of economic downturns.

4. Require transit agencies to rank alternatives according tothe net dollar cost per unit of each non-monetary benefit.

Net cost per: Alternative Hour of Transportation User Benefit Ton of Ozone Abated Ton of Greenhouse Gases Abated BTU of Energy Saved
A $3.50 $2.00 $11.00 $0.20
B 5.00 3.00 10.00 0.10
C 4.00 2.50 9.00 0.15
D 10.00 6.00 100.00 0.25
E 15.00 4.00 150.00 0.40

Calculating the cost per unit of benefit for severalalternatives and non-monetary benefits should produce a tablesomething like the one above. In the table, "net cost" is theannualized capital cost plus annual operating cost minus annualfares. In this table, no alternative is most cost effective by allnon-monetary benefits, but two alternatives - D and E - are clearlynot cost effective by any measure.

5. Require that alternatives that rank low be eliminated fromconsideration.

The FTA should require that alternatives that are clearly notcost effective - such as alternatives D and E in the above table -be rejected. This process gives transit agencies the flexibilitythey desire to consider a broader range of benefits while imposingrigor on the cost-effectiveness analysis to insure that federalresources are not spent on cost ineffective projects.


On June 3, 2010, the Federal Transit Administration (FTA) askedfor "public input on how to improve its calculation of 'costeffectiveness,' including whether FTA should measure quantifiablebenefits other than reduced travel time. In addition, FTA seekscomment on how it should evaluate environmental benefits andeconomic development effects."

Cost effectiveness is an essential criterion for allocatingresources at any time. Even during economic boom periods,government funds and resources are limited. Almost everyone losesif those funds are spent on costly projects that provide fewbenefits. The current recession combined with record federal debtmakes cost effectiveness in federal spending more important thanever.

Cost effectiveness is often confused with benefit-cost, but thetwo differ significantly. A benefit-cost analysis attempts tocalculate all benefits and costs in dollars (or some other commondenomination). When properly calculated, any project whose benefitsexceed the costs can be considered worthwhile. If resources arelimited, projects can be ranked by their benefit-cost ratios andthose with the highest ratios should be undertaken first, but suchranking is not absolutely essential so long as the benefits of eachselected project exceed the costs.

When it is not possible to express all of the potential benefitsof investments in dollar terms, however, a cost-effectivenessanalysis is called for. For such an analysis, benefits that cannotbe expressed in dollars must still be quantified using some othermeasure or measures such as hours of time saved, tons of abated airemissions, or accident fatalities avoided. Then the costs indollars are divided by the benefits to calculate the cost per hour,ton, fatality, or whatever is the benefit.

Note that a benefit-cost analysis produces a clear thresholdthat allows decision makers to reject projects: any project whosebenefit-cost ratio is less than one should not be funded. Acost-effectiveness analysis has no such clear threshold. Instead,for a cost-effectiveness analysis to work, it is vital that anyproposed project be compared with a full range of alternativeprojects that can potentially produce similar benefits. Projectsshould be selected only when their dollar cost per unit of benefitis clearly lower than that of any of the alternatives.

The FTA's existing cost-effectiveness process for New Starts isinadequate because it fails to recognize the need to consider afull range of alternatives. Instead, it sets an arbitrary thresholdof approximately $24.49 dollars per hour of "transportation systemuser benefit," (the exact dollar amount varies with inflation),above which projects are deemed to not be cost effective.

There are two major problems with this process. First, thisarbitrary threshold would potentially allow many cost-ineffectiveprojects to be approved for two reasons:

  1. Projects that cost $5 per hour of benefit were ranked equallywith projects that cost $24.49 per hour of benefit, when in factthe former were far more cost effective.
  2. Projects that cost less than $24.50 per hour were consideredcost effective even if alternative projects in that corridor ormetropolitan area could produce the same benefits at a far lowercost.

As noted in the advance notice of proposed rulemaking, some"members of the transit community" objected to the FTA'scost-effectiveness process for an additional reason: The arbitrarythreshold required the FTA to reject any projects that cost morethan about $24.50 per hour of user benefit even if they producedother benefits that were not measured in terms of "hours oftransportation system user benefits."

The Cato Institute proposes a new cost-effectiveness processthat will remedy all of the defects of the FTA's current process,including the transit community's complaint that other benefits arenot considered in the process. This process includes the followingsteps:

  1. Identification of a full range of alternative projects.
  2. Identification of key non-monetizable benefits of thosealternative projects.
  3. Estimates of the costs and monetary benefits of eachalternative project.
  4. Estimates of the non-monetary benefits of each alternativeproject.
  5. Ranking the alternative projects in terms of dollars of netcost per unit of each key non-monetary benefit.
  6. Eliminating low ranking projects from consideration.

1. Develop a Full Range of Alternatives

The goal of a cost-effectiveness analysis is to find theleast-cost way of attaining designated outputs or goals, andidentification of a full range of alternatives is a crucial step inany such analysis. Leaving out any alternatives that could be morecost effective than the proposed action effectively biases theanalysis towards the proposal.

For New Start projects, the FTA should specify that alternativesinclude, at the very least:

  • No build;
  • A baseline alternative as defined in the existing rule andincluding, at the very least, improvements in existing servicesthat the transit agency would be likely to make over the lifetimeof the proposed New Start project if that project is notbuilt;
  • A significant reduction in system-wide fares;
  • In the case of light- and heavy-rail projects, alternativesshould include a range of bus-rapid transit and/or express busesalternatives operating on existing roads, new general purposehighway lanes, exclusive busways, and "virtual" busways(high-occupancy toll or HOT lanes with priority for buses andvanpools and dynamically priced to keep lanes fully utilized butfree-flowing);
  • In the case of commuter-rail projects (defined as passengerservice that operates only, or mainly, during rush hours),alternatives should include long-haul commuter bus service onexisting roads and (if not now available) on new HOV or HOTlanes.

In addition to no-build and baseline alternatives, the FTAshould specify that alternatives to streetcar and other Small Startprojects include:

  • Increased frequencies of local bus service;
  • A local circulator bus or a network of circulator buses orother special bus routes, recognizable by distinctive liveries;and
  • Designation or construction of exclusive bus lanes or lanesdedicated to buses and selected other vehicles such as taxis, sharetaxis, and bicycles.

Other alternatives should be included depending on which keynon-monetary benefits are included in a cost-effectivenessanalysis. For example, if a transit agency counts abated greenhousegas emissions as a benefit, it should consider alternative fueledbuses, trolley buses, and other ways of achieving that benefit. Inaddition, transit agencies should give serious consideration toalternatives submitted by members of the public.

2. Identify Key Non-Monetizable Benefits

While hours of transportation system user benefit are animportant potential benefit of transit projects, transit agenciesshould be allowed to consider other benefits as well. However, anybenefit considered in a cost-effectiveness analysis must be a trueoutput, not merely a means to an end. For example, reducing airpollution is an output; persuading auto drivers to take transitinstead is merely a means to an end.

The FTA should continue to require transit agencies to calculatethe cost per hour of transportation system user benefits. Inaddition, agencies should be allowed (and in some cases the FTA maywant to require) to consider cost effectiveness in terms of costper:

  • Passenger mile of mobility within the metropolitan area;
  • Hour of reduced delay due to congestion;
  • Ton of abated air pollution;
  • Ton of abated greenhouse gas emissions;
  • BTU of energy saved;
  • Life saved from avoided accidents;
  • Other true outputs that cannot be measured in dollarterms.

Assessments of air pollution and greenhouse gas emissions shouldinclude the emissions from power plants providing electricity forelectrically powered transit. They should also consider the effectsof cold starts, i.e., that emissions of some pollutants, such asvolatile organic compounds, are greatest when motor vehicle enginesare cold, so a short trip can produce almost as much pollution as along trip.

Some measures that should not be considered because they aremeans to an end, not outputs themselves, would include:

  • Changes in local population densities or other measures ofurban form;
  • Vehicle miles traveled;
  • Vehicle trips;
  • Vehicle ownership rates.

Some transit agencies want to consider the economic developmentresulting from a transit project as one of the non-monetarybenefits. The FTA should discourage this, however, for at least tworeasons.

  1. The effects of new transit projects on economic development areoften highly exaggerated. Most, if not all, of the transit-orienteddevelopments that have been built along light-rail and streetcarlines in Portland, Oregon, for example, received millions ofdollars in tax-increment financed (TIF) subsidies. These subsidies,not the rail transit, are what led to the developments.
  2. Local economic development is often a zero-sum game. As RobertCervero and Samuel Seskin concluded in "An Evaluation of theRelationships Between Transit and Urban Form" (TCRP ResearchResults Digest, June, 2005), "Urban rail transit investments rarely'create' new growth, but more typically redistribute growth thatwould have taken place without the investment."

At the very least, transit agencies that wish to considereconomic development as a benefit of transit projects shouldinclude TIF and other subsidies on the cost side of the project andshould estimate the effects of those subsidies on development withboth the proposed transit project and alternatives to it. Toaccount for possible zero-sum games, they should also consider theeffects on economic development on a metro-wide basis.

3. Estimate the Costs and Monetary Benefits

The direct costs of New Starts and Small Starts transit projectsinclude the forecast annualized capital and annual operating costsof the entire transit system. The FTA should continue to requirethat capital costs be amortized over their useful lifespan at a 7percent discount rate. While this seems straightforward, mosttransit agency analyses fail to disclose three important costs:overruns; capital renewal and replacement of worn-outinfrastructure; and the threat to existing transit services whennew projects have financial problems.

Cost overruns: In 1989, Department ofTransportation researcher Don Pickrell found that then-recent railtransit projects had gone over their original projected costs by anaverage of 46 percent. He noted that, "The systematic tendency tooverestimate ridership and to underestimate capital and operatingcosts introduces a distinct bias toward the selection ofcapital-intensive transit improvements." Since then, transitagencies have somewhat improved their forecasts of futureridership, but recent rail transit projects are still costing anaverage of 40 to 50 percent more than the projections made at thetime of the original cost-effectiveness/alternatives analysis.

Danish planner Bent Flyvbjerg argues that planners of publicworks should use "reference-class forecasting" to account for suchoverruns. For example, if similar projects have gone 50 percentover projections made at the time of the original alternativesanalysis, then planners doing an alternatives analysis should add50 percent to the projected cost of a proposed new project. The FTAshould require this as a part of its cost-effectiveness analysis.The FTA should also require transit agencies that have not donereference-class forecasting to redo their cost-effectivenessanalysis under the new rules if their projected costs escalate bymore than 10 percent since their original analysis.

Capital renewal and replacement: America'slegacy rail transit systems - systems more than 30 to 40 years old- today suffer from deferred maintenance. A recent FTA reportrevealed that the nation's seven largest transit systems requiredabout $50 billion to bring them to a state of good repair, some 90percent of which was due to deferred maintenance on rail lines.Moreover, most of those systems were not even spending enough moneyto maintain their existing state of poor repair, much less returnthen to a state of good repair.

Despite this, most of the dozens of cities that are building orplanning new rail lines have no financial plans for renewing thoselines when they start wearing out in 25 to 30 years. They arerelying heavily on federal funding for construction, and in somecases have promised voters that they will end local taxes neededfor local matching funds when construction has been paid for. Thismeans they will have little or no money for capital renewal andreplacement. The FTA should require agency cost-effectivenessanalyses to include capital renewal costs, including replacement ofvehicles and other worn-out infrastructure for at least 50 yearsafter the project opens.

Threat to existing transit services: Monetarybenefits include fares and transit advertising. Since fares are, ineffect, a measure of transit ridership, transit agencies need notconsider transit ridership as one of the key non-monetary benefitsof transit projects. However, one serious problem with a largenumber of capital-intensive transit projects is that they haveforced transit agencies to cannibalize their bus systems to pay forthe fixed-guideway systems.

At various stages in the lifespan of rail transit systems, costoverruns, revenue shortfalls, and long-term maintenance costs haveled at least half the nation's metropolitan areas that have bothrail and bus transit to severely cut bus services. The FTA requirestransit agencies to show that building a fixed-guideway system willnot take funding away from the existing transit system, but thisrequirement should be made more rigorous. For example, the FTAshould require a stress test asking what would happen to transitagency operations if, at any time when the agency is building orrepaying bonds sold to finance fixed-guideway projects, the taxrevenues needed to support the agency decline for several years dueto a recession.

4. Estimate the Non-Monetary Benefits

Transit agencies should quantify the annual outputs (or savings)of each of the key non-monetary benefits under each of thealternatives. The FTA should continue to require that estimates ofbenefits be made for the entire urban area served by the transitagency, not just the corridor to be served by the proposed project.This will ensure that the benefits in the corridor, such as transitridership gains or economic development, are not offset byzero-sum-game losses (or slower rates of growth) elsewhere in theurban area.

In estimating benefits, transit agencies should be required touse the current travel demand forecasting model of the localmetropolitan planning organization (MPO) for all estimates ofridership, revenue and costs that relate to ridership. The FTAshould discourage modification of the MPO model for transit studiesunless those changes (including modification of land use densities,socioeconomic assumptions and demographic forecasts) are clearlydocumented and reviewed in the public record and certified by theMPO. Under no circumstances should any transit agency develop itsown travel demand-forecasting model for estimating patronage forany proposed new transit project or line.

5. Rank the Alternative Projects

Compiling the data would result in a table resembling the onebelow. "Net cost" is the annualized capital and capital renewalcost plus annual operating cost minus fares. In this table, none ofthe alternatives ranks best for all of the key non-monetarybenefits, but alternative C ranks either first or second for all ofthe criteria. That would probably be the most cost-effectivealternative, though a case could be made for alternatives A or B.Alternatives D and E are clearly not cost effective.

Net cost per: Alternative Hour of Transportation User Benefit Ton of Ozone Abated Ton of Greenhouse Gases Abated BTU of Energy Saved
A $3.50 $2.00 $11.00 $0.20
B 5.00 3.00 10.00 0.10
C 4.00 2.50 9.00 0.15
D 10.00 6.00 100.00 0.25
E 15.00 4.00 150.00 0.40

In practice, it is more than likely that the ranking will beroughly similar for all of the non-monetary criteria. This isbecause projects that have a significant effect on such outputs ascongestion delay should also have parallel effects on airemissions, energy consumption, and other environmentalmeasures.

6. Eliminate Low-Ranking Projects fromConsideration

A project should be considered cost effective only if it ranksat or near the top of the list (in terms of least net cost per unitof benefit) for all of the key non-monetary benefits. Projects thatare not cost-effective, such as alternatives D and E in the abovetable, should not be considered for federal funding.

The current cost-effectiveness rule applies a nationally setarbitrary number to rule out federal funding for local transitprojects. The Cato Institute's proposed cost-effectiveness processcompares local transit proposals with alternatives in the samecorridor or metropolitan area, and only rules out projects that areless cost effective than local alternatives.

At the national level, the FTA should still classify projects as"high" through "low" depending on whether they cost, say, less than$12 through more than $30.50 per hour of transportation system userbenefit. But no project would be completely ruled out for simplybeing "low" as long as it is still the most cost-effective projectin its metropolitan area (though any transit agency unable to finda project costing less than $30.50 per hour of user benefitprobably isn't trying hard enough).

Local variations in calculation procedures will probably resultin a significant range of results for projects that are, in fact,similarly cost effective. Thus, a project in one metropolitan areathat is ranked "high" should not have clear priority over a projectin another metropolitan area that is ranked "medium-high." However,calculation procedures should be consistent enough from onemetropolitan area to another than a project that is ranked "high"should have clear priority over one that is ranked, say, "medium"or lower.


In sum, the Cato Institute proposes that the FTA adopt acost-effectiveness rule that requires transit agencies submittingNew Starts and Small Starts proposals to:

  1. Consider a full range of alternatives, including, asappropriate, various forms of bus-rapid transit and otherimprovements in bus service;
  2. Calculate the costs of alternatives including long-termmaintenance costs and the use of reference-class forecasting toestimate cost overruns;
  3. Estimate the cost per hour of transportation system userbenefit as well as the cost per unit of other key non-monetarybenefits that are true outputs, not means to an end;
  4. Continue to base calculations on urban-area effects, notcorridor effects, so as to eliminate the possibility ofzero-sum-games.

Taken together, these proposals should significantly improve theeffectiveness of federal transit grants while they give transitagencies the flexibility they need to solve local transportationproblems.