Making Effective Fixed-Guideway Transit Investments: Indicators of Success (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

V O L U M E 2 Research Report

C O N T E N T S 2-1 Chapter 1 Introduction 2-2 1.1 Indicator-Based Methods 2-5 1.2 Historical Best Practices: Pushkarev and Zupan 2-6 1.3 FTA Investment Criteria 2-8 Chapter 2 Literature and Data Review 2-8 2.1 Previous Studies 2-9 2.2 Data 2-12 Chapter 3 Focus Groups: Phase I 2-12 3.1 Participants 2-13 3.2 Results 2-16 3.3 Conclusions 2-17 Chapter 4 Conceptual Framework 2-17 4.1 Defining Transit Project Success 2-18 4.2 Levels of Analysis 2-19 4.3 Identifying Indicators of Transit Project Success 2-21 4.4 Observation Set 2-28 Chapter 5 Quantitative Analysis Methods and Findings 2-30 5.1 Project-Level Models 2-33 5.2 System-Level Models 2-37 5.3 Estimating Uncertainty in Model Outputs 2-37 5.4 Input from Focus Groups: Phase 2 2-39 5.5 Response to Practitioner Input 2-42 5.6 Summary of Results and Comparison with Previous Studies 2-43 Chapter 6 Case Studies: Overview 2-44 6.1 Settings 2-44 6.2 Project Attributes 2-45 6.3 Indicator-Based Planning Methods 2-46 6.4 Potential Usefulness of the TCRP Project H-42 Method 2-48 6.5 Synopses 2-50 Chapter 7 Spreadsheet Tool: Technical Notes 2-51 Chapter 8 Conclusion 2-51 8.1 Implementation 2-52 References

2-1 C H A P T E R 1 This research report describes a method for estimating the likely success of proposed fixed-guideway rail projects. The method is more complex than typical indicator-based techniques used by transit agencies, but simpler than four- step forecasting models, FTA analysis requirements, or other advanced evaluation methods. The success metrics are based on predictions of project-level ridership, predicted changes in system-level transit usage, and estimated capital costs. The metrics can help decision-makers gauge the potential success of investments, based on the unique characteristics of the corridor and station areas to be served and the metropolitan areas in which they are located. The research was conducted for TCRP Project H-42, “An Exploration of Fixed Guideway Transit Criteria Revisited.” The objective of the project was to identify conditions and characteristics necessary to support fixed-guideway transit system projects and provide guidance on evaluating proposed projects. The researchers’ task was to provide an analytical framework and a set of tools in the form of a handbook and spreadsheet tool that decision-makers at all levels could use to determine whether conditions are present to support the success of their proposed investment. As part of the project, the research team: • Reviewed relevant literature and data sources to identify ways that transit system success is measured. • Conducted two rounds of focus groups and interviews with transit professionals in the private sector, public sec- tor, and academia. • Prepared a preliminary list of transit project success mea- sures and possible indicators or predictors of that success. • Assembled a geographic data set of fixed-guideway transit stations and networks in the United States, covering 3,263 transit stations in 44 metropolitan areas. • Collected data at the station, project, and metropolitan area levels, including system and station ridership, agency operating costs, project capital costs, regional and local demographics, employment, gross domestic product (GDP), gas prices, parking availability and pricing, regulatory restric- tiveness in land uses, rail and highway networks, and transit service characteristics. • Conducted regression analyses to identify station-area, corridor, and metropolitan-area factors that are the most significant predictors of project-level ridership and system- level passenger-miles traveled (PMT). • Developed a spreadsheet tool based on the regression anal- ysis results. • Carried out case studies of six transit projects in differ- ent parts of the United States, reviewing public reports and other materials, conducting site visits, and interview- ing transit planners, metropolitan planning organization (MPO) officials, and consultants who worked on the projects. Given the lengthy, costly, and uncertain process that cur- rently guides the process of evaluating, prioritizing, and fund- ing transit projects, the research team sought to simplify and inform a preliminary evaluation of projects by identifying a series of indicators of project success that could be applied without requiring extensive analysis. The spreadsheet tool, based on such indicators, enables local agencies to identify projects most worthy of further development and support. The research report summarizes and discusses the litera- ture review, data sources, focus groups, and interviews with transportation professionals conducted for the project. The data collection process is outlined, as is the rationale behind the observations and variables included in the final data set for analysis; the analysis approach is described, and the models are presented that best predict project ridership and system-wide transit usage based on information about the region, the corridor, and proposed project characteristics. Case studies are included that were conducted on selected U.S. fixed-guideway projects that provide qualitative success measures and indicators to supplement the findings from the Introduction

2-2 quantitative analysis. The cases offered a reminder that other factors often weigh in project decisions, and that heuristic indicators will continue to play an important role. The case studies also helped the researchers revise the analysis and improve the usability of the spreadsheet tool. The spreadsheet tool is available for download from the TCRP Report 167 web page, which can be accessed at www. trb.org by searching “TCRP Report 167”. The accompany- ing handbook, presented as Volume 1 in TCRP Report 167, provides an overview of the project, summarizes the use of indicator-based methods, and offers guidance on the spread- sheet tool and how to use it. 1.1 Indicator-Based Methods The characteristics of fixed-guideway transit projects and their surrounding corridors may serve as predictors, or indica- tors, of project success, as defined in various ways. Indicator- based methods provide an analysis approach for predicting success that is simpler than common four-step transportation models using zonal data. Although they are not a substitute for those methods, indicator-based methods can be used to conduct an initial evaluation of corridor alternatives. The indicator-based method developed by this study is rela- tively sophisticated, based on empirical research, and focused on project ridership and system usage rather than other suc- cess measures. Development of the method involved extensive original analysis of data about existing fixed-guideway transit projects in the United States. Local agencies might use this method to: • Assess whether it is worthwhile to expend funds on detailed project planning studies, • Compare various corridors within a region to see which offers the greatest potential for a transit project, or • Test various project and land use scenarios within a partic- ular corridor to identify those that deserve more detailed study. This report focuses on ridership, system patronage, and cap- ital costs. The research does not deal directly with operating costs, user and non-user benefits, or hard-to-quantify impacts such as network effects, social equity, and environmental improvements, any of which could justify some projects that would not otherwise make the grade. Further, the research- ers designed a mode-agnostic model, meaning that the model does not assume ridership bonuses for more desirable modes. This approach has both strengths and limitations, which are discussed at more length in Chapter 4. Indicator-based methods for assessing transit opportuni- ties have been used in practice for many years. For example, in 1976, New York’s Regional Plan Association suggested transit mode suitability criteria based on size of the downtown and residential density (Table 1.1). The Regional Plan Association’s recommendations were followed by Pushkarev and Zupan’s research, which led to the book Urban Rail in America. (Pushkarev and Zupan 1982) Today, transit planners rely on guidelines such as these when they develop system plans and identify potential new tran- sit routes, often using a variety of indicators to make relative comparisons between corridors within their regions to iden- tify those with the greatest transit potential. Many of the indi- cators used are similar to those already noted—the amount of population and employment in a corridor, population and employment density—but they may also include other fac- tors, such as the presence of transit-supportive policies, the level of highway congestion, the availability of right-of-way, and public support. For example, A Toolbox for Alleviating Traffic Congestion (Institute for Transportation Engineers 1989) offers general guidelines as follows: Light rail transit is most suitable for service to non-residential concentrations of 35 to 50 million square feet. If rights-of-way can be obtained at grade, thereby lowering capital costs, this threshold can be lowered to the 20 million square foot range. Average residential densities of about 9 dwelling units per acre over the line’s catchment area are most suitable. For longer travel distances where higher speeds are needed, rapid transit is most suitable for non-residential concentrations beyond 50 million square feet and in corridors averaging 12 dwelling units per acre or more. Commuter rail service, with its high speed, relatively infrequent service (based on a printed schedule rather than regular head- ways) and greater station spacing is suitable for low density residential areas—1 to 2 dwelling units per acre. However, the volumes required are only likely in corridors leading to non- residential concentrations of 100 million square feet or more, found only in the nation’s largest cities. The San Francisco Bay Area Metropolitan Transportation Commission (MTC) has adopted a set of housing density Transit Vehicle Mode Minimum Downtown Size, Square Feet of Contiguous Non- Residential Floor Space (millions) Minimum Residential Density, Dwelling Units per Acre Local bus 2.5 4 to 15a Express bus 7 3 to 15a Light rail 21 9 Heavy rail 50 12 Commuter rail 70 1 to 2a a Varies with type of access and frequency of service. Source: Regional Plan Association, Where Transit Works: Urban Densities for Public Transportation. New York, 1976. Table 1.1. Transit investment suitability criteria according to regional plan association, 1976.

2-3 thresholds by transit mode that projects are expected to meet before the MTC programs funds (Table 1.2). According to the MTC’s Resolution 3434, “Each proposed physical transit extension project seeking funding through Resolution 3434 must demonstrate that the thresholds for the corridor are met through existing development and adopted station-area plans that commit local jurisdictions to a level of housing that meets the threshold” (MTC 2006). The Utah Transit Authority calculates a Transit Prepared- ness Index to identify those parts of its service area that have the characteristics to support a successful transit investment (Utah Transit Authority 2005). The index (see Figure 1.1) relies on five criteria to identify the best places in the region for improving transit service: 1. Transit-oriented development (TOD) or mixed-use zoning (up to 40 points); 2. TOD or mixed-use development in general plan (up to 10 points); 3. Bicycle/pedestrian plan (up to 10 points); 4. Amenity proximity score based on walkscore.com (up to 10 points); and 5. Intersection density based on walkscore.com (up to 30 points). Consulting firms have developed proprietary indicator- based tools. One such tool is the Transit Competitiveness Index (TCI) (see Figure 1.2). This tool offers a way to score different travel markets in terms of how well transit is likely to compete with the automobile. The TCI accounts for vari- ous transportation and land use characteristics—trip vol- umes, land use density, parking cost, and congestion—along with trip purpose and household characteristics to produce a numeric score. Depending on the score, individual markets are characterized as strongly competitive, marginally competitive, marginally uncompetitive, or uncompetitive. Further informa- tion about the TCI is available at: http://www.mtc.ca.gov/planning/tsp/TCI-DRAFT- PRIMER.pdf The Portland Metro (the MPO for the Portland, Oregon region) applied an interactive web-based build-a-system tool as part of the public involvement process for its High Capac- ity Transit System Plan (see Figure 1.3). According to Metro’s plan, BART HRT LRT BRT CR Ferry Housing threshold (Average housing units per station area) 3,850 3,300 2,750 2,200 750 HRT = heavy rail transit; LRT = light rail transit; BRT = bus rapid transit; CR = commuter rail Source: San Francisco Bay Area MTC Resolution 3434, Attachment D-2, as revised July 27, 2005. Table 1.2. MTC housing density thresholds. Source: Utah Transit Authority, 2005 Figure 1.1. Transit preparedness index for Weber County.

2-4 Source: MTC and Cambridge Systematics, Inc. Figure 1.2. Sample use of the TCI. Source: Portland Metro Figure 1.3. Build-a-system tool, Portland, Oregon.

2-5 along a transit corridor of 100 to 150 square miles required a minimum threshold of 12 dwelling units per residential acre along the corridor and 50 million square feet of non-residential space in the downtown. Light rail transit (LRT) operating at 5-minute peak-hour headways along a 25 to 100 square mile corridor required only nine dwelling units per residential acre and 20 to 50 million square feet of non-residential space. Thresholds to support bus service varied, based on service fre- quency, from four to 15 dwelling units per residential acre and 10 to 35 million square feet of non-residential space. Commuter rail transit operating 20 trains per day was supported by just one or two dwelling units per residential acre on corridors to the largest downtown in the region. Urban Rail in America continued the previous research and focused on rail, using data from 24 CBDs in the Tri-State area of New York, New Jersey, and Connecticut. The authors cali- brated a decay function demand model to estimate trips from a residential area to the downtown as a function of down- town non-residential floor space, residential area population distribution, and the distance between the two. CBD non- residential attractors that were farther away from the resi- dential population exerted less influence and generated fewer trips to the downtown than those that were located closer. The relative importance of non-residential development outside of the CBD also influenced travel demand; areas with fewer competing attractions outside of the downtown attracted more trips to the CBD. The authors incorporated total esti- mated demand for downtown trips into a mode share model to determine the attractiveness of fixed-guideway transit ver- sus the private automobile. Dramatic changes have occurred in metropolitan area devel- opment patterns, the work force, economic conditions, and [The] tool allowed community members to explore trade-offs between corridors and build their own high capacity transit sys- tem. With the build-a-system tool, community members learned about centers that could be served by high capacity transit and to compare corridors based on ridership, travel time, operations cost, capital cost, and environmental benefits (Portland Metro 2009). Metro’s interactive tool is more fully described at http://www.oregonmetro.gov/index.cfm/go/by.web/ id=26680. 1.2 Historical Best Practices: Pushkarev and Zupan The research for TCRP Project H-42 has many precedents, including Public Transportation and Land Use Policy (Pushk arev and Zupan 1977) and Urban Rail in America (Pushkarev, Zupan, and Cumella 1982). These studies inspired rules of thumb regarding the feasibility of different levels of transit investment in given corridors. The 1977 study used non- residential central business district (CBD) floor space, dwelling units per residential acre near stations, and distance to the CBD to estimate transportation demand across a variety of transit modes, ranging from dial-a-ride taxis to heavy rail transit (HRT). Per-passenger operating costs were calculated for each of the modes, and on that basis the potential service frequency (based on demand) was estimated in terms of average daily trip origins produced per square mile. The authors suggested minimum threshold residential densities and downtown sizes that would be required to support various levels of cost- effective service across different modes (Table 1.3). For exam- ple, heavy rail rapid transit at 5-minute peak-hour headways Mode: Service Minimum Dwelling Units per Residential Acre Size of Downtown Local bus: minimum (20 buses per day) 4 10 million square feet non-residential Local bus: frequent (120 buses per day) 15 35 million square feet non-residential LRT: (5-minute peak-hour headways) 9 (corridor of 25 to 100 square miles) 20 to 50 million square feet non-residential HRT (rapid): (5-minute peak-hour headways) 12 (corridor of 100 to 150 square miles) 50+ million square feet non-residential Commuter rail: (20 trains per day) 1 to 2 Largest in region Table 1.3. Transit-supportive residential density and employment thresholds (adapted from Pushkarev and Zupan 1977).

2-6 (49 U.S.C. § 5309). This discretionary grant program provides capital assistance for new fixed-guideway systems (New Starts and Small Starts), corridor-based BRT projects (Small Starts), and capacity expansions on existing fixed-guideways (Core Capacity). In recent years, the program has been funded at close to $2 billion per year. As local transit project sponsors typically rely on substantial funding from FTA, many state and regional transit funding policies mirror those of FTA (Deakin et al. 2002). Thus, it is important to understand the research being conducted under TCRP Project H-42 in light of FTA’s Section 5309 programs. The Section 5309 New Starts and Small Starts Program is currently the federal government’s primary method of fund- ing new fixed-guideway transit investments for both the con- struction of new fixed-guideway systems and extensions to existing fixed-guideway systems. Commuter rail, HRT, LRT, BRT, monorails, automated people movers, and streetcar projects are eligible to apply for financial assistance. New Starts and Small Starts funding is allocated to major cap- ital projects on a discretionary basis—one of the few instances when U.S. DOT funding is not distributed by formula. To help manage the competition for funds, FTA evaluates New Starts and Small Starts projects within a multicriteria analysis framework. The framework provides a structured approach for developing a project rating, based on a set of criteria and a series of weights. Under MAP-21, FTA’s rating system uses six project jus- tification criteria, three local financial commitment criteria, and a five-point rating scale (high, medium-high, medium, medium-low, and low) for each of the criteria (Table 1.4). FTA guidance offers more specifics on the measures, recom- mends the weights for each criterion, and specifies the break- points that FTA will use when applying the five-point ratings scale. Projects must receive at least a medium rating on both justification and financial commitment to move to the next phase and ultimately be considered for a grant. Some of the rating criteria are assessed qualitatively while others (e.g., mobility improvements, cost effectiveness, envi- ronmental benefits) are quantified using transportation plan- ning models. Models are used, for example, to estimate the number of daily riders expected to use a project, and the change in vehicle-miles traveled (VMT). Project ridership is combined with capital and operating/maintenance cost estimates to assess gasoline prices since these seminal studies were carried out. There has also been a renewed investment in fixed-guideway transit projects. APTA reports that there are now 27 CR rail, 15 HRT (subway) and 27 LRT systems in the United States. In addition, BRT has been adopted in several municipalities as a new alternative to traditional transit modes, allowing com- munities historically priced out of rail technology to develop cost-effective transit networks. As of 2013, APTA counts five BRT systems in the United States. Since 1980, ridership on commuter, heavy, and light rail has grown from 2.52 billion to 4.47 billion trips per year, and passenger-miles have grown from 17.5 billion to 29.5 billion (APTA 2012). Data and analysis tools also have changed. Developments in research methods, more readily available land use and transportation data, and ubiquitous computing and geo- graphical information system (GIS) technology have pushed both the state of the art and the state of the practice of indi- cator development and measurement. It is now possible to incorporate or create more explicit and accurate measures such as gravity-model-based accessibility indexes for tran- sit and auto; more-refined measurements of population and employment characteristics within specific walking-distance buffers around transit stations; congestion data; and parking prices. This research effort has benefited from these advances. The direction the research team took in TCRP Project H-42 is in some ways quite different from previous indi- cator-based approaches. With more data and analysis tools have come the ability to make better estimates and more information about the uncertainty of predictions. More data and tools also made it possible to characterize ridership and cost along a continuum rather than providing thresholds at which particular technologies can be used, and to consider the effects of projects on patronage at the system level. This flexibility is useful and warranted. Ridership can vary greatly; system impacts can be distinct from project-level measures; and the capital cost and operating costs of different technolo- gies implemented in different settings can vary greatly. 1.3 FTA Investment Criteria The single largest source of funds for transit improvements in the United States is the FTA Capital Investment Program Project Justification Criteria Local Financial Commitment Criteria Mobility improvements Cost effectiveness Environmental benefits Congestion relief Land use Economic development Current financial condition Commitment of funds Reliability/capacity of the financial plan Table 1.4. FTA MAP-21 project criteria.

2-7 There has been interest in applying streamlined evalu- ation measures (or warrants) to FTA’s project evaluation process, and MAP-21 specifically encourages the use of warrants. TCRP Project H-42 sought a simplified way for local agen- cies to evaluate proposed fixed-guideway transit alternatives based on relatively simple indicators of project success or merit that might be applied, without requiring extensive analysis, at least as a screening phase. With the procedure presented in TCRP Report 167, local agencies can quickly develop an initial estimate of the ridership likely on a project before spending scarce resources on more detailed studies. cost effectiveness. FTA reviews the models, the model inputs, and the model outputs. To reduce the burden of these reviews, in September 2013 FTA released a simplified travel forecasting model for predicting transit ridership on fixed-guideway proj- ects. The FTA model is the Simplified Trips-On-Project Soft- ware (STOPS). A project’s land use and economic development ratings are based on FTA’s qualitative assessment of the current land use conditions, plans and policies for future land use, and affordable housing considerations. Financial ratings are based on projections of project costs and revenues, along with some- what qualitative assessments of the reliability and completeness of underlying forecasts.

2-8 C H A P T E R 2 TCRP Project H-42 began with a review of previous stud- ies and a search for data sources to (1) ascertain relevant mea- sures of success from varying definitions of what constitutes successful transit systems, and (2) identify and describe what is currently known about indicators of success, or character- istics that determine whether a transit project will likely be successful. 2.1 Previous Studies Recent work continues to confirm the importance of pop- ulation density (Taylor et al. 2009) and employment density (Barnes 2005) as predictors of transit ridership. Income mea- sures (Taylor et al. 2009), measures of network configuration (Thompson and Brown 2006, 2010, 2012; Thompson et al. 2012), service frequency (Evans 2004), bus line connections (Kuby et al. 2004) and park-and-ride spaces (Kuby et al. 2004) are additional indicators found by other studies. Guerra and Cervero (2011) studied more than 50 nationwide HRT, LRT, and BRT projects, and determined that jobs and population in the service area, number of park-and-ride spots, frequency of service, and GDP were correlated with transit ridership. Additional often-cited predictors of transit use include population characteristics such as education level, immigrant status, renter status, and car ownership (Taylor et al. 2009; Chatman and Klein 2009; Kuby et al. 2004); service character- istics such as fare (Guerra and Cervero 2011; McCollom and Pratt 2004; Kohn 1999), revenue vehicle-miles (Kohn 1999) and speed (Guerra and Cervero 2011); average station dis- tance to the CBD (Guerra and Cervero 2011; Kuby et al. 2004); transit network service coverage (Thompson and Brown 2006, 2010, 2012; Thompson et al. 2012); weather (Kuby et al. 2004); and fuel price (Guerra and Cervero 2011). Researchers have also investigated indicators such as trip destination type (Barnes 2005) and centrality, which measures relative accessi- bility of each station to all other stations determined by aver- age travel times (Kuby et al. 2004). Some studies differentiate the significant indicators of transit usage by mode, and recent work has found that the strength and nature of influential fac- tors vary by transit type (Thompson and Brown 2012). One element that might predict automobile use but that is often excluded from these studies is the cost of private auto use as measured by congestion indexes and parking prices. Indicator-based analysis represents only one of many pos- sible approaches to examining the likely demand for transit. Other aggregate demand methodologies might generate com- parative statistics across different transit systems, though their conclusions are sometimes based on heuristic rules instead of predictive models. For example, recent reports for the Brook- ings Institution have emphasized the importance of job acces- sibility via transit as a sign of a successful transit system and formed conclusions about the lack of such transit accessibil- ity across 371 transit providers in 100 of the nation’s largest metropolitan areas (Tomer et al. 2011; Tomer 2012). Other travel-demand estimation methods use disaggregate data on households or individuals. They might include a tra- ditional four-step model that employs a sequential framework based on four choice dimensions, or disaggregate models that use survey data to explain individual-level behavior around discrete choices (Small and Verhoef 2007). Few travel-demand studies have investigated the importance of multimodal inter- actions, impacts over time, the relative costs of transit versus auto, or parking availability. Successfully carrying out an analysis of these potential indicators is a significant challenge because data are difficult to assemble. There are also relatively few systems to study and compare in order to produce robust and reliable statistical results. In summary, previous studies of transit lines and systems have concluded that multiple indicators determine transit project success, including population concentration near tran- sit stations, the relative cost of automobile travel, and transit service characteristics such as fares, speed, access, and fre- quency. They also found that household income and minor- ity status are correlated with ridership within cities, although Literature and Data Review

2-9 some researchers contest these two indicators. The presumed impact of household income might be influenced by the fact that fixed-guideway transit systems across the United States tend to serve commuters with higher average incomes, and minority status might merely be a proxy for other, unmeasured elements, such as transit dependence or captivity. A summary of the measures and indicators of success that were considered for inclusion in the analysis is included in Table 2.1. 2.2 Data Before beginning the data collection and subsequent analy- sis, the research team catalogued and described existing data on transit capital, operating, maintenance, disposal, and life cycle costs; transit agency and private household expenditures; transit networks; car parking; employment and population densities; levels of mixed-use development; and transit- and auto-based accessibility measures (Table 2.2). For many of these measures there is considerable variation in the unit of analysis (e.g., state, transit agency, metropolitan area, census tract and block, or household), the survey period (e.g., short-, mid-, or long-term) and the update frequency (e.g., decennial, annual, quarterly, or monthly). It is challenging to find data on network attributes, intermodal facilities, parking costs and availability, BRT systems, and urban design characteristics. For a more complete summary of the data reviewed, see Appendix A. 2.2.1 Catchment Area Analysis The research team compiled regional demographic infor- mation for the nationwide analysis of metropolitan areas from the 2000 U.S. Census and 1-year American Commu- nity Survey (ACS) for each year from 2005 through 2009. ACS data was collected by metropolitan area and census data by county, which was then aggregated up to the metropoli- tan area through either a summation or weighted average. Included in the information gathered were characteristics of the population (race, median age), households (occupancy, tenure, median rent and value), the economy (median house- hold income, per capita income, percentage of population below poverty line), and the workforce (workers per person, commute mode to work, vehicles per household). The census and ACS regional economic information was supplemented with metropolitan area economic data from the U.S. Bureau of Labor Statistics (BLS) and the U.S. Bureau of Economic Analysis (BEA), including job counts, unemployment figures, personal income levels, and GDP from 2000 through 2009. The researchers also incorporated metropolitan area demo- graphic information through an analysis of the characteristics of catchment areas around each station. Census 2000 block and block-group data was spatially applied to the station catchment areas that cut around or through them, and the catchment area information was then aggregated up to the metropolitan level. At the block level, census data was collected on age, race Table 2.1. Summary of measures and predictors of success considered for analysis. Measures of Success Predictors of Success 1. Cost and Ridership Metrics (Primary) Cost (capital and operating) Average cost per passenger, per passenger- mile, per mile, per hour of time savings, per new transit trip. Operating cost recovery ratio Ridership Ridership totals, change in ridership, ridership per capita 2. Economic Cost-benefit Analysis (Primary) Net present value > 0 Marginal benefits > marginal costs (including external costs such as congestion and pollution) 3. Land Use Impacts (Secondary) Increased development and densification Higher land value & property tax revenues 4. Equity Measures (Secondary) Benefits and costs to disadvantaged individuals, populations, or regions 1. Transit Supply Costs/Revenues Capital costs Change in operating costs Service supply (frequency by time of day) Transit fares Network attributes (e.g., route alignment) 2. Transit Demand Socioeconomics of user Costs and travel time of alternatives (e.g., car, local bus) Land use and transit ridership (built-environment factors) Employment and population density Station-area characteristics (e.g., distance from CBD) 3. Development Potential Available land Strong real estate market Permissive regulations Targeted infrastructure expansion Tax incentives

2-10 Indicator Data Source Data Notes Employment density and diversity; size of job centers Census, county business patterns, Longitudinal Employer-Household Dynamics (LEHD), Bureau of Labor Statistics (BLS), Bureau of Economic Analysis (BEA) Workers by industry, at different spatial levels Higher density increases potential ridership and may serve as a proxy for higher road costs when concentrated in centers. Resident and regional characteristics and income U.S. Census, American Community Survey (ACS), Public Use Microdata Sample (PUMS) , BEA, BLS Consumer Expenditure Survey (CES) Household and per capita income, rents, ethnicity, age In general, lower-income neighborhoods will generate more riders. Expanding service in disadvantaged areas can also help improve social equity. Transit network attributes National Transportation Atlas Database (NTAD), GIS Link-to-node ratio, accessibility indexes Transit service characteristics Transit agencies, FTA, Google Earth Route-miles, stations, park-and-ride spaces, bus line connections, service frequency, speed, track grade, opening year Parking Colliers International, Parking In Motion (PIM) North America Central Business District parking rates, individual garage rates Less parking per capita and higher market-rate parking can prompt motorists to switch to transit. Congestion/travel speeds Texas A&M Transportation Institute (TTI), FHWA Average daily traffic per freeway lane, relative travel conditions in peak period versus free-flow Congested Corridors Report Fuel costs National Household Travel Survey (NHTS), GasBuddy Land use regulatory restrictiveness Pendall et al. Zoned densities, growth management Tools Neighborhood walkability Walkscore.com Walk score Weather conditions National Climatic Data Center (NCDC) Average temperature, precipitation Fares NTD, transit agencies Fare revenues/ Passenger-miles Fares are a control for investigating impacts, not a predictor as such. Table 2.2. Possible predictors of transit success considered. (including percent Hispanic), household occupancy, and tenure. At the block-group level, census data was collected on house- hold size, household income, automobile ownership, commute mode, and commute duration. In addition to the census spatial demographic information, block-level data was incorporated from Longitudinal Employer-Household Dynamics (LEHD) on job counts by employment location from 2002–2008, bro- ken down by service industry and income group. For the first step in the station catchment area demographic analysis, maps were created of the station catchment areas. Each station was assigned to its respective block/block group using the geographic areas defined by ESRI Census 2000 TIGER/Line Data. Around each station, straight-line-distance buffers of 0.25, 0.5, and 1 mile were created for urban rail sys- tems and 0.5, 1, and 3 miles were created for CR systems so that the researchers could test measures taken for different- sized areas around stations. Thiessen polygons were used to ensure that each station’s catchment area was mutually exclu- sive of neighboring station catchments. Census blocks were then clipped to each buffer to create shapefiles that contained all complete and partial census blocks within 0.25, 0.5, and 1 mile of urban HRT, LRT, or BRT stations and within 0.5, 1,

2-11 • Census blocks that are entirely composed of water should not have any demographic data (population, jobs, etc.) assigned to them, so they would not affect the analysis. Some demographic indicators were available only at the block-group level. Rather than re-creating catchment areas using block-group shapefiles, the researchers aggre- gated the census block catchment area shapefiles up to the block-group level. In these cases, the clipped area (within the catchment area) of each block within the block group was added up and then divided by the land area (as reported by the census) of the containing block group. Demographic information was then multiplied by this fraction in a similar fashion. The study team then assigned census data to a catchment area based on that land area ratio. If an entire block/block group was within the bounds of a catchment area, the land area fraction would be equal to 1 and the full census count for a given demographic variable would be allocated to that catchment area. If only a portion of a block/block group fell within a catchment area, the team applied the land area fraction and allocated only that percentage of the census count to the catchment area. A non-count census variable (e.g., median age) was assigned to a catchment area by tak- ing weighted averages based on the catchment’s population size. Finally, the study team aggregated the characteristics of each station catchment area up to the regional level for the nationwide analysis of metropolitan areas through either a summation or a calculated average (in some cases weighted by population or households). and 3 miles of CR stations (Figure 2.1). To construct the panel data set, the researchers repeated this process for each year from 2000 to 2009, because the opening of a new station in a given year at times forced the realignment of the catchment area around a nearby station. In the next step, the research team assigned demographic and employment data to each station catchment area. First, the fraction of land area of each block or block group falling within a given catchment area was calculated. Census blocks that were not clipped to coastlines were used in the catch- ment area analysis. Although it was initially thought that this might pose a problem with demographic data being incor- rectly assigned to water areas, the team decided its analysis would not be affected, for the following reasons: • The Census’ own description of how census block geogra- phies are created states that water areas within block groups are excluded from land areas and assigned a separate block number. Even water that is not along a coast or river (i.e., a pond located within a land census block) is taken out and assigned to the largest water block in the block group. Total area covered by the catchment areas remains the same. Figure 2.1. Graphical representation of GIS catchment area creation process.

2-12 C H A P T E R 3 The study team carried out two rounds of focus groups and interviews with transit professionals and academics, to inform initial decisions about the research and to provide feedback on our initial results. In the first round, the researchers met with professionals in the transit industry to discuss basic concepts for framing the research, such as how to define a successful transit project, what factors have been examined in the past to help identify potentially successful projects, what factors would be useful for future informal alternatives analyses of transit projects, and what readily available data are currently used to support the incorporation of certain predictors of transit success. Participants in the first phase of discussions agreed that defining success, and evaluating projects based on success standards, is a complex process. Some participants, particu- larly those working for transit agencies or consulting firms, believed that different projects should be evaluated using dif- ferent criteria. Interviewees provided a wide array of poten- tial measures of a project’s success, including ridership levels, improved regional efficiency and mobility, economic devel- opment, and the creation of a transit-friendly environment. A variety of predictors of success were also suggested, including quality of service, cost savings versus the private auto, corri- dor density, supportive local land use policies, a demonstrated commitment to transit in the region, and the integration of the new project into the existing system. The results from the second round of interviews and focus groups are discussed in Section 5.4. 3.1 Participants The first meeting was a 90-minute focus group of eight participants conducted during the APTA Rail Conference in Boston in June 2011. Participation was by invitation only, with participants from transit agencies, MPOs, consulting firms, and academic institutions from various-sized metro- politan areas representing a spectrum of transit technologies (HRT, LRT, CR, and BRT). Participants were chosen from among the conference attendees based on their knowledge of transit project evaluation and their thought-leadership posi- tions within the industry, with the goal of seeking a variety of perspectives. The focus group was followed by a series of telephone interviews with other participants to follow up on ideas presented during the focus group meeting and to help balance the range of participants. Focus group participants at the 2011 APTA Rail Confer- ence were: • Alan Lehto, director of project planning, TriMet, Portland, Oregon. TriMet is the public transportation agency of the Portland metropolitan area. • Tom Jenkins, principal consultant, InfraConsult, Los Angeles, California. InfraConsult is a firm specializing in the development and financing of infrastructure projects. • Jim Parsons, principal consultant, Parametrix, Seattle, Washington. Parametrix is a firm specializing in the engi- neering, planning, and environmental elements of infra- structure projects. • Liz Rao, vice president and public transit chair, HNTB, Denver, Colorado. HNTB is a firm specializing in the engi- neering, planning, construction, financing, and operations of infrastructure projects. • Mike Shiffer, vice president for planning, strategy & tech- nology, TransLink, Vancouver, British Columbia. TransLink is the regional transportation authority of metro politan Vancouver. • Kim Slaughter, senior vice president of service design & development, METRO, Houston, Texas. METRO is the metropolitan transit authority of Harris County, Houston, Texas. • Bill Woodford, president, AECOM Consult, Arlington, Virginia. AECOM Consult is a firm specializing in the management of and technical support for infrastructure projects. Focus Groups: Phase I

2-13 Subsequent telephone interviews were held in July 2011 with: • Scott Rutherford, professor, Department of Civil & Envi- ronmental Engineering at the University of Washington, Seattle, Washington. • Steve Polzin, transit research program director, Center for Urban Transportation Research at the University of South Florida, Tampa, Florida. The Center for Urban Transporta- tion Research provides technical support, policy analysis, and research support. • David Ory, principal transportation planner/analyst, Metro- politan Transportation Commission, Oakland, California. The Metropolitan Transportation Commission (MTC) is the transportation planning, coordinating and financing agency for the nine-county San Francisco Bay Area. • Nat Bottigheimer, assistant general manager of planning and joint development, Washington Metropolitan Area Transit Authority (WMATA), Washington, D.C. WMATA is the transit authority of the national capital area. 3.2 Results Both the focus group and the telephone interviewees were asked the following questions: • How would you define success for a fixed-guideway transit project? • How would you assess the success of the FTA New Starts program? • Other than the FTA New Starts criteria, what indicators of success have you used or seen used in the planning phase to identify potentially successful fixed-guideway transit proj- ects? What other indicators of success do you think should be used in planning-level evaluations of transit alternatives? • What tools and data are needed to calculate or utilize these indicators? • Do you have any other suggestions with regard to research direction? Sections 3.2.1 through 3.2.5 summarize the responses to these questions. 3.2.1 How would you define success for a fixed-guideway transit project? Consistent with the participants’ varying views on what constitutes a successful project, they offered a diverse and, in some cases, conflicting set of potential success measures. Some concrete assessment factors included ridership, service quality, and manageable costs. Others, less easily measured, included development around transit, improved regional efficiency and mobility, and the creation of a transit-friendly environment. 3.2.1.1 Ridership and Rider Benefits Several participants suggested potential success measures based on changes in ridership on the line, including abso- lute ridership figures or new transit trips recorded. According to one participant, success is “simply about riders,” because “all other goals (such as congestion reduction, air quality improvement, even land use impacts) are premised on provid- ing accessibility and, essentially, on people using the line.” The same participant emphasized PMT as the ultimate measure, “better than simply riders, because the element of distance is a valuable measure in terms of mobility accomplished.” One participant suggested that ridership figures are important to local decision-makers: “Politicians care about ridership.” Other participants emphasized the financial element of tran- sit system operations in the discussion of ridership, suggest- ing cost per passenger as the more important determinant of project success. Other participants felt that, in addition to increased ridership, an important success measure was an improved trip experience for those already using transit. One participant suggested that a successful project made a transit line “fast, reliable, frequent, and safe,” more generally provid- ing “better service for existing trips.” 3.2.1.2 Land Use Changes Participants also emphasized land use changes in response to a transit project. A few felt particularly strongly that it was important for a transit agency to “get a reasonable or expected return on its investment,” and that success was all about “what you get in return.” For one participant, return was largely expressed in terms of development around tran- sit, whether in the form of new development within a region or more efficient development patterns around stations. An interviewee noted that “it’s important to ask whether any development happened. . . . It all comes back to the critical question of land use.” Another warned to be “careful how benefits are quantified and how the study/impact areas are bounded. . . . One-hundred units built near a station do not mean that 100 households have been created; they have just been shifted from elsewhere.” More broadly, some participants discussed measuring suc- cess through general improvements in efficiency and mobility across the region. One interviewee emphasized that “the con- versation about transit success should be more about improv- ing efficiency—this is the best form of economic growth that transportation can provide. By improving efficiency you reduce labor costs by reducing the cost and difficulty of the commute. Improving efficiency is the primary goal.”

2-14 Another potential measure suggested by one participant was the creation of a transit-friendly environment, which could reflect a project’s success through “metrics that reveal an orientation toward transit activity and transit investment.” One such metric might be the presence of “a high bike/ped mode share of all trips,” which “provides a more comprehen- sive look—a 24-hour set of outcomes.” Other potential figures include the percentage of facilities that are usable by persons with disabilities (“as many ADA-oriented facilities as pos- sible”) or the percentage of children and senior citizen users, both reflecting adequate transit “accommodations for those who are not able to or shouldn’t drive.” 3.2.1.3 Cost-Benefit Analysis Many participants, especially those representing public transit agencies, saw the need for a quantifiable cost-benefit test to assess the success of a transit project. As one participant put it, “success is somehow quantifiable based on whether benefits exceed costs—essentially a cost-benefit test.” A focus group participant noted that when his agency evaluates mul- tiple project proposals, it also “assigns costs to alternatives and trades off the benefits with the costs.” The challenge arises from the fact that benefits include not only transportation but also social and environmental factors: “the cost-benefit cate- gories compare apples and oranges.” He added that the easiest way to define the “kind of value the project is creating” might be through “land value and density of development,” but he stressed that “there could be others.” 3.2.1.4 Higher-Resolution Goal Setting Finally, one theme that occurred in most discussions was that success should also be measured through the achieve- ment of project-specific local goals. As one interviewee put it, a project can be considered successful if it “solves the problem it was supposed to solve.” For some of these success measures, a project’s potential success might be apparent by looking at corridor character- istics, but others could require considerable project-specific forecasting and analysis. According to the interviewees, suc- cess indicators would need to approach success from multiple scales (regional, system, corridor, local, etc.) and from mul- tiple perspectives (FTA, regional, transit agency, etc.). 3.2.2 How would you assess the success of the FTA New Starts program? Federal law does not clearly establish goals for the New Starts program. Although the law authorizing the federal funding programs for transit directs FTA to rate projects in terms of cost effectiveness, it says little about how effective- ness is to be measured. Land use and economic development are also called out in the law as rating factors, with little direc- tion on federal goals and objectives. Participants in the focus group and interviews agreed that it is essential to define what makes a project successful before indicators for predicting success can be identified. One focus group participant emphasized the “need to define what a suc- cessful project is, then think about what the warrants are to achieve that.” However, participants also tended to agree that the projects funded through the program have had varying goals, and that they serve different markets using different technologies. Some projects are advanced by their local spon- sors with the intent of reducing transit travel time, but others are meant to provide accessibility or promote development. As one participant succinctly stated, “Different values lead to different decisions.” Without more specific objectives, the success of a project is often measured in terms of whether or not it was completed on time and on budget, or whether or not initial ridership projections were achieved. Suggesting that a one-size-fits-all approach does not match the diversity of local goals and project types, some partici- pants said that projects with different goals and characteristics should be categorized and evaluated under different criteria. One participant suggested that the evaluation process should “put a project into one of a number of categories” (a typology of projects). One set of success indicators could be used for projects meant to achieve time savings, and another set could be used for projects seeking to improve access or promote economic development. Other participants noted that federal goals for the New Starts program may differ from local goals for a project. Some participants suggested that there should be one set of indica- tors that local agencies might use to predict whether a project could be justified in a corridor, and a second set of indicators that FTA might use to determine whether a project is deserv- ing of federal funding. One participant emphasized that there should be room for a locality to say “Yes, it looks like it’s a reasonable investment,” while FTA employs a separate pro- cess for “deciding which projects receive the limited amount of federal funds available.” Some participants also suggested that the indicators might be used by FTA to assess the level of scrutiny a project should receive. If high ridership benefits are obvious according to an indicator-based method, FTA might give a project’s demand forecast little review. If ridership benefits are unlikely, however, FTA might use the indicators as the basis for immediately reject- ing a project for funding. FTA would then focus its oversight on those projects that are not immediately rejected but that are not obviously likely to achieve high ridership benefits. One partici- pant recalled this policy in practice, where the “level of scrutiny in ridership forecasts on the busy lines was less than the rigor- ous calculations that were required for the uncertain projects.”

2-15 Focus group participants and interviewees strongly empha- sized that the success of a transit project is a function of its relative “cost savings versus taking a car,” or essentially “how difficult it is to operate an automobile.” The cost of automo- bile use can be measured in monetary terms: one financial cost that participants commonly mentioned was “the price of parking,” more specifically, parking management programs that involved the “absence of free/heavily subsidized garages, and the absence of large lots.” However, an interviewee pointed out that “a high price of auto[mobile travel] . . . is not always expressed in dollars,” and that time costs caused by congestion or road network complexity can be just as important. Participants also mentioned several land use elements as critical factors in leading to a more successful transit proj- ect. One interviewee suggested that an existing opportunity for dedicated right-of-way should be considered, as it could indicate time savings on the line and also reduce the proj- ect’s costs of construction. Density was also an often-cited element, including “development density” and “density of the street grid” around stations. Participants stressed the importance of local land use policies that support TOD. As one interviewee expressed it, “If you build it, they will come” can apply only if “local land use policies and market demand allow it to happen. Policies can prevent development from occurring around new projects.” The current presence of a successful bus system in the region was also mentioned as an important predictor of the success of a potential rail project. One focus group member felt that “to do rail, a city must show they have an existing commitment to bus.” Another supported the sentiment by suggesting that one start by looking at whether “there is a market for bus when considering whether rail makes sense.” However, a few members of the focus group countered that current success with bus service is not always a valid indica- tor, because rail might work where a bus route is not feasible due to “geographic constraints” or “reliability issues” that make a bus service ineffective in the corridor. An interviewee stressed the importance of examining the existing conditions and capabilities of a line when extension projects are proposed. This interviewee recommended exam- ining the strength of current transit ridership within the cor- ridor, for example, “riders on existing bus routes along the arterial,” and proposed that the success of a project hinges on the project sponsor’s familiarity with the proposed technol- ogy. The interviewee also stressed the importance of “hook- ing it all together” and “reducing the complication associated with doing something different,” adding that “once the cen- tral system is established it does not make sense to incorpo- rate different technologies as the system expands.” Finally, participants suggested that the success of a transit project could be predicted through its connection to the rest of the system, including not only new linkages formed but At least one participant suggested that federal funding be linked to project outcomes, making local jurisdictions more accountable for a project’s success. One participant specifi- cally noted that “things that make a project successful (e.g., design quality, development density, supportive land use policies) have nothing to do with how the project is funded.” 3.2.3 Other than the FTA New Starts criteria, what indicators of success have you used or seen used in the planning phase to identify potentially successful fixed-guideway transit projects? What other indicators of success do you think should be used in planning-level evaluations of transit alternatives? In response to queries on potential indicators of success, participants in the focus group and interviews offered a variety of suggestions, including quality of service (e.g., convenience and reliability), cost savings versus private auto, density sur- rounding the corridor, supportive local land use policies, the region’s demonstrated commitment to transit, and the inte- gration of the new project into the existing system conditions (with respect to proposed technology and impacts on network connectivity). Many participants expressed the view that the observed quality of service of a transit line could act as a measure of success and that proposed service characteristics could help predict a project’s success. Convenience was one central factor mentioned, including proximity of the line to trip origins and destinations, providing people with the “ability to walk to and from stations,” as one interviewee put it. In particular, partici- pants stressed the importance of linking people with jobs. One participant noted that office proximity is more important than household proximity when an individual makes the decision to commute by transit. A participant summarized results from the 2005 WMATA Development-Related Ridership Survey, adding that “for workers to commute by transit to suburban offices, offices must be much closer to stations than worker households need to be. . . . People will walk a good distance for access to transit, but not to their final destination.” In addi- tion to the presence of jobs in a destination area, participants stressed the importance of having a destination area with a gen- erally vibrant and walkable environment. One interviewee in particular noted the value of “the ability to do things by foot in the area—the availability of options once you’re there,” adding that “this depends on safety, activity, and general vibrancy near stops.” Other essential features of service quality, according to one participant, included “frequent service with a long span of service” and ensured reliability of the proposed service, which would allow individuals to “work [transit] into their lifestyle.”

2-16 happened in the past . . . identifying the corridor character- istics associated with previous New Starts projects as use- ful points of comparison for new ones.” Other participants proposed that the research project might look at changes in travel over the last few decades, and use that knowledge to update the quantitative thresholds presented in Urban Rail in America (Pushkarev, Zupan, and Cumella 1982). As one interviewee observed, “The question is more compli- cated now, with wider ranges of variation in transit perfor- mance and cost. It is important to determine what ‘x’ level of demand can justify a ‘y’ level of investment.” This person further suggested that this study’s scope remain narrow and concrete, focusing on transportation benefits and avoiding more subjective and political measures like economic devel- opment, job creation, or contingency uses. The interviewee bemoaned the fact that “parties today often dismiss the fun- damental cost-benefit performance analysis and instead use these other [subjective political] justifications to rationalize the decision to move forward with a transit project.” 3.3 Conclusions Participants saw the primary challenge of measuring the success of transit projects as incorporating different assess- ment factors into a cost-benefit analysis. They suggested a wide array of potential measures of a project’s success, includ- ing ridership levels, service improvements, cost control, a general return on investment, improved regional efficiency and mobility, and creation of a transit-friendly environment. A variety of potentially simple predictors of success also were suggested, including quality of service, cost savings ver- sus the private auto, corridor density, supportive local land use policies, a demonstrated commitment to transit in the region, and the integration of the new project into the exist- ing system conditions. also spillover effects along the line. One participant warned about the importance of modeling the project’s “impact on the rest of the system—whether the project relieves or puts pressure on the core and what shifts in inbound and out- bound ridership occur.” Generally, participants suggested looking for whether a region employed a “sound system/ network planning process.” 3.2.4 What tools and data are needed to calculate or utilize these indicators? The research team received no direct recommendations from interviewees for specific tools and data that could poten- tially be used to calculate relevant indicators and predictors of success, but the participants did discuss on a general level their opinions about how simple or complex the tools should be. One participant suggested that success indicators be “sim- ple and understandable” for “local leadership and public or community stakeholders,” and recommended that success indicators be kept from becoming too complicated through “breaking things down into chunks.” Another participant disagreed, stating that project analysis is inherently complex and “you cannot simplify and get very accurate answers— there is too much subjectivity.” This idea of complexity in the process was echoed by an interviewee’s expressed sentiment that “simple sketch tools” cannot replace modeling, because “simple analytics do not make the problems we are trying to solve any simpler.” 3.2.5 Do you have any other suggestions with regard to research direction? Several participants offered additional recommendations on the general direction of the TCRP Project H-42 research. One participant suggested that the study “look at what has

2-17 C H A P T E R 4 Numerous possible measures of transit investment success exist. After considering several alternatives, the TCRP Proj- ect H-42 team focused on transit ridership at the project and system levels, because these are strong direct measures of the benefits of transit, although they are by no means perfect or appropriate in all cases. This chapter includes a discussion of the different kinds of indicators modeled as predictors of transit ridership for the two types of data: a cross-section of fixed-guideway projects and a time-series of metropolitan- level data on rail and bus passenger-miles traveled (PMT). Summary statistics are provided about the 55 transit invest- ments included in the analysis, as well as the 18 metropolitan areas with fixed-guideway transit investments that occurred from 2002 to 2008, the period of study for which the research team could assemble complete data for statistical analysis. (Notice that the entire MSA data set consisted of 244 met- ropolitan areas.) The data collection process is described in more detail in Appendix B to this report. 4.1 Defining Transit Project Success No definition of “success” for a fixed-guideway transit project is universally accepted. Project goals vary by region, by city, and by corridor, and they can be broad and multi- faceted. Standards that might be used to classify completed projects as highly successful, moderately successful, or unsuc- cessful simply do not exist. The literature reviews and focus groups in this project yielded a range of definitions but no definitive metric for measuring success. A project may be perceived as successful if it is built on time and on budget, or if ridership exceeds expectations. A tran- sit system’s success also can be measured through economic cost-benefit comparisons, its impacts on land uses, local mea- sures of equity and environmental effect, or its impacts on congestion and VMT. A person’s view of a project’s success may also depend on his or her perspective—a transit agency general manager or the agency’s board of directors may define success differently than a transit rider, a taxpayer, or a fund- ing partner does. Measuring success is largely driven by the primary goals of the city and the proposed project. In a city like Stockholm, which is committed to be a zero-carbon city by 2050, the ability of a proposal to reduce VMT by automobiles per capita is the primary concern. In other cities, overriding goals might be enhanced mobility or economic productivity. From an economics standpoint, a successful project is one whose benefits exceed its costs. Table 4.1 compares capital cost per mile by mode for proj- ects in this study. A full accounting of a transit project’s direct and indirect costs and benefits is analytically challenging. Many of the benefits and externalities are difficult to quantify and cannot be assigned a dollar value, such as a transit proj- ect’s contribution to making a city more livable. Another way to measure success might be to assess how fully a completed project meets the goals it was intended to achieve. The goals of fixed-guideway projects are many and varied, however, and they are often difficult to measure. For example, one proj- ect goal might be to improve access for poor people, but it is extremely difficult to establish a monetary value for this goal. Identifying a comprehensive and widely acceptable defini- tion of success proved to be elusive. The researchers there- fore focused on measures of success that can be quantified and that generally correspond with a range of project goals: project-level ridership, changes in system-wide transit use, and project-level cost. Though incomplete as a measure of success, the expected ridership on the project and the expected effect on the system’s usage as a whole, in combination with the cost of the project, provide valuable information to help establish a corridor’s potential for fixed-guideway transit. A simple model of capital cost was added to enable a rudimentary cost-benefit analysis. The study team used ridership because the number of pas- sengers offers one direct measure of the number of people who benefit. When a new transit project is proposed, one of the first things people want to know is how many people the system Conceptual Framework

2-18 is expected to carry. Increases in system-wide patronage can also serve as a proxy for a project’s mobility and accessibility benefits, as well as sustainability benefits such as reductions in automobile use, air pollutant emissions, and energy consump- tion, to the extent that increased system ridership indicates that more people are choosing to leave their cars at home and take transit instead. These benefits exist only if former motorists switch to transit, and if large shares of future trips by transit would otherwise have been taken by car. Ridership gains of fixed-guideway transit projects could instead come from for- mer bus riders, or from new trips not previously made. To some degree, ridership can also be viewed as a proxy for potential land use and economic development benefits. The more riders a project attracts, the more likely it is that the project will help stimulate growth. Changes in transit ridership result from, and can be viewed as a measure of, the improved transit speed and reliability produced by a project. Ridership is also a convenient indicator of success because transit ridership data can be both readily collected and statis- tically correlated with corridor conditions. Thus, the TCRP Project H-42 research identifies the conditions that are likely to lead to increases in transit ridership given investment in a new fixed-guideway transit project. The researchers also con- sider a project’s capital cost in relation to these measures of success. Recognizing the value of other potential measures of success and understanding the importance of employing a more-refined multiple-indicator approach when different projects are evaluated based on different metrics, the research team chose to address these additional measures and issues through case studies (see Chapter 6). These approaches are particularly relevant in the realm of policy-making. Given data limitations and the highly focused scope of work for TCRP Project H-42, the researchers were unable to address these additional success factors in the quantitative analysis for this study. The principal objective of this study was to create a simple method grounded in empirical analysis, using measures that are distinctive and intuitive. The additional measures add a level of complexity that makes them difficult to implement in practice and therefore incompatible with the project’s specific goals. Multicriteria performance evaluations for urban public transit systems involve multilevel hierarchies and subjective assessments of decision alternatives, expanding on the widely understood simple metrics of system use that are incorporated in the analysis. One example of a multiple-indicator metric was used for a study conducted in Istanbul in 2004 (Gercek et al. 2004). The authors evaluated three alternative rail tran- sit network proposals by using the Analytic Hierarchy Process (AHP), a multicriteria decision support system. The AHP facil- itates decision-making by organizing perceptions, experiences, knowledge, and judgments—the forces that influence the decision—into a hierarchical framework with a goal, scenar- ios, criteria, and alternatives of choice. Research by Yeh, Heng, and Chang implemented a fuzzy multicriteria analysis (MA) approach in a case study evaluating the performance of 10 bus companies in Taiwan (Yeh et al. 2000). In this methodology, the subjectivity and imprecision of the evaluation process were modeled as fuzzy numbers by means of linguistic terms. 4.2 Levels of Analysis When comparing the transit potential of different corri- dors, or the potential of different alternatives within a cor- ridor, the use of two complementary measures of project ridership is suggested: • Project-level ridership addresses the number of people who use a project on a daily basis, measured as average weekday boardings and alightings at project stations. Project-level ridership includes new riders attracted to transit, such as former automobile drivers who switch to transit or future travelers by transit who would otherwise have used a private car. It also includes existing riders, such as people who previously took the bus but who now ride the new fixed-guideway system and may benefit from faster travel time, improved reliability, or greater comfort. • System-level ridership addresses annual PMT across the entire system, as defined by rail and bus PMT, data that are reported on a yearly basis by transit agencies to FTA, and are collected in the National Transit Database. This metric represents the amount of new transit use that is expected once the project is in service. PMT takes into account the greater regional mobility that may occur when a single fixed-guideway project links riders to a regional system. It captures new riders and the length of their trips, but it does not incorporate existing riders whose trip length on transit does not change, even if these riders benefit from faster travel time. Compared with project-level ridership, the change in system-level PMT offers a better indicator of a project’s likely impact on overall highway congestion, emissions, and energy consumption. Table 4.1. Capital cost per mile of study set projects by mode. HRT LRT BRT Commuter Rail Cost/mile (2009 $) $251.2 million $61.0 million $49.8 million $10.5 million

2-19 Although project-level ridership is a fundamental compo- nent of a transit project’s success, success should also be con- sidered in the context of the entire regional transit system. Project-level ridership alone fails to account for possible shifts in modes between new transit projects and existing services such as parallel bus lines. System-wide PMT allows examination of changes in transit use across all lines and modes in the system, controlling for any possible shifts in mode and line. A regional approach has been noticeably absent from previous research on travel demand associated with fixed-guideway transit projects. Project-level and system-level measures are complemen- tary and offer different perspectives on a project’s benefits. A project that does well in one dimension may not do well in the other. An urban circulator, for example, may attract a significant number of riders. Given that circulator trips are typically short, however, the project may have little impact on PMT unless it makes longer-distance transit travel more convenient. A CR project, on the other hand, could have a larger impact on PMT even if ridership is not as high, because CR trips tend to be much longer. Neither measure alone tells the full story. Not surprisingly, then, the indicators of project ridership success and PMT success are somewhat different. A potential disadvantage of using PMT is that it may not be well suited to measuring transit use in larger metropolitan areas that are experiencing rapid suburbanization and seek to reduce vehicle travel distances. In crowded, dense cities, time spent in travel may be a better measure of success than distance traveled. The example of the urban circulator and the CR line is again instructive: the former could strengthen a downtown area, whereas the latter could contribute to decentralization. The primary levels of the analysis in this study were U.S. Census-defined metropolitan areas and individual transit line projects in the United States. In addition to gathering data at both of these levels, the research team collected spatial information and information at the individual station level, which was aggregated to the project level and metropolitan area for analysis. 4.3 Identifying Indicators of Transit Project Success A variety of factors can potentially influence the ridership levels of a fixed-guideway transit project. Some factors are attributes of the system itself. Such internal factors include service reliability, fare, frequency, vehicle speed, and service amenities such as comfort. Route alignment and connectivity of the transit network also are important factors that are par- tially controlled by the transit agency, and improving these elements of the system can potentially increase ridership and contain costs. Other potentially influential elements are outside a transit agency’s control. Such external factors include characteris- tics of the service population and surrounding metropolitan area. Population growth, improved economic conditions, and certain demographic attributes tend to increase ridership lev- els. Characteristics of the built environment also can play a significant role. Higher population and employment densi- ties, as well as mixed-use and more walkable neighborhoods, may lead to more transit use. One final potentially significant predictor of transit ridership is the relative cost of the auto. When driving is costly—that is, when congestion levels, park- ing prices and gas prices are high—people are more likely to choose transit over the auto. All of these factors are potentially significant indicators of the future success of a transit project. For the analysis used in this study, the researchers grouped them into four distinct categories: system characteristics, service population and metropolitan area characteristics, land use, and relative merit of alternative modes. Each of the categories is composed of a complex set of attributes that change with spatial scale and time, and the study team generated a set of variables that can be used to objectively quantify these attributes. 4.3.1 Project and System Characteristics Transit use can be related to characteristics of the transit service provided. When an agency improves the quality or expands the coverage of its transit service, ridership tends to increase. Conversely, when an agency increases fares, rider- ship is expected to drop. However, results vary significantly in studies that examine the price elasticity of transit use, as effects have been shown to differ by geographic location, time of day, and income level. The strength of this relation- ship between a transit project’s service and its ridership levels depends on the metrics used to quantify service. The simplest measure of a transit system’s service is its extent, which can be quantified at the project level as the com- bined length of the routes. At the system level, the directional route-miles of all fixed-guideway and bus lines can be an indicator of both the physical service area and the number of people being served by the network. Another service character- istic is a project’s level of connectivity to other transit networks, possibly measured by the number of bus lines to which it con- nects. All of these transit system characteristics may increase transit accessibility, which is the primary influence on transit ridership. More connections are likely to be correlated with higher transit use because in these areas more direct travel to many destinations is possible via transit. Some stations or lines are more frequently accessed by car than transit, and in this case service might be measured by the number of park-and- ride spaces provided at the line’s stations. Using service characteristics of a project or system presents a conundrum, because transit service decisions are made not only to increase patronage but also in response to demand.

2-20 team considered several aspects of land use and built a set of variables that capture the characteristics that facilitate public transportation. It is readily observed that many big cities have fixed- guideway transit but many small cities do not. The size of the city has implications for the number of potential users, the number of activities accessible by transit, and amount of capital available for building transit. The size of the city could be quantified in many ways, including the population, the area, the GDP, and the number of jobs. Residential and employment density was considered by examining catchment areas around each of the stations. By analyzing the number of jobs, the number of residents, and the interaction between job and resident figures in the catch- ments, relationships between urban densities and transit use can be identified. Density of workers in different industries, and residents of different types of housing or different income levels, is also potentially relevant. In TCRP Project H-42, such measures were found to be highly predictive of transit use. The idea of density is further explored through accessibility measures. Since the connection between density and transit use might be more specifically based on access to activities, gravity- based accessibility measures might be a more helpful way of describing the intensity of land use. (A gravity model predicts the number of trips from an originating zone to destination zones as a function of the attractive power of destination zones and the distance to each of them from the originating zone.) Because transit is often accessed by foot or bike, the walk- ability of the area around transit stations might be an impor- tant indicator of transit use. Although this characteristic can be nuanced, the researchers quantified walkability using the Street Smart Walk Score algorithm. This metric takes into account the accessibility of various amenities including retail, institutions, and dining. Additionally, the algorithm factors in the density of intersections and block lengths. Walk Scores are calculated on a range of 0 to 100, and higher walk scores might be expected to influence transit use by making the walk to and from transit stops more interesting and useful. In some cases, transit use may be influenced by the preva- lence of a specific industry. Using the North American Industry Classification System, the study team generated indicators of 20 industries using the number of jobs in each station catchment from the LEHD data. This allowed the models to distinguish between a dense commercial district and a dense industrial dis- trict, for example. Cases were found of differential effects on ridership or PMT by industry, as is discussed in Chapter 6. Land use also affects transit use through the presence of specific facilities and institutions. During the case studies, transit agencies reported that universities, stadiums, hos- pitals, museums, airports, and hotels can be important trip generators. These land uses can be more difficult to quan- tify without local knowledge, but it is possible to substitute Overflowing parking lots may be expanded; more bus lines may be added to a busy terminal; and very dense service areas are more likely to have undergrounded rights-of-way for transit. The researchers conducted analysis with and without those characteristics, knowing that including them as pre- dictors of success was problematic but wanting to compare results both ways. The study team was able to predict rider- ship fairly well without including transit system characteris- tics, suggesting that these characteristics may change partly in response to demand, either prospectively or over time. 4.3.2 Service Population and Metropolitan Area Characteristics Transit use might also be associated with characteristics of a project’s service population and surrounding metropoli- tan area. Such factors reflect the influences of sociodemo- graphic and environmental conditions on transit patronage. Resident age could be an important indicator of ridership, as both younger and older people tend to use transit more than people of middle age. Specific age data might include median age, the number of residents over age 65, or the num- ber of residents under age 18. In addition, certain population groups, such as university students, are often associated with higher transit use. Studies have shown that higher transit use is also linked to certain racial or ethnic group status (Taylor et al. 2009) and recent immigrant status (Chatman and Klein 2009). This effect can be isolated by examining concentra- tions of race, ethnicity, or immigration status. Other potentially influential characteristics of the metro- politan area are external to policy-makers. For example, weather and climate may affect transit ridership. Harsh envi- ronmental conditions might cause people to travel by automo- bile instead of waiting for transit and walking to connections, or it might lead to increased ridership as people who would normally walk or bike choose to make their trips by tran- sit. The National Climatic Data Center (NCDC), run by the National Oceanic and Atmospheric Administration (NOAA), quantifies weather and climate by providing metropolitan- level estimates for annual precipitation, percent of possible sunlight, average temperature, days with highs above 90°F, days with lows below 32°F, and average snowfall. Finally, though the study team did not include it in this analysis, it is possible that local crime rates or personal safety concerns on transit or at transit stations could heavily affect its use. 4.3.3 Land Use Many people have a mental conception of transit-friendly land use, but quantifying it requires more detail than a single number on a range from friendly to unfriendly. The research

2-21 combined 2008 population of this subset of 18 metropolitan areas in the data set totaled 75.6 million people, representing one-quarter of the total U.S. population. Table 4.3 provides a detailed descriptive summary of the 18 metropolitan areas. Figure 4.1 shows PMT per capita by transit for those metro- politan areas in 2008. Among the 18 metropolitan areas with a fixed-guideway transit investment during the 2002 to 2008 period, the highest intensity of transit ridership is in the San Francisco–Oakland– Fremont, CA, metropolitan area, based on per capita fixed- guideway transit PMT in 2008. This metropolitan area is the seventh largest of those modeled, with 4.3 million residents in 2008. 4.4.2 Fixed-Guideway Transit Projects of Study The ridership models include 55 fixed-guideway transit projects in 21 U.S. metropolitan areas (Table 4.4). The data set includes 13 HRT projects, 36 LRT projects, three CR proj- ects, and three fixed-route BRT projects. Because of the small number of CR projects in the data set, the researchers do not recommend the use of this model with proposed CR investments. Likewise, data was not available for streetcar or urban circulator projects—which may require entirely different indicators of success—so this model should not be applied to such investments, and no projects of that mode are in the data set. Although there is also good reason to use caution when using the model for BRT projects, better estimates from similar methods are not currently possible relevant jobs using industry data categories (e.g., healthcare industry workers as a proxy for hospitals). 4.3.4 Competition from Other Modes The service provided by the transit network should be con- sidered relative to the attractiveness of alternative modes. In particular, attributes that characterize the speed and conve- nience of driving—factors generally out of the control of a transit agency—may influence transit use. Transit use might be expected to increase when the cost of driving is high. This characteristic of a system can be quan- tified through gas prices and parking prices. Parking prices in the downtown area and parking prices in the catchments might have independent effects on transit use, depending on the type of trips that are dominant on a transit route. The level of investment in driving infrastructure might also determine use of public transit. One way of quantifying this is through the number of highway lane-miles, and aver- age daily traffic per lane-mile. 4.4 Observation Set Between 1974 and 2008 32 metropolitan regions had fixed- guideway transit and 126 fixed-guideway transit projects were completed in the United States (Appendix C). Ideally, the TCRP Project H-42 research would incorporate informa- tion on all of these projects and metropolitan regions, but the set of observations that were modeled is limited to the projects for which the team could secure data on ridership and its indicators. For example, the researchers’ set of system- level observations was restricted by the availability of LEHD employment data, which was used to construct measures of employment near proposed stations. The data that were available from 2004 to 2008 do not cover some areas of the United States with significant transit investments, notably Washington, D.C.; Charlotte, North Carolina; and Boston. The best-fit models in this study are therefore based on a subset of recently completed transit projects. 4.4.1 Metropolitan Areas of Study The PMT model includes data from 244 MSAs, 18 of which had a fixed-guideway transit investment occur during the study’s 7-year data set (2002–2008) (Table 4.2). Because transit service density and the sheer size of the New York City metropolitan area make it an outlier, the research team excluded the region in most of the analysis. The 18 regions varied in population between 1.1 and 12.8 million people in 2008. The largest metropolitan area studied was the Los Angeles–Long Beach–Santa Ana, CA, region; the smallest was the Salt Lake City, UT, region. The Atlanta–Sandy Springs–Marietta, GA Baltimore–Towson, MD Chicago–Naperville–Joliet, IL–IN–WI Cleveland–Elyria–Mentor, OH Dallas–Fort Worth–Arlington, TX Denver–Aurora, CO Los Angeles–Long Beach–Santa Ana, CA Miami–Fort Lauderdale–Miami Beach, FL Minneapolis–St. Paul–Bloomington, MN–WI Philadelphia–Camden–Wilmington, PA–NJ–DE–MD Pittsburgh, PA Portland–Vancouver–Beaverton, OR–WA Sacramento–Arden-Arcade–Roseville, CA St. Louis, MO–IL Salt Lake City, UT San Diego–Carlsbad–San Marcos, CA San Francisco–Oakland–Fremont, CA San Jose–Sunnyvale–Santa Clara, CA Table 4.2. Metropolitan areas included in analysis.

2-22 Table 4.3. Descriptive summary of metropolitan areas included in the analysis, 2002–2008. Descriptor Mean SD Min Max n Annual passenger-miles (thousands)a 961,443 1,064,339 142,510 4,154,660 124 Operating cost per thousand passenger- miles (millions, $2009) $713 $253 $364 $1,636 124 Population within 1/2 mile of stations 401,263 581,464 37,019 2,218,951 124 Annual passenger-miles per person residing within 1/2 mile of stations 3,941 2,670 628 11,973 124 Population of metropolitan area (thousands) 4,107 2,919 1,002 12,768 124 Annual passenger-miles per person residing in metropolitan area 206 122 74 609 124 Percent of metropolitan area population within 1/2 mile of stations 8% 7% 1% 26% 124 Population per total land area within 1/2 mile of stationsb 23 21 7 87 124 Jobs within 1/2 mile of stations 387,648 385,696 99,085 1,673,264 124 Annual passenger-miles per job within 1/2 mile of stations 2,332 878 582 4,675 124 Labor force of metropolitan area (thousands) 2,112 1,462 536 6,548 124 Annual passenger-miles per job in metropolitan area 396 233 147 1,156 124 Percent of metropolitan area jobs within 1/2 mile of stations 18% 9% 7% 35% 124 Jobs per total land area within 1/2 mile of stationsa 28 13 9 65 124 Retail, entertainment, and food jobs within 1/2 mile of stations 61,110 58,983 14,549 249,071 124 Higher-wage jobs within 1/2 mile of stations 182,314 184,598 28,216 884,079 124 Population under 18 within 1/2 mile of stations 31,060 37,272 3,873 132,413 124 Directional route-miles of system 4,282 2,797 1,471 14,214 124 Average walk score of stations 68 6 58 84 124 Real GDP (millions, $2005) $208,921 $151,659 $47,847 $695,513 124 Per capita income $40,615 $6,597 $29,892 $62,427 124 Average daily traffic per highway-lane 16,274 3,073 7,377 20,425 124 Congestion index 6,955,203 16,200,000 66,095 71,500,000 124 Average gas price by county $2 $1 $1 $4 124 a Passenger-miles include rail and bus services. b Differs from Pushkarev and Zupan, who used residential land area.

2-23 Figure 4.1. 2008 annual fixed-guideway transit passenger-miles per person, by metropolitan area with fixed-guideway transit included in analysis. 0 100 200 300 400 500 600 700 Dallas-Fort Worth-Arlington, TX Sacramento--Arden-Arcade--Roseville, CA St. Louis, MO-IL San Jose-Sunnyvale-Santa Clara, CA Cleveland-Elyria-Mentor, OH Pittsburgh, PA Minneapolis-St. Paul-Bloomington, MN-WI Atlanta-Sandy Springs-Marietta, GA Miami-Fort Lauderdale-Miami Beach, FL San Diego-Carlsbad-San Marcos, CA Portland-Vancouver-Beaverton, OR-WA Denver-Aurora, CO Los Angeles-Long Beach-Santa Ana, CA Baltimore-Towson, MD Philadelphia-Camden-Wilmington, PA-NJ-DE-MD Salt Lake City, UT Chicago-Naperville-Joliet, IL-IN-WI San Francisco-Oakland-Fremont, CA Annual PMT per Resident of Metropolitan Area, 2008 Table 4.4. Fixed-guideway transit projects included in analysis. State City Project Name Mode AZ Phoenix Metro Light Rail LRT CA Los Angeles Long Beach Blue Line LRT Red Line Segments 1,2,3 HRT Green Line LRT Pasadena Gold Line LRT Orange Line BRT CA Sacramento Sacramento Stage I LRT Mather Field Road Extension LRT South Phase 1 LRT Sacramento Folsom Corridor LRT CA San Diego Blue Line LRT Orange Line LRT Mission Valley East LRT CA San Francisco Initial BART HRT BART SFO Extension HRT CA San Jose San Jose North Corridor LRT Tasman West LRT VTA Tasman East and Capitol Segments LRT VTA Vasona Segment LRT CO Denver Central Corridor LRT Denver Southwest Corridor LRT Denver Southeast (T-REX) LRT FL Miami Metrorail HRT South Florida Tri-Rail Upgrades CR (continued on next page)

2-24 Table 4.4. (Continued). State City Project Name Mode Portland Airport MAX LRT Portland Interstate MAX LRT LRT PA Philadelphia SEPTA Frankford Rehabilitation HRT TX Dallas S&W Oak Cliff and Park Lane LRT North Central LRT UT Salt Lake City North South Corridor LRT University and Medical Center Extension LRT WA Seattle Seattle Central Link Light Rail Project LRT a Jersey City and Newark, New Jersey, belong to New York City’s metropolitan area. Trenton, New Jersey, is part of the Philadelphia metropolitan area. GA Atlanta North/South Line HRT North Line Dunwoody Extension HRT IL Chicago O'Hare Extension (Blue Line) HRT Orange Line HRT Metra North Central and SW Corridors CR Douglas Branch HRT MD Baltimore Baltimore Metro HRT Central Line LRT Three extensions LRT MN Minneapolis Hiawatha Corridor LRT NJ Jersey Citya Hudson-Bergen MOS 1 and 2 LRT NJ Newarka Newark Elizabeth MOS-1 LRT NJ Trentona Southern NJ LRT System LRT NY Buffalo Buffalo Metro Rail LRT OH Cleveland Cleveland Healthline BRT OR Eugene Eugene EmX BRT OR Portland Portland MAX Segment I LRT Portland Westside/Hillsboro MAX LRT because this study includes almost all possible fixed-guideway BRT projects currently operating in the United States. Appendix F provides a summary of the projects included in the ridership models for TCRP Project H-42, and Appendix C provides information for 71 other projects completed in the past 40 years that the researchers did not include because of age or because data were missing. Figure 4.2 shows the 55 projects within their respective tran- sit networks and metropolitan areas across the United States. The 55 transit projects included in the ridership model opened as early as 1974 and as recently as 2008. The projects range in size from 1 to 72 route-miles in length and from 2 to 33 stations. The longest projects are typically CR lines, whereas the systems with the most stations are often a city’s first investment in a particular transit mode. Such projects were termed initial, and the database includes two initial HRT projects, 11 initial LRT projects, and one initial BRT project, for a total of 14 “initial” projects. In aggregate, the projects represent 849 bidirectional route- miles of fixed-guideway (approximately 88 below grade and 130 elevated track) with 774 stations and 151,564 transit agency-owned parking stalls. The total cost of constructing the projects in 2009 dollars was $54.4 billion. Table 4.5 pro- vides a more detailed descriptive summary of the 55 transit projects. Figure 4.3 shows the distribution of average weekday rider- ship by transit project, and Figure 4.4 shows the distribution of average weekday ridership per guideway-mile by transit project. The research team deliberately did not establish a typology of indicators according to fixed-guideway transit type (e.g., initial versus expansion project), transit mode (e.g., LRT, HRT, CR, BRT) or by urban setting (e.g., based on surround- ing densities or whether location is a CBD, central city, inner suburb, or outer suburb). The approach was instead to run analyses that included appropriate measures to render vari- ables representing type and mode statistically insignificant, given that such measures are imprecise. Other indicators were sufficient to predict ridership according to the statistical tests used, enabling the method to avoid relying on somewhat arbi- trary definitions of HRT, LRT, and BRT—categories that have large overlaps in service quality and capital cost. Although the researchers did not model differences in indicator effects among metropolitan areas of different sizes, measures of city size were tested extensively. (Note: for the rudimentary capital cost model appearing in the spreadsheet only, mode is included to help estimate capital cost.)

Figure 4.2. Fixed-guideway transit projects included in analysis. Table 4.5. Descriptive summary of fixed-guideway transit projects included in analysis. Descriptor Mean SD Min. Max. n Average weekday ridership 28,470 41,092 1,065 284,162 55 Total capital cost (millions, $2009) $950 $1,137 $26 $6,960 55 Route-miles 15 15 1 72 55 Percent at grade 69% 35% 0% 100% 51 Percent below grade 15% 27% 0% 100% 51 Percent elevated 16% 27% 0% 100% 51 Number of stations in alignment 13 8 2 33 55 Opening year 1998 8 1974 2008 55 Age of project 10 8 0 34 55 Frequency of trains in peak AM hour 13 6 4 26 55 Number of bus lines that connect to stations 54 59 0 339 55 Transit-owned parking stalls per station 3,087 4,563 0 29,778 51 Jobs within 1/2 mile of stations 70,355 63,719 4,819 311,300 55 Population within 1/2 mile of stations 55,754 53,159 1,709 269,182 55 Population of metropolitan area (thousands) 5,424 4,657 348 18,969 55 Average daily parking rate within 1/2 mile of stations $10 $5 $2 $26 44 Average daily parking rate in the CBD $15 $8 $4 $38 55 Average county gas price $3 $0 $3 $4 55 Capital cost per thousand riders (millions, $2009) $50 $45 $4 $211 55 Capital cost per route mile (millions, $2009) $93 $124 $4 $755 55

2-26 0 50000 100000 150000 200000 250000 300000 Newark: Newark Elizabeth MOS-1 San Jose: Tasman West Chicago: Metra North Central San Jose: VTA Capitol Segment - Connected to Tasman East Portland: Portland Airport Max San Jose: Tasman East Salt Lake City: Medical Center Ext. San Jose: VTA Vasona Segment Chicago: Metra Southwest Corridor San Diego: Mission Valley East Baltimore: Three extensions Sacramento: Sacramento Folsom Corridor Eugene: Eugene EMX Sacramento: Mather Field Road Extension Salt Lake City: University Ext. Portland: Portland Interstate MAX LRT Trenton: Southern New Jersey Light Rail Transit System Denver: Denver Southwest Corridor Atlanta: North Line Dunwoody Extension Sacramento: South Phase 1 San Jose: San Jose North Corridor Dallas: North Central Cleveland: Cleveland Healthline Chicago: Douglas Branch Denver: Denver Southeast (T-REX) San Francisco: BART SFO Extension Seattle: Seattle Central Link Light Rail Project Chicago: O'Hare Extension (Blue Line) Los Angeles: Orange Line San Diego: Orange Line Los Angeles: Pasadena Gold Line Buffalo: Buffalo Metro Rail Baltimore: Central Line Los Angeles: Red Line (Segment 1) Los Angeles: Red Line (Segment 3) Minneapolis: Hiawatha Corridor Los Angeles: Green Line Sacramento: Sacramento Stage I Salt Lake City: North South Corridor Chicago: Orange Line Portland: Portland Westside/Hillsboro MAX Denver: Central Corridor Miami: South Florida Tri-Rail Upgrades Baltimore: Baltimore Metro Jersey City: Hudson-Bergen MOS 1 and 2 Phoenix: Metro Light Rail San Diego: Blue Line Philadelphia: SEPTA Frankford Rehabilitation Los Angeles: Red Line (Segment 2) Dallas: S&W Oak Cliff and Park Lane Miami: Metrorail Portland: Portland MAX Segment I Los Angeles: Long Beach Blue Line Atlanta: North/South Line San Francisco: Initial BART Average Weekday Ridership Figure 4.3. Average weekday ridership, by fixed-guideway transit project included in analysis.

2-27 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Chicago: Metra North Central San Jose: Tasman West Trenton: Southern New Jersey Light Rail Transit System Chicago: Metra Southwest Corridor Miami: South Florida Tri-Rail Upgrades Portland: Portland Airport Max Sacramento: Sacramento Folsom Corridor Baltimore: Three extensions San Jose: Tasman East San Jose: San Jose North Corridor San Diego: Mission Valley East San Jose: VTA Capitol Segment - Connected to Tasman East San Jose: VTA Vasona Segment Denver: Denver Southeast (T-REX) Dallas: North Central Denver: Denver Southwest Corridor Newark: Newark Elizabeth MOS-1 San Diego: Orange Line Baltimore: Central Line Sacramento: Mather Field Road Extension Portland: Portland Interstate MAX LRT Seattle: Seattle Central Link Light Rail Project Los Angeles: Green Line Los Angeles: Orange Line Sacramento: South Phase 1 San Diego: Blue Line Eugene: Eugene EMX Los Angeles: Pasadena Gold Line Sacramento: Sacramento Stage I Los Angeles: Long Beach Blue Line Cleveland: Cleveland Healthline Portland: Portland Westside/Hillsboro MAX Phoenix: Metro Light Rail Salt Lake City: North South Corridor Salt Lake City: Medical Center Ext. San Francisco: BART SFO Extension Dallas: S&W Oak Cliff and Park Lane Chicago: Douglas Branch Minneapolis: Hiawatha Corridor Jersey City: Hudson-Bergen MOS 1 and 2 Miami: Metrorail Chicago: O'Hare Extension (Blue Line) Salt Lake City: University Ext. Baltimore: Baltimore Metro Chicago: Orange Line Buffalo: Buffalo Metro Rail Portland: Portland MAX Segment I San Francisco: Initial BART Los Angeles: Red Line (Segment 3) Atlanta: North Line Dunwoody Extension Atlanta: North/South Line Denver: Central Corridor Los Angeles: Red Line (Segment 2) Los Angeles: Red Line (Segment 1) Philadelphia: SEPTA Frankford Rehabilitation Average Weekday Ridership per Directional Route Mile Figure 4.4. Average weekday ridership per directional route mile.

2-28 C H A P T E R 5 The project team tested more than 140 different factors that might be expected to influence project-level ridership or system-level PMT on transit. Multiple regression analyses were conducted using the projects and cities for which the researchers had complete data. The model-building process proceeded down multiple parallel tracks, while always being constrained by having a relatively small number of observa- tions available. The first step was to specify models based on utility theory, focusing on the relative costs of transit and automobile use as reflected in the number of near- station households and workers, parking costs, congestion, and transit connectivity, along with built-environment measures and neighborhood sociodemographic attributes. Because of the small data set and the problem of correlated variables, the research team also tested several stepwise approaches to model-building. Several important variables reflecting transit agency decisions—such as the number of bus lines serving fixed- guideway transit stations, parking availability, and the per- centage of the track at grade—are highly correlated with rail use, but are problematic as predictors, because they both reflect and generate demand. Therefore, this report presents results with and without these highly correlated service variables and discusses the differences. For both project-level ridership and system-level PMT, the research team built parsimonious regression models (the simplest plausible models with relatively few predictive indi- cators) that reflected a physical explanation of the factors that drive transit use, such as access to stations, origins and desti- nations of different types on the transit network, the costs of driving, and the size of the metropolitan area. Generally, the starting point was a complete model reflecting a utility theory of transit use, to the extent possible with aggregate data. The researchers then began to pare down the model, rejecting vari- ables that the analysis showed to be insignificant predictors of ridership. This approach was complemented by building a model based on the significance of a larger set of variables. Here the variables that were considered to be most impor- tant (based on theory and their significance) were tested first. New variables were incrementally added and retained if they were significant and improved the goodness of fit. The results of this approach were used to inform the set of variables in the complete theoretical model. The model-building process used in this study was iterative and exhaustive. The combination of population and employment near sta- tions with parking costs in the downtown area was highly correlated with project ridership. System-level PMT was correlated with the population of the metropolitan area; the number of higher-wage jobs and leisure (retail, enter- tainment, and food) jobs within ½ mile of stations in the metropolitan area; and the interaction of road congestion, population, and employment near stations. Transit travel speed and frequency were not significant predictors of proj- ect ridership when controlling for other factors. These are endogenous and likely are determined by transit managers in response to anticipated or actual demand, so this result is less surprising than it might at first seem. Several other factors have been thought by transit manag- ers to affect transit use, or have been shown to be correlated with transit patronage in previous research. These factors include mixed-use development near stations, walkability (as measured using walk scores), whether the project serves the downtown area, key trip generators such as stadiums or universities, and the weather in the area. In some cases, such as local intersection density, the researchers were unable to acquire a direct measure of these factors within the budget and timeframe of TCRP Project H-42. In the remaining cases, however, the researchers tested the measures and found that including them did not improve the performance of the models (see Chapters 4 and 5). Table 5.1 provides summary statistics for the indicators of greatest statistical significance in explaining project rider- ship and system-wide PMT on transit. A full list of the indi- cators considered in the research and their contributions as Quantitative Analysis Methods and Findings

2-29 predictors of success can be found in Appendix E. For values of significant indicators for each of the 55 projects in the ridership model, see Appendix F. Many of these indicators of project ridership and system- level PMT are outside the control of local transit agencies or local governments; however, jobs and people within ½ mile of stations could be affected by public policy. In the longer term, transportation and land use planning decisions are likely to affect congestion and the monetary price of travel, including parking costs. Nevertheless, it is primarily the station-area-specific data that are relevant to comparing different corridors and station locations in terms of their potential for success. The daily parking rate indicator deserves special mention. Although the price of parking was found to correlate with ridership, the price of parking may actually be a reflection of a variety of conditions that are positive for transit ridership, such as density and transit-supportive public policies. Most importantly, however, it reflects the relative cost of rail tran- sit’s chief competitor, the private automobile. When automo- bile travel is relatively costly and there are many near-station jobs and residents, the project can be expected to have higher ridership. As in previous research, the study team for TCRP Proj- ect H-42 found employment and population densities to be highly predictive of a proposed transit project’s success, but Variable Name Description Obsa Mean Median Project Ridership Model Jobs near stations Employment within ½ mile of project stations 55 70,355 46,107 Population near stations Population within ½ mile of project stations 55 55,754 42,224 Transit utility (Jobs population parking rate)/106 55 113,077 30,695 CBD parking rate Daily parking rate in the CBD 55 15 14 Project age Age of project 55 10 7 Ridership 55 28,470 21,350 Predicted Ridership 55 28,470 19,344 Metropolitan Area PMT Model Jobs near stations Jobs within ½ mile of fixed-guideway stations in metropolitan area 141 b 250,112 187,042 Population near stations Population within ½ mile of fixed-guideway stations in metropolitan area 141 b 239,984 112,926 Leisure jobs near stations Retail, entertainment, and food jobs within ½ mile of fixed-guideway stations in metropolitan area 141b 38,611 26,380 High-wage jobs near stations Jobs with salaries exceeding $3,333/month within ½ mile of fixed-guideway stations in metropolitan area 141b 118,844 84,359 Congestion indexc Total VMT divided by number of freeway lane-miles in MSA (FHWA) 1,641 10,275 10,339 MSA jobs Overall employment in MSA (LEHD) 1,888 211,323 86,621 MSA population Overall population of MSA (BEA) 1,888 706,284 289,937 MSA leisure jobs Retail, entertainment, and food jobs in MSA (LEHD) 1,888 44,533 18,973 MSA high-wage jobs Jobs with salaries exceeding $3,333/month in MSA (LEHD) 1,888 72,267 26,222 PMT 1,888 84,309 6,775 a The ridership model has a single observation for each investment, whereas the PMT model records an observation for each year in each MSA. b Catchment variables are summarized only over MSA-years in which catchment population was positive (i.e., those in which fixed-guideway transit was operating). c Variable does not vary by year—multiple observations have repeated values Table 5.1. Summary statistics for model variables.

2-30 the interaction of residents and jobs near stations was found to be particularly important in conjunction with high park- ing costs. This measure captures the exponentially increasing value of a well-connected network of origins and destinations. 5.1 Project-Level Models The research team’s initial statistical model of project-level ridership was designed to capture the following concepts associated with high transit use: • A large number of workers, shoppers, and residents have good access to stations. • The relative time costs of driving versus transit are high. • The project is connected to a larger network serving activity centers and other residents. • Jobs and housing are balanced over the project and/or system. Ridership is reported in different ways by different agencies. For this project the researchers used average weekday ridership, measured as the average of non-summer weekday boardings and alightings on project stations. Multivariate regression mod- els were constructed using data from existing BRT, LRT, and HRT projects in the United States. The regressions extracted potential indicators from the database of 600 independent variables discussed in Chapter 5. The richness of the data set emphasizes the importance of finding a parsimonious model, as including all of the presumptively relevant indicators is simply not possible. From the original group, the study team sought independent variables and interactions between inde- pendent variables that were most effective at predicting rid- ership. For those variables that describe characteristics of the station catchments, consistent catchment sizes within the sets of employment and household variables were preferred in order to improve usability, based on focus group feedback and case studies. The researchers opted to use a ½-mile catchment after determining there was little loss of precision from specifying different catchment sizes and that the ½-mile catchment tended to perform as well or better than the ¼-mile or 1-mile catch- ments for various variables specified on a station-area basis. 5.1.1 Findings The first of the final models expresses ridership as a func- tion of jobs and population around the stations, parking rates in the CBD, the percent of the alignment at grade, the number of park-and-ride spaces, and the age of the project (Table 5.2). Specifically, the ridership is predicted by Equation 1: 0.12 _ 0.04 _ 393.64 0.05 9,971.61 % 3.38 & 707.94 8,235.44 (1) R P Jobs P Pop P R P R Age Rate Int Grade [ ] [ ] [ ] [ ] [ ] [ ] [ ] = + − + − + + + Although this is the best-fit model mathematically, it includes variables that may be endogenous. (See Section 5.1.3 for more on endogenous variables.) A more theoretically defensible model omits the number of park-and-ride spaces, expressing ridership in terms of jobs and population around the stations, percent at grade, parking rates in the CBD, and the age of the project, as shown in Equation 2: 0.16 _ 0.01 491.9 0.08 3,294.39 17,846 % 913.39 4,431.84 (2) R P Jobs P P R D Age Pop Rate Int Grade Grade [ ] [ ][ ] [ ] [ ] [ ] [ ] = − − + + − + + A comparison of predicted and actual ridership for the model including park-and-ride spaces is shown in Figure 5.1 and for the model omitting park-and-ride spaces in Figure 5.2. Notice that the scatter for the latter is a bit larger than for the model including park-and-ride spaces, but not dra- matically so. Table 5.3 shows five project-level models, as follows: 1. The final endogenous model; 2. The final defensible model, which is the model used in the spreadsheet tool; 3. Model C, illustrating the impact of including endogenous variables for level of service and the number of bus connec- tions available at stations, and showing that these variables are not statistically significant; Variable Name Abbreviation Definition Catchment jobs P_Jobs Jobs within 1/2 mile of project stations Catchment population P_Pop Pop. within 1/2 mile of project stations CBD parking rate P_Rate Daily parking rate in CBD Ridership model interaction term R_Int (I_Jobs × I_Pop × P_Rate)/(1 million) Percent at grade %_Grade Percent of alignment at grade Missing at-grade values dummy D_Grade 1 if %_Grade info missing; 0 if not Number of park-and-ride spaces P&R Number of park-and-ride spaces Project age Age Age of the project Table 5.2. Summary of variables in final ridership models.

2-31 town parking rates are high and the project serves many jobs and residents, ridership tends to strongly increase. This inter- action term contributes more than any other term to the fit of the ridership model. At first glance, the sign on the parking rate coefficient might seem counterintuitive. However, the influ- ence of parking cost also gets picked up through its association with job and population densities in the interaction term. The combination of concentrated housing and employment with parking charges boosts transit ridership more than downtown parking rates alone. The net effect of parking rates on transit ridership is positive because the effects of the interaction term eclipse the effects of the CBD parking rate. Finally, as the equa- tions indicate, transit ridership tends to rise as projects mature, and a project that is entirely at grade has fewer riders than does a subway or elevated rail line. 4. Model D, including indicator variables for HRT and BRT modes. Notice that neither coefficient is statistically sig- nificant; and 5. Model E, showing the simplest model expressing weekday ridership in terms of near-station jobs and population. This model has the largest sample and both variables are significant, but the fit is relatively poor. As shown in Table 5.3, it is the interaction between jobs, pop- ulation, and downtown parking cost that best predicts ridership. Jobs and residents near stations are not statistically significant on their own in the best model (Table 5.3, Column 2—Final Models, Defensible). The effect of the CBD parking rate illustrates the interaction between driving costs and transit convenience. When down- 50,000 0 50,000 100,000 150,000 200,000 250,000 300,000 0 50,000 100,000 150,000 200,000 250,000 300,000 Pr ed ic te d Ri de rs hi p Actual Ridership Figure 5.1. Predicted versus actual ridership for the endogenous model. 50,000 0 50,000 100,000 150,000 200,000 250,000 300,000 0 50,000 100,000 150,000 200,000 250,000 300,000 Pr ed ic te d Ri de rs hi p Actual Ridership Figure 5.2. Predicted versus actual ridership for the defensible model.

2-32 centage of the variation in ridership that is explained by each variable. As shown in Figure 5.4, the interaction term by itself explains about 62 percent of variation; jobs within ½ mile of stations, another 20 percent; and variations in percent at grade, an additional 16 percent. 5.1.3 Endogenous Variables Some variables that are intuitively associated with high ridership both cause and are caused by transit use. Some- times these endogenous variables represent attributes that could be retrospectively adjusted to accommodate high tran- sit use. For example, a transit agency might increase the num- ber of park-and-ride spaces, the frequency of service, or the number of bus connections if demand exceeds the planned capacity. Other variables are prospectively adjusted because 5.1.2 Comparing Variable Impacts In understanding the contribution of different variables to the explanatory power of the model, one useful method is to compare the beta weights, which normalize the variable coefficients from the model by the standard deviation of the variable (these are sometimes also referred to as standard- ized regression coefficients). These are unitless coefficients between 0 and 1, reflecting the relative predictive power of variables in the model. Beta values for the defensible model are shown in Figure 5.3. Notice that the interaction term has the largest magnitude coefficient relative to the range of the variable. Catchment jobs, population, and CBD parking rate are more influential in combination than they are individually. Another illustration of the relative influences of the vari- ables is a partial R2 analysis. The partial R2 represents the per- Variable Name Final Models Rejected Models Endogenous Defensible Model C Model D Model E Catchment jobs 0.117** 0.155 0.0646 0.122** 0.324** Catchment population 0.0384 -0.0140 0.00103 0.0441 0.309* CBD parking rate -393.6 -491.9* -354.2 -462.7 Ridership interaction term 0.0455** 0.0773*** 0.0441* 0.0470** Percent at grade -9,971.6* -17,846.2* -10929.1* -3028.4 Missing at-grade dummy 3,294.39 Park-and-ride spaces 3.383** 3.170* 3.139** Age of project 707.9** 1,040.3** 574.3* 659.0* Number of bus lines 100.4 Level of service 340.2 HRT dummy variable 7,757.3 BRT dummy variable 880.2 CONSTANT 8,235.4 20,672.69** 5,917 2,854 -11,258.3 Number of observations 50 55 50 50 56 Adjusted R2 0.939 0.894 0.942 0.939 0.656 * p < 0.05, ** p < 0.01, *** p < 0.001 Table 5.3. Summary of project-level ridership models. 0.16 0.10 0.02 0.25 0.71 0.3 0.2 0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Percent At Grade CBD Parking Rate Catchment Populaon Catchment Jobs Interacon (Jobs×Pop×Parking Rate) β Weight Figure 5.3. Beta weights for defensible ridership model.

2-33 5.2 System-Level Models Another measure of the success of a transit project is its impact on the entire metropolitan-wide transit system. The research team’s second set of analyses examined incremental changes in annual system-wide PMT on rail and bus. This measure is intended to capture the impact of the project with- out double-counting usage that may have shifted from other transit routes. The early PMT models for this study included only MSAs with fixed-guideway transit, for a total of 18 metro- politan areas after attrition due to missing data. The research- ers ultimately estimated a more comprehensive model that included 244 MSAs with available data. The two final models (described in Section 5.2.2) predict negative increments 50% and 80% of the time, respectively, when used to retroactively project PMT for the completed investments in the project-level database. This surprising finding is discussed in more detail in Section 5.3 It should be noted that falling PMT over multiple years is not uncom- mon, regardless of whether any investments in transit were made. Of all city-years in the data set, 59% show stagnant or dropping PMT. Furthermore, flat or falling PMT is more common in city-years with fixed-guideway transit in place (79/122, or 65% of city-years in the data set show less than 5% growth). 5.2.1 Method of Analysis The goal of this method is to identify indicators of transit use at the metropolitan level. By comparing predictions for the current state of the system and the system after a tran- sit project, the overall impact of the project can be estimated. Specifically, incremental changes in the annual metropolitan PMT due to individual fixed-guideway transit projects are identified. The researchers used a panel data set composed of 244 cities observed over 7 years (2002–2008). One technique for addressing panel data is a fixed-effects model, which calcu- lates a unique constant baseline PMT value for each metro- politan area to capture metropolitan-level differences. The fixed-effect technique is useful for making predictions for there is reason to believe the project will have high ridership. Examples of these variables include percent of the alignment that is at grade, or the design speed of the system, in which added construction or capital expense is acceptable because of high anticipated use. Although there is no doubt that tran- sit ridership increases in response to more frequent and faster service, determining the exogenous component of demand associated with parking, bus service, and other service mea- sures is a difficult task. It can only be presumed that transit agencies make the best possible decisions about service fac- tors and, therefore, that the effects of such variables reflect judgments about existing levels of demand (or likely future levels of demand) rather than actually causing ridership. Also notable, based on discussions from the case stud- ies, is that including parking spaces and bus connections might be confusing from the user’s perspective. It would be incorrect in many cases to infer that building more park- ing spaces without otherwise changing the project would increase ridership. As a compromise, the researchers did include one endogenous measure—the percent of the proj- ect at grade—in the project-level ridership model. This measure can only be prospectively adjusted, but it may be highly correlated with other service characteristics such as travel frequency, speed, and reliability. One indicator, project age, exhibits some endogeneity that has a minor impact on the predictive power of the model. The age of the project is a useful predictor of use with older projects experiencing higher ridership. This variable captures three phenomena: maturation, prioritized selection, and attrition. The most straightforward interpretation of project age is that the transit project will mature as travelers adjust their behavior and land use responds. Additionally, cities tend to prioritize projects with high expected ridership, so older projects also tend to be those with high demand. The age of the project may also affect estimates because unsuccessful projects may be discontinued and are therefore absent from the sample. Including age is less problematic than including other endogenous variables because it is not possible for an agency to increase the age of the project in the same way that it could add a park-and-ride lot or make service more frequent. 0% 6% 16% 20% 62% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Catchment Populaon CBD Parking Rate Percent At Grade Catchment Jobs Interacon (Jobs×Pop×Parking Rate) Paral R2 Figure 5.4. Partial R2 values for defensible ridership model.

2-34 pared to the value predicted by the final MSA-level model. For reference, the figure includes a black line representing a perfect prediction. 5.2.2 Findings The final model expresses system-wide annual PMT in terms of the metropolitan area’s population, congestion level, and information about the ½-mile radius catchment areas around all rail stations in the region, including popu- lation; jobs; the number of jobs associated with food, shop- ping or entertainment; and the number of high-wage jobs. The values for the catchment measures change in any year after a new project comes into service, so the data represent both within-city and across-city variation. Table 5.4 pre- sents a summary of the variables used in this study’s final PMT models. The final catchment-level model is specified as shown in Equation 3: PMT 2.54 0.223 8.44 3.28 1.01 0.0610 0.0147 18,977 (3) Catch Jobs Pop LeisJobs HWJobs Cong Int Pop Catch Catch Catch Catch PMT MSA BEA = − − + + − + − −( ) The final MSA-level model was not used in the spread- sheet tool. This model includes MSA-wide employment and population data derived from the same LEHD source as the catchment area data. These additional variables were included to control for how metropolitan-level characteristics affect system-wide transit use. Including these terms resulted in the specification shown in Equation 4: metropolitan areas included in the analysis, but it is some- what less reliable in terms of its ability to make predictions for metropolitan areas that were not included. Also, as the number of metropolitan areas in the sample grows, fixed- effects models become less efficient. Alternatively, random- effects models fit a distribution of variation between cities rather than estimating a specific value for each one, based on characteristics of the cities such as their total popula- tion or their climate. Statistically speaking, a random-effects model is a more efficient technique for a panel of many met- ropolitan areas, because the metropolitan-level characteris- tics can be described by a small set of variables (as compared to the fixed-effect models that require one variable for each metropolitan area). There is an important distinction in the interpretation of the estimated coefficients for random-effects and fixed-effects models. In random-effects models, the coefficients represent a combination of between- and within-city effects, whereas the coefficients from the fixed-effects model describe only the average of the within-city effect of the variables. For both approaches, the researchers assume that the omitted variables accounted for in the city-level effect are stable over the period of the project. The composition of the final system-wide PMT model was selected from a set of 93 variables and interactions (see Appen- dix E). The researchers selected ½-mile catchments around fixed-guideway stations in the metropolitan area to be consis- tent with the project-level model. Both fixed- and random-effects regressions were estimated. A Hausman test indicated that a random-effects approach was superior for the final models. Figure 5.5 shows the actual PMT observed in the 244 modeled MSAs (not increments) com- 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 0.0 1.0 2.0 3.0 4.0 5.0 Pr ed ic te d Ri de rs hi p M ill io ns Actual Ridership Millions Figure 5.5. Predicted versus actual PMT for the best-fit model.

2-35 level model, contains employment data only from within ½ mile of fixed-guideway stations in the MSA. This model is applied in the spreadsheet tool. The second model adds a set of metropolitan area variables for employment from the LEHD. The third model is the same as the MSA-level model except that, instead of using the BEA count of MSA population, it uses less recent but more easily obtainable MSA data from the 2000 U.S. Census. The census MSA population and congestion index vari- ables are the only ones that do not vary by year. Given that the metropolitan data set is built in MSA-years, these variables will have many repeated values across observations, whereas other variables will have a unique value for every data point. Just as for the ridership model, beta values (normalized coef- ficients) were calculated for the variables in the two final PMT models. The strongest station-based influences on passenger- miles of transit usage were the number of high-wage jobs and the number of leisure jobs near stations (Figure 5.6). Metropolitan-level measures such as the overall MSA popu- lation employment and jobs by type also exert large influences on PMT. The impact of population is intuitive, as the largest urban areas would be expected to have the busiest transit sys- tems. Interestingly, both catchment and MSA employment show the same trend: leisure and high-wage jobs both have a strong positive impact on PMT, whereas other types of employment have a negative influence when controlling for population growth, high-wage jobs, and leisure jobs. PMT 2.21 0.661 7.41 3.16 1.12 0.479 0.332Jobs 0.486HWJobs 0.189LeisJobs 0.273Pop 64,450.4 (4) MSA MSA MSA MSA MSA Census LeisJobs HWJobs Cong Catch Catch= − − + + − + − + + + −( ) As expected, population and jobs in the station catchments remain important indicators of system-wide PMT. The story is complex, however: population and jobs near stations, when interacted with metropolitan congestion, yield positive PMT gains. In addition, both higher-wage and leisure jobs are associ- ated with higher system-wide PMT. Leisure jobs—those held by workers in retail, food, accommodation, entertainment, arts, and sports—may represent workers who commute on fixed- guideway transit, but the measure may also capture the impact of activity centers and dense, transit-friendly destinations often found in large cities that are not readily measured with variables such as mixed-use entropy indexes and walk scores, neither of which were statistically significant in our testing. Near fixed- guideway stations, high-wage jobs may cause a system-wide boost in PMT if those workers are less likely to use bus services but are willing to patronize new fixed-guideway service. The size of the region (expressed as the population) and the FHWA congestion index alone are not statistically significant, but they are included in the model to control interacting variables. Table 5.5 displays three models—two final specifications and one alternative. The first model, referred to as the catchment- Variable Name Abbreviation Definition Jobs near stationsa JobsCatch Jobs within 1/2 mile of system stations Population near stationsa PopCatch Pop. within 1/2 mile of system stations Leisure jobs near stationsa LeisJobsCatch Number of retail, food, accommodation, entertainment, arts and sports jobs within 1/2 mile of system stations High-wage jobs near stationsa HWJobsCatch Number of jobs earning more than $3,333 per month within 1/2 mile of system stations FHWA congestion index Cong Avg. weekday VMT/freeway lane- mile in metropolitan area PMT interaction term IntPMT (S_Jobs × S_Pop × Cong)/(1 billion) MSA jobs JobsMSA Total MSA employment MSA high-wage jobs HWJobsMSA Number of jobs earning more than $3,333 per month in MSA MSA leisure jobs LeisJobsMSA Number of retail, food, accommodation, entertainment, arts and sports jobs in MSA MSA population PopMSA Population of MSA (2000 Census) a Measured within 1/2 mile of all fixed-guideway rail stations in the region, excluding commuter rail stations. Table 5.4. Summary of variables in final PMT model.

2-36 it may also be likely that they already do so. As has been dis- cussed, higher-income workers are less likely to choose to ride a city bus, but may find train or BRT service more appealing. Therefore, fixed-guideway alignments serving higher-income workers might be more likely to add PMT to the system. The study team also tested whether the number of units of rental housing and/or the number of office jobs near stations was associated with PMT, but the results did not improve the model. Alternative measures of the utility of driving were tested, including the number of freeway and arterial lane-miles, lane- miles of each type per square mile, per capita length of freeways and arterials, and the year-to-year change in these values, gas Before settling on the final models, many alternative approaches were tested. The inclusion of low-wage jobs was tested because low-wage workers are generally more likely to use transit. Unexpectedly, low-wage employment—tested only for catchments, not for the full MSA—was found to have a significant negative effect on PMT, though the model fit was not much improved. This result may be because low-wage employment indicates declining economic fortunes more than the presence of potential transit riders. Another possible reason for the result is that high-wage employment may bet- ter reflect added transit ridership than low-wage employment when new fixed-guideway transit lines come online. Although workers making a lower wage are more likely to ride transit, Variable Name Final Census Catchment-Level MSA-Level MSA Variables Catchment jobs -2.542*** -2.608*** -2.212*** Catchment population -0.223___ -0.202___ -0.661*** Catchment leisure jobs 8.441*** 8.299*** 7.412*** Catchment high-wage jobs 3.279*** 3.464*** 3.157*** FHWA congestion index -1.088___ -1.282*__ -1.123* PMT interaction term 0.061*** 0.056*** 0.048*** MSA jobs 0.120*__ -0.322*** MSA high-wage jobs -0.076___ 0.486*** MSA leisure jobs 0.355___ 0.189 MSA population (U.S. Census) 0.273*** MSA population (BEA) 0.147*** 0.115*** Constant -18,977.000___ -29,783.5*____ -64,450.4 # of observations 1,641 1,641 1,641 Cluster-specific variance 145,053.9***__ 141,380.8***__ 147,803.0*** Other variance 14,624.4***__ 14,531.2***__ 13,129.8*** BIC score 37,789.2_____ 37,781.0_____ 37,519.3 *p < 0.05, ***p < 0.001, BIC = Bayesian information criterion. Table 5.5. Metropolitan-level PMT models. 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1 Catchment jobs MSA jobs Catchment populaon Congeson score MSA leisure jobs Interacon (jobs×pop×congeson) MSA high wealth jobs Catchment leisure jobs Catchment high wealth jobs MSA populaon β Value Figure 5.6. Beta values for final PMT model (MSA level).

2-37 If it is assumed that increasing metropolitan population and an improving economy over those 5 years should have naturally resulted in increased bus service, bus service might hypothetically be allowed to stagnate when investments in rail were made. The data do not, however, strongly support the diversion of service hypothesis. 5.3.2 Increasing Efficiency Although data were not available to test this hypothesis, it is possible that rail investments decrease the miles traveled on transit for some passengers, resulting in a negative overall incre- ment. Rail lines can sometimes provide more direct routes than bus service. Furthermore, whereas a bus rider might be willing to make multiple transfers to board and alight close to the indi- vidual’s origin and destination, fixed-guideway systems encour- age using alternate modes (driving, walking, or bicycling) to access stations farther from home. Both of these effects may decrease overall PMT. On the other hand, rail lines also allow passengers to travel farther than bus in a given time, so that might imply that travel distances would actually get longer with increased mobility. The estimates of incremental PMT for each project in the study’s database were calculated as the difference between the model predictions for the entire metropolitan area with and without the project. The incremental difference is sometimes on the same order of magnitude as the error in the estimate of overall PMT. Some small negative increments are, in fact, within the margin of error for the model, implying that the project has no statistically significant effect on system-wide PMT. The spreadsheet tool reflects this by reassigning small negative increments of PMT to zero. The same issue applies to small positive increments as well, although these values are not reassigned in the spreadsheet tool. 5.4 Input from Focus Groups: Phase 2 Development of a spreadsheet tool to make it possible for transit agencies or other interested parties to estimate rider- ship on proposed fixed-guideway transit projects and the proj- ects’ impacts on system-wide PMT was an element of TCRP Project H-42. The spreadsheet tool provides a simple way to apply the indicator-based models to compare corridors, align- ments, and modes in terms of these success measures. During the second round of focus groups and interviews, participants were shown an initial mock-up of the spread- sheet tool and invited to comment on its utility for local plan- ning. Many participants said that the tool would be useful for an initial evaluation of potential transit projects, helping prioritize alternatives, providing a means for scenario testing, and demonstrating to the community the implications of dif- ferent options. Comments were offered suggesting additional prices, parking costs, and congestion measures. Testing these variables in the final model demonstrated that the researchers’ congestion score (based on FHWA data on freeway use per lane- mile) was the most significant measure of automobile utility. The research team used lagged variables to test the signifi- cance of the relationship between changes in a given year and the resulting impact in subsequent years. Instead of using inputs such as population, employment, ridership, and so forth for the year when PMT is measured, the study team tested val- ues lagged by 1, 2, or 3 years to see whether there was evidence of the investment maturing and ridership stabilizing. The results did not change very much, likely because a longer lag (e.g., 10 years) may be the likely period over which ridership effects are felt. Testing lags longer than 3 years are not possible with existing data. 5.3 Estimating Uncertainty in Model Outputs The model outputs are based on a fit to data that shows natu- ral variation, or scatter. Even if the measurements are assumed to be free of error and the functional form to be correct, uncer- tainty remains associated with the fit. The uncertainty in the estimated coefficients is summarized in the variance-covariance matrix. For both the project-level ridership models and the metropolitan-level PMT models, the variance-covariance matrix generated during the modeling process is used to allow the user to estimate the error in predicted transit use for the proposed project. Uncertainty in the incremental PMT esti- mates is about 20 percent on average, with smaller projects typically having less certainty as a percentage of the estimate. Because the negative increments were a constant through- out multiple tests of the model’s robustness and several alter- native data sets, the researchers concluded that they are legitimate findings. Possible explanations include diversion of service and increasing efficiency. 5.3.1 Diversion of Service It is possible that agencies may pay for rail investments in part by diverting resources from existing services, or that they respond to new rail lines by downgrading or closing parallel bus routes. The resulting drop in PMT might over- whelm the added miles from the rail investment. To test this, the research team observed the changes in bus and rail seats provided per capita over the MSAs in the sam- ple set (from the National Transportation Database, or NTD). MSAs with increasing rail seats between 2002 and 2007 showed a decrease in the number of bus seats, whereas MSAs with no increase in rail seats showed an increase in the number of bus seats; however, these changes were on average so small as to be negligible.

2-38 Telephone interviews were also conducted with three indi- viduals who expressed interest during the mid-year meeting of TRB’s Metropolitan Policy, Planning and Processes Commit- tee (ADA20): Mary Archer, Marin County (California) Transit; Elizabeth Schuh, Chicago Metropolitan Agency for Planning; and Tom Schwetz, Lane County (Oregon) Transit. The focus groups and the telephone interviews started with a PowerPoint presentation through which the research team summarized the research goals, introduced the analytical approach, and presented a preliminary mock-up of the spread- sheet tool’s input and output screens. In the focus groups, participants were particularly interested in seeing how their projects compared with others in the database. Participants in the APTA rail focus group were generally more interested in the policy/planning implications of the find- ings than in potential applications of the model. For instance, in an initial version of the model, they wanted to know why the radius for calculating station-area density would differ for residential and employment development. At that time the best predictors were ½ mile for population near stations and ¼ mile for employment. One participant noted that his agency had considered locating a station near an office campus but did not do so because the campus density was not “transit oriented.” Knowing the importance of jobs within a quarter-mile of stations, he said that they might have fought harder to put the station close to the corpo- rate campus, or to have planned for branch lines serving that center and other nearby job centers. Participants also showed interest in the mode-neutral nature of the model, while noting that FTA now allows agencies to factor in the added appeal of rail. Participants in one of the focus groups said that the traditional approach to New Starts planning emphasized finding solutions to a current or future transportation problem. One participant wondered if a focus on project- level ridership signaled a movement away from problem- solving. This participant pointed out that the tool does not tell the user whether or not a specific problem has been solved, nor does it address whether conditions for existing riders have been improved. The participant also noted that his agency has used land use density thresholds as a success indicator for proposed rail stations, giving them a lever for influencing projects before they receive formal submissions for consideration. A participant in another focus group also questioned whether the spreadsheet tool implied that projects with a lower cost per rider were by definition more successful. This participant gave an example of a project that is considered to be a local success, despite being at the high end of the cost per rider range, because fare revenues from that project exceed operating costs. Although cost per rider might be useful as part of a multicriteria project evaluation considering return capabilities that would make the tool more useful, includ- ing improved visualization of the tool’s outputs in the form of charts and tables. Overall, participants expressed no sig- nificant concerns about the difficulty of generating the input data for the tool, apart from the time required. Many partici- pants requested that the handbook provide clear and detailed information about the underlying mechanisms of the spread- sheet tool to give them more confidence in the validity of its results and to better explain the tool’s outcomes to interested parties. More information about the second round of focus groups and interviews appears in Appendix H. In Phase 2, the research team made efforts to broaden the outreach to include MPOs and both large and small tran- sit agencies. One focus group was held during APTA’s June 2012 Rail Conference in Dallas, which tends to attract tran- sit agency professionals. As with the first focus group, par- ticipants were selected from among the conference attendees based on their knowledge of transit project evaluation and their leadership roles within APTA. To broaden the partici- pation, representatives of the local MPO also were invited. Participants included: • Doug Allen, deputy general manager for planning and development, Capital Metro, Austin, Texas, and vice chair of APTA’s policy and planning committee • Matt Sibul, director of planning for the Utah Transit Authority, Salt Lake City, Utah • Barb Weigle, planner with Dallas Area Rapid Transit, Dallas, Texas • Kay Shelton, planner with Dallas Area Rapid Transit, Dallas, Texas • Chad Edwards, planner with North Central Texas Council of Governments, Dallas, Texas • Hua Yang, planner with North Central Texas Council of Governments, Dallas, Texas • Cheryl King, planning director for the Metropolitan Atlanta Rapid Transit Authority, Atlanta, Georgia • Kim Slaughter, planning director for Houston Metro, Houston, Texas Two smaller focus group discussions were subsequently held to solicit broader input. The first was held on June 13 at the Houston-Galveston Area Council (HGAC), the MPO serving the Houston, Texas, region. Participants included Ashby Johnson, deputy director of planning, and four mem- bers of the HGAC planning staff. The second small focus group was held July 18 at the MTC, the MPO serving the San Francisco Bay area. Participants included David Ory, head of the MTC’s systems analysis group; Carolyn Clevenger, a senior planner with the MTC; and Dave Vauten, an MTC staff planner.

2-39 5.5.1 CBD Employment Earlier work identified jobs in the CBD as an important indicator of transit use. This analysis uses jobs located within ½ mile of stations to predict transit use at both the project and the metropolitan level. To test whether CBD jobs have a dis- tinct and separate impact on ridership and PMT, the research- ers defined CBD jobs in three ways. First, the researchers nominated one station in each metropolitan area as the central station and counted the number of jobs within ½ mile of sta- tions that are themselves within ½ mile or less of that central station. The second definition expanded the size of the CBD to include all stations within 1 mile of transit distance to the central station. The third definition used catchment jobs for stations that fall inside the CBD as defined by the U.S. Census Bureau in 1982. As Table 5.6 shows, only one of the CBD catchment variables—residents near stations within ½ mile of the central station—was significant in the project-level ridership model. It shows a counterintuitive negative contribution. The model’s overall goodness of fit does not change significantly, so the research team retained the more parsimonious model without CBD measures. At the metropolitan level, the CBD catchment variables likewise do not improve the model. As shown in Table 5.7, only the third definition improves the fit significantly, but the CBD variables are still statistically insignificant. Model D shows a model in which only CBD station areas are counted, and the fit is distinctly worse. This result does not necessarily indicate that CBD employ- ment is irrelevant as an influence on ridership and PMT. Rather, it indicates that the inclusion of jobs located near sta- tions allows for the greater job density of the CBD to be incor- porated in the model, because the CBD contains stations. CBD employment is not independently a significant influence on ridership or PMT. 5.5.2 Exclusion of Renovation Projects Of the 72 projects used in the study set, six involved reha- bilitating existing stations, five of which were included in the estimation data set. Three of these projects were in Chicago (Metra North Central, Metra Southwest Corridor, and CTA Douglas Branch); one was in Miami (South Florida Tri-Rail Upgrades); and one was in Philadelphia (SEPTA Frankford Rehabilitation). The Tri-Rail and Frankford Projects were classified as enhancement projects in the study’s data set, indicating that they consisted solely of the improvement of existing infrastructure. An additional project in Chicago— the Metra UP West, was excluded from the final data set because of incomplete data. To test the impact that including these projects had on the ridership models, the researchers compared the final on investment and other measures, some might consider a focus on cost per rider alone to be incendiary. The participant suggested that the handbook explain how the user should interpret and use this information as part of a multicriteria project evaluation. This suggestion was echoed by one of the telephone inter- viewees, who pointed out that cost per rider needed to be considered in context, such as the cost of other alternatives. This interviewee stated that a variety of factors enter into local decisions on a transit project—impacts on business, property values, parking, vegetation, traffic volumes, emissions, jobs— and suggested that the researchers consider how to estimate some of these other factors based on ridership results, or at least point users to other sources they might use to estimate a fuller array of factors. Participants in the APTA rail focus group also suggested other variables that might be considered in the regression analyses: • Regional and/or corridor characteristics – Number of CBDs – Quality of pedestrian access to stations • Characteristics of the project – Extension or new line – Does project provide a one-seat ride to the CBD, or is transfer required? • Special events – Seats at sports venues within ¼ mile of stations, number of events, attendance – Convention center size – Hotel rooms within ¼ mile • Key trip generators – Hospital beds in catchment area – Commuting students in catchment area • Weather – Days of sunshine – Inches of precipitation, snow Focus group attendees added that the tool should consider the walkability of pedestrian access to stations, not merely the distance. They also recommended that the research team try to expand the tool to include CR. An interviewee suggested that the research team consider an operations and maintenance cost output as well as capital cost. 5.5 Response to Practitioner Input Many additional variables, including walkability, entropy indexes, and specific industry types, were tested and cre- ated in response to input from focus group participants and case study interviewees. None of these additional measures improved the models significantly, but a few of the factors are discussed in more detail in this section.

2-40 Variable Name Final Model CBD Test Models Defensible ½-mile 1-mile Census-Defined Catchment jobs 0.155 0.280* 0.307** 0.308* Catchment population -0.014 -0.0888 -0.0795 -0.0585 CBD parking rate -491.9 -387.8 -386.7 -247.6 Ridership interaction term 0.0785*** 0.0815*** 0.0752*** 0.0654*** Percent at grade -17,846.2* -92.7 -516.2 -1,111.5 Missing at-grade dummy 3,294.9 -20,899.9*** -21,115.9** -19,099.8** Project age 913.4* 968.8** 953.3** 742.7* ½-mile CBD jobs -0.0532 ½-mile CBD population -1.664* 1-mile CBD jobs -0.126 1-mile CBD population -0.795 Census-defined CBD jobs -0.0932 Census-defined CBD population -0.89 CONSTANT 20,672.7** 21,763.0** 21,278.0** 17,896.9* # of observations 55 55 55 55 Adjusted R2 0.894 0.911 0.906 0.902 *p < 0.05, **p < 0.01, ***p < 0.001 Table 5.6. Comparison of project-level models that include CBD catchment variables. Variable Name Final Model CBD Test Models Defensible ½-mile 1-mile Census-Defined 1 Census-Defined 2 CSA population 0.0161 0.0202 0.0188 -0.0143 -0.0228 Catchment jobs -2.742*** -3.017*** -2.836*** -3.500*** Catchment population -1.305** -1.201** -1.067* -0.938* Catchment leisure jobs 8.816*** 9.505*** 8.927*** 8.655*** 3.031 Catchment high-wage jobs 4.486*** 4.442*** 4.415*** 4.524*** 3.559*** Congestion index -5.001 -3.478 -4.502 -11.42 0.733 PMT interaction term 0.0633*** 0.0691*** 0.0552** 0.0748*** -0.0231 ½-mile CBD jobs 0.441 ½-mile CBD population 35.58 1-mile CBD jobs 0.492 1-mile CBD population -10.41 Census-defined CBD jobs 0.753 -0.976 Census-defined CBD population -3.186 -12.02 Constant 798,410.7** 375,261 909,602.0** 1,099,246.7*** 714,128.9* # of observations 124 124 124 110 110 Adjusted R2 0.714 0.717 0.711 0.752 0.633 *p < 0.05, **p < 0.01, ***p < 0.001 Table 5.7. Comparison of metropolitan-level models with CBD catchment variables.

2-41 each of which is associated with a set of codes developed for the North American Industry Classification System (NAICS). The researchers used an entropy index (Cervero & Kockelman 1997) to estimate the balance within the subset of jobs listed in Table 5.9. The entropy index can be expressed as follows: ln ln 3 ln ln 3 ln ln 3 (5)E P P P P P PA A B B I I∑= − × + × + × PA is the fraction of jobs in the catchment that belong in group A, PB is the fraction jobs in the catchment that belong in group B, and PI is the fraction jobs in the catchment that belong in group I. Perfect balance between the three groups would yield an entropy index of 1. This was done at the proj- ect and metropolitan levels. In both cases, the job diversity measure did not improve the goodness of fit; thus, it was not included in the final models. The job diversity index is also a means of addressing a set of variables considered to be important by practitioners. Some agencies stated that they found the presence of dry cleaners, defensible model with three test alternatives (see Table 5.8) as follows: 1. Adding a dummy variable for projects including sta- tion rehabilitations (the column marked Rehab. Stations Dummy). 2. Adding a dummy variable for projects classified as enhance- ments (the column marked Enhancement Dummy). 3. Removing rehabilitation and renovation projects from the estimation data set (the column marked Rehab. Projects Removed). Job diversity and the importance of retail jobs were sug- gested as important indicators of transit use during conversa- tions with several regional planners in the case study phase. The modeling process in TCRP Project H-42 used retail jobs as one component of the leisure-based jobs category when predicting PMT. To measure job diversity, the researchers created a success metric that is designed to measure the bal- ance of jobs that are associated with non-work activities. As shown in Table 5.9, three groups of activities were created, Variable Name Final Model Renovation Test Models Defensible Rehab. Stations Dummy Enhancement Dummy Rehab. Projects Removed Catchment jobs 0.155 0.163* 0.165* 0.167* Catchment population -0.014 -0.0475 -0.0424 -0.0504 CBD parking rate -491.9 -594.0* -386.5 -479.9 Ridership interaction term 0.0785*** 0.0818*** 0.0794*** 0.0808*** Percent at grade -17,846.2* -14,536.9* -14,548.6* -13,646.1* Missing at-grade dummy 3,294.9 -9,217.7 -2,549.8 -7,367.6** Project age 913.4* 1,040.1** 1,038.9** 1,065.5** Rehab. stations dummy 19,567.2 Enhancement dummy 27,200.9*** CONSTANT 20,672.7** 18,794.0** 15,755.9* 16159.9* # of observations 55 55 55 50 Adjusted R2 0.894 0.901 0.907 0.908 *p < 0.05, **p < 0.01, ***p < 0.001 Table 5.8. Comparison of project-level models with rehabilitation and enhancement project adjustments. Label Activities Codes A Shopping, beauty salons, mechanics, laundry, religious activities, restaurants, and banks NAICS codes 44/45, 81, 72, 52 B Real estate, lawyers, accountants, notaries, arts, entertainment, recreation NAICS codes 53, 54, 71 I Schools, doctors, dentists and hospitals, public services NAICS codes 61, 62, 92 NAICS = North American Industry Classification System. Table 5.9. Definition of groupings for non-work employment analysis.

2-42 thing which the team’s review of prior research suggested has not been done systematically before. Similar to the original work of Pushkarev and Zupan as well as subsequent studies, the researchers for this study found that population density and employment density were highly pre- dictive of transit ridership; unlike those studies, however, the researchers found that the combination of jobs, residents, and high parking cost or high road congestion is much more influ- ential than any of those indicators on their own. Additional indicators that this study shares with other recent research include income measures, measures of network configuration, service frequency, local bus connections, and park-and-ride spaces. The research team found some often-cited predictors of success to be insignificant when controlling for other factors. Insignificant factors include population characteristics such as education level, immigrant status, renter status, and car own- ership; service characteristics such as fare, frequency, revenue vehicle-miles, and speed; average station distance to the CBD; transit network service coverage; weather measures; and fuel price. The study described in this report does not investigate other cited indicators (e.g., trip destination type) or street net- work design characteristics (e.g., intersection and street den- sity and percent of intersections that are 4-way intersections, an indicator of both density and connectivity). Although this study differentiates by mode and finds that mode-specific dummy variables are not significant, the project-level data set is too small to enable a test of how the influence of indicators like jobs and residents near stations may vary by mode (HRT, LRT, and BRT). Conversely, one element that is incorporated into the model presented in TCRP Report 167 but often is excluded from other studies is the relative costs of transit versus pri- vate automobile—namely congestion measures and parking prices and availability. This study finds that high congestion is a significant indicator of transit use in conjunction with concentrations of jobs and residents near stations, but not by itself. drug stores, flower shops, and other specific retail services to be important. In TCRP Project H-42 the data were not detailed enough to examine each of these niches individually; however, the larger industry categories into which they fall were each tested, and each was found not to be statistically significant. Certain institutions and facilities are also considered to be important indicators of transit use. Specifically mentioned in focus groups and case studies were schools (quantified by number of students or desks), hospitals (quantified by num- ber of hospital beds), and stadiums (quantified by number of seats). Although the presence of one of these institutions may have a significant impact on ridership, the researchers were unable to gather reliable data across the panel, and therefore these indicators were not tested. Weather and climate were suggested at a case study visit and subsequently included. The researchers tested various metrics from the NCDC, including number of days with high or low temperatures, percent of possible sunlight, and average tem- perature, precipitation, and snowfall. When controlling for other factors, these weather characteristics were not significant predictors of transit use at the project or metropolitan level. 5.6 Summary of Results and Comparison with Previous Studies The analysis in this study used aggregate demand models to investigate the impact of numerous indicators on rider- ship and PMT for the largest possible set of fixed-guideway projects, incorporating cross-sectional and time-series data. Other indicator-based research has similarly attempted to predict the success of transit lines using causal models that incorporate numerous characteristics of the system itself and surrounding conditions. The work for this project examines almost all of the same indicators as those other causal studies, and includes some that others ignore. This study also tested system-wide impacts of fixed-guideway investments, some-

2-43 C H A P T E R 6 The research team conducted case studies of transit proj- ects in six metropolitan areas, reviewing public reports and other materials, conducting site visits, and interviewing more than 50 transit planners, MPO officials, and consultants who had worked on the projects. The cases were used to help the researchers understand how transit planning decisions had been made and the nature of any indicator-based evaluations that had occurred. This included determining what measures of transit project success were considered by planners and how those success measures might have predicted the future performance of a project. The study team also discussed TCRP Project H-42’s proposed indicator-based method to understand how it might be used by planners, and what improvements would be helpful. This section summarizes the findings from the case studies. Detailed write-ups are avail- able in Appendix I. The study team focused on HRT, LRT, and BRT projects because the project’s data set has few commuter rail (CR) examples and no streetcar examples. For three of the case study sites, the researchers analyzed the planning and per- formance of CR projects in the same region, drawing conclu- sions about the indicator-based method that may not apply to the HRT, LRT, and BRT cases. The cases were as follows: • Lynx South Line (Charlotte, North Carolina) • North Central DART Extension (Dallas, Texas) • Emerald Express BRT line (Eugene, Oregon) • Interstate MAX (Portland, Oregon) • University and Medical Center Extensions (Salt Lake City, Utah) • Metro Branch Avenue Extension (Washington, D.C.) The research team sought out diverse cases in order to iden- tify differences between projects and to better understand the potential implementation of the proposed indicator-based method across different project types (see Case Study Diversity). The team also ensured that it had access to project planning documentation and that planners were available for interviews. Finally, the researchers looked for outlier cases, such as Port- land’s Interstate MAX and the Branch Avenue Extension of WMATA’s Green Line. Interstate MAX was a below-average project, based on net cost per passenger-mile ratios (Guerra and Cervero 2011), in a region lauded for its high-quality tran- sit planning. By contrast, the Green Line was one of the most cost-effective projects (Guerra and Cervero 2011). Case Study Diversity Modal: One HRT project, four LRT projects, and one BRT project (with three CR lines in the same metropolitan areas also discussed). Geographic: Two Pacific, one Mountain West, one Sunbelt, and two East Coast. Metropolitan area size: Ranging from 350,000 to 6 million residents. Project context: Two cases were the first fixed- guideway projects in a region and four cases were system expansion projects. Transit funding: Four projects received federal New Starts funding, and five received some form of direct federal support. No two projects had the same mix of project funding sources. Stakeholders: Three of the projects passed through multiple jurisdictions. Plans for two of the projects crossed state lines. The research team visited the case study projects, met with transit agency and MPO staff, and reviewed documents archived by the project sponsors. Transit project consultants were also interviewed. During the site visit in Washington, D.C., 2 days Case Studies: Overview

2-44 were spent at FTA reviewing document archives and speaking with staff. Although case study settings and case study projects varied considerably, very similar indicator-based planning practices emerged across all the cases. Qualitative indicators and mea- sures of success were often given more credence in decision- making than quantitative indicators and technically derived measures. Because the indicator-based method developed as part of this project focused on ridership and PMT and did not include other success measures, many interviewees believed the use of the quantitative method would be limited, and that changes and additions would be necessary for the tool to be more useful. 6.1 Settings Differences in case study settings may have contributed to the different ways projects were evaluated and planning decisions were made. One element that differentiates these case studies is distinct cultural and physical transit orienta- tion. The Dallas project, Washington, D.C., project, and the southern portion of the Charlotte project were planned in settings that were auto-oriented at the time, which contrib- uted to their auto-oriented features. Eugene, Portland, and Salt Lake City were built, at least partly, in arterial medians close to the cores of urban areas. These settings influenced the planning philosophies that guided the projects. Whereas planners in Washington, D.C., and Dallas, Texas, tended to consider potential patrons of their rail system extensions to be park-and-ride users or bus riders, planners in Char- lotte, North Carolina, and Eugene, Oregon, thought of their patrons as arriving by a mix of modes. In Portland, Oregon, planners suggested that the average rider they planned for was a pedestrian (Interviewee AA, in-person conversation, 8/7/12). Macro regional factors also play a role in how transit proj- ects perform. For instance, the Dallas North Central Line failed to achieve ridership projections, whereas the Wash- ington, D.C., Branch Avenue Extension met projections. The Dallas and Washington, D.C., regions have similar total population and employment, but Washington, D.C., now has four times the transit route-miles and four times the number of people living near its rail system stations. Parking rates in the CBD, an indicator of supply-and-demand dynamics, are three times as high in D.C. as in Dallas—in fact Dallas has one of the lowest average CBD parking rates of any major city. These factors may help explain the fivefold difference in regional transit ridership between the two locations. In spite of differences in the settings of these case stud- ies, their transit planning processes were similar. Several fac- tors that could be expected to produce different approaches among the case studies did not seem to play a role, including the mix of transit modes and the size of the metropolitan area. Federal environmental policy and funding require- ments may have led to a consistent transit planning process, as well as the use of consulting firms. If stakeholder agencies lacked fixed-guideway planning capacity, as was the case in Charlotte where no light rail previously existed, consulting firms were engaged to lead the planning. Even in locations where fixed-guideway projects had previously been devel- oped, consultants were consistently retained to aid with planning. 6.2 Project Attributes Differing project characteristics may have contributed to the ways they were evaluated and how planning decisions were made. The interviews conducted in TCRP Project H-42 suggest that transit planners carry out system planning and project planning differently—primarily varying the priori- tization of the decision criteria they consider—depending on those characteristics. Motivations for transit projects and typologies of transit facilities varied greatly. For example, some projects were motivated largely by a desire to support changes to the regional urban form and land use patterns (Charlotte, Eugene, and Portland), but others were moti- vated by automobile traffic mitigation or mobility concerns (Washington, D.C.; Dallas; and Salt Lake City). This led plan- ners to use different prioritization of success measures when considering modes, alternatives, station locations, and other project attributes. At the same time, some of the cases were primarily envisioned as walk-up services (Portland, Eugene, and Salt Lake City), whereas others were envisioned as park- and-ride facilities early in the planning process (Washing- ton, D.C., and Dallas). Again, the expected role of the project informed the prioritization of various measures of success and, therefore, success criteria used by planners during the planning process. Variance in planning philosophies and characterizations of projects might be partly attributable to the timing of the plan- ning processes. Most of the case study projects were added to regional plans during the 1980s, and planning was carried out in the 1990s and 2000s. These more recent planning processes strongly considered land use impacts, economic development at each station, and other current-day concerns. However, the case studied in Washington, D.C., although opened in 2001, was actually planned in the 1950s when priorities focused on decongesting central cities and facilitating travel between rap- idly expanding suburbs and the CBD. Transit planning has changed over time, based on changing values and advances in the state of the art. Different priority was given to success metrics depend- ing on project differences, but the researchers were struck by how similar the success metrics, evaluation techniques, and

2-45 planning processes actually were across projects that var- ied by size, mode, and other features. All of the cases used very similar indicator-based methods to develop early tran- sit plans and to quickly assess the potential for various pro- posals to be successful. Additionally, qualitative indicators and measures were often given more credence in decision- making than quantitative indicators and technically derived measures. Some of these similarities may be explained by the same nationwide factors that were enumerated in the prior section, including adherence to federal policies by national consulting firms. 6.3 Indicator-Based Planning Methods The case study research suggests that TCRP Project H-42’s indicator-based method will be situated within an already- robust set of indicator-based transit planning methodolo- gies. Several transportation planning agencies noted that their regions have recently employed robust indicator- based transit project prioritization methodologies (Inter- viewee AB, in-person conversation, 8/7/12; Interviewee AC, in-person conversation, 8/20/12). During the planning of every case study project, various kinds of indicator-based methods—some heuristic, some empirical—were used to propose transit alignments, compare and contrast project alternatives, and justify the selection of a particular pro- posal, and typically included goals in addition to ridership and capital cost. Multiple interviewees stated that transit planning is an art and a political process, not a science (Interviewee AD, telephone conversation, 8/24/12; Interviewee AA, in-person conversation, 8/7/12). Project stakeholders invariably dis- cussed the need to balance multiple objectives beyond rider- ship and capital cost as they planned transit projects. Notably, the goals associated with implementing the six case study tran- sit projects were consistent across all the projects, although they were prioritized differently. Those goals are listed in Table 6.1. The transit planning literature often focuses on predicting project success based on specific technical planning approaches and sophisticated planning tools, such as four-step transpor- tation models; however, the research team identified nearly 20 different simple criteria being used by planners to predict whether a transit project would be successful according to one or more of the goals enumerated above. These criteria can be described as rule-of-thumb procedures for predicting project success and making determinations about route options or alternative station locations (Table 6.2). Considering that most technical approaches—even the proposed indicator-based method—require users to describe a transit project before an evaluation can be made, it is obvi- ous that less-technical methods were employed to develop the test cases. Across every case study, transit planning deci- sions seemed to rely more on the rules of thumb than on the outputs of the technical evaluations. Though not technically complex, the rule-of-thumb meth- ods helped transit planners address the immense complexity of designing and building a transit project. The case stud- ies illustrate several balancing acts among various interest groups, among conflicting objectives, and between technical analysis and heuristic evaluations. Measure of Project Success Example Metrics Evaluated Before Operations Abbreviation Ridership Modeled riders per day, riders per day per station, and riders per mile R Sustainability Modeled mode shift (i.e., choice ridership), VMT, air quality (particulate matter) S Real estate impacts Projects proposed during transit planning, billions of dollars in private real estate investment since stations were announced RE Economic development Qualitatively assessed through anecdotes, case studies, and business community’s advocacy (also see real estate impacts) ED Consolidated bus operations Modeled operating costs BUS Congestion relief Modeled hours of congestion on parallel roadways C Project completion Passed local, regional, and state votes; completed federal process steps; won funding; set project delivery date; opened for revenue service PC Dependent riders Non-auto households in proposed station areas, low-income households in proposed station areas DR All goals were observed in each case study. Table 6.1. Project goals discussed by project stakeholders.

2-46 successful (Interviewee AA, in-person conversation, 8/7/12). In Charlotte, planners found that the Bush Administration’s singular focus on cost effectiveness based on cost per hour of travel-time savings led them to make cost-saving changes to their project that ultimately produced a short-sighted invest- ment (Interviewee AD, telephone conversation, 8/24/12). Although fully aware that their long-range planning models predicted that future light rail extensions would require ser- vice on the line to use longer trains, Charlotte reduced the capital cost of their initial light rail facility by limiting station platform sizes to those required to accommodate initial rid- ership demand. Subsequently, when extending the rail line, they were forced to disrupt operations to lengthen the plat- forms on the existing segment. Project-level and system-level patronage were seen as dis- tinct by some interviewees. Interviewees in Salt Lake City stated that some of their latest projects have not achieved the ridership they anticipated, but system-wide ridership gains have resulted from operational changes on the trunk 6.4 Potential Usefulness of the TCRP Project H-42 Method The study team demonstrated the preliminary spreadsheet tool and discussed how interviewees might employ it. One objective of TCRP Project H-42 is to “identify conditions and characteristics that are necessary to support alternate fixed- guideway transit system investments.” As noted in the prior section, the case studies suggest that transit planners balance numerous objectives for which certain “conditions and char- acteristics” are relevant to some planners but uncorrelated with or counterproductive to others. Although ridership was universally regarded as a measure of transit project success, it was one of many success measures under consideration. In Portland, transit planners implemented a project to sat- isfy other objectives of the transit agency, local governments, community members, and other stakeholders, despite the fact that according to several quantitative measures of suc- cess the project was expected to have low ridership and be less Criterion (Rule-of-Thumb) Measure of Project Success Ch ar lo tte D al la s Eu ge ne Po rt la nd Sa lt La ke C ity D .C ./M D Provide fixed-guideway transit where bus ridership is already high R / BUS X X X X X Select high-visibility corridors where patrons will feel safe R X Connect CBD with suburban park and rides near a congested belt loop R / S / C / BUS X X X Minimize stations to maximize speed R / S / C X X Minimize grade crossings and in-street operations to maximize speed R / S / C X X X X X Provide fixed-guideway transit in corridors where parallel highway infrastructure is heavily congested R / S / C X X X Connect multiple employment centers R / S / C X X X X Connect major regional destinations R / ED X X X Place alignment in close proximity to commercial property R / ED X X Place stations in busy locations where “eyes on the street” provide a sense of safety R X Provide transit in high-demand travel corridors where alternate capacity is prohibitively expensive ED X X X X Maximize the number of stations ED, RE X X X X Place alignment along corridors with ample development potential to facilitate urban growth as described by local land use plans or regional plans RE X X X X Provide fixed-guideway transit in corridors where inexpensive right-of-way can be easily accessed PC X X X X X X Maximize distance between alignment and single family neighborhoods; minimize taking of residential property PC X X X X Identify corridors that can help garner local political support for further transit system investment PC X X X Select corridors that garner congressional support PC X X X Locate stations in low-income areas or in communities of color DR / PC / ED X X X Provide service that has average travel speeds greater than existing bus routes R / BUS X X X X Provide substantial bus layover facilities at stations BUS X X X Table 6.2. Criteria discussed by project stakeholders.

2-47 based on regressions of national data points and stated that they would likely rely instead on locally calibrated regional models (Interviewee AF, in-person conversation, 8/13/12; Interviewee AL, in-person conversation, 8/7/12). In one instance, modelers felt the lack of mode specification was problematic because their local research found significant differences in perceived wait times for various transit modes (Interviewee AM, in-person conversation, 8/7/12). Other modelers suggested that they would likely use the model and share the results with other staff and with board members as supplemental evidence if the results corroborated their opinions and the regional model (Interviewee AG, in-person conversation, 8/30/12). Stakeholders also noted that the tool and the existing regional models suffered similar issues related to their granu- larity. For example, one interviewee noted that the method did not address local street grids, station-area aesthetics, and other factors that were considered influential in transit plan- ning decision-making (Interviewee AK, in-person conver- sation, 8/30/12). Planners tend to rely on rules of thumb for these matters, even when there are conflicting views. For exam- ple, Charlotte transit planners argued that the line should limit the number of at-grade roadway crossings (calling them con- flict points) because they would slow operations and detract from the appeal of the service. Simultaneously, Charlotte land use planners prioritized keeping the rail alignment at grade and maximizing the density of local roadways near stations to promote connectivity and attractive urban form. They believed this would allow more people to physically reach the stations and would overcome any psychologi- cal impediment to service access that might be caused by grade separation (Interviewee AN, in-person conversation, 8/30/12). Some interviewees suggested that the method would be a helpful tool for specific circumstances when expected rider- ship was considered in a non-technical manner. For exam- ple, one interviewee thought the model might be usable to quickly compare potential projects within a regional sys- tem plan to produce an initial prioritization of projects for review by elected officials (Interviewee AP, in-person conver- sation, 8/13/12). This interviewee saw this use of the model as a low-risk situation, because elected officials typically ignored staff ’s technical prioritizations unless they supported their position. In half of the case study cities, transit planners thought the tool could be given to citizens and local officials who were demanding obviously infeasible rail projects, pro- viding those constituents with clear evidence of the short- comings of such projects without requiring more complex (and costly) regional modeling exercises (Interviewee AQ, in-person conversation, 8/08/12; Interviewee AM, in-person conversation, 8/7/12; Interviewee AC, in-person conversa- tion, 8/20/12). line services that were enabled by those projects (Interviewee AE, in-person conversation, 8/20/12). Interviewees in Dallas were interested not in the total num- ber of riders on the line but in attracting incremental choice riders to the facility to relieve traffic congestion on a parallel highway (Interviewee AF, in-person conversation, 8/13/12). This observation suggests that, at a minimum, this study’s project-level ridership model should be used in combination with its system-wide PMT model. Given the lack of a “choice rider” output variable, however, Dallas may not have used the spreadsheet tool had it existed when they were planning their project. Although cost considerations were foremost—given that a project could not be built if it exceeded the limits of local funding and federal matching capacities—local planners sel- dom considered cost per rider as a success measure. Several interviewees said that they would be more likely to use a rider per mile metric, saying they saw it as a more intuitive met- ric for transit agency board members and the public (Inter- viewee AG, in-person conversation, 8/30/12; Interviewee AF, in-person conversation, 8/13/12; Interviewee AH, in-person conversation, 6/5/12). Most of the interviewees reported that they considered development density and the connection of activity centers when designing a project. Charlotte transit staffers were staunch promoters of downtown job growth and real estate development to help justify their investment in light rail rather than enhanced bus services (Interviewee AD, telephone con- versation, 8/24/12). Portland regional planners first argued against Westside Express Service, the region’s commuter rail (CR) facility, because of the low densities along the line (Interviewee AI, in-person conversation, 8/7/12). However, land use density thresholds had little explicit influence on transit technology choices. In most instances, the mode of transit was dictated by the system that was being extended or by the funding sources available to the sponsoring agency. Eugene’s transit agency was one exception, using density thresholds to argue for the less expensive BRT mode rather than light rail. Even though interviewees felt that the model provided excellent predictions of ridership on past projects, they wor- ried that its method lacked face validity and that it would be susceptible to criticism. For example, planners in several cities focused on travel-time competitiveness as a predictor of ridership. They felt that any model that did not include a proxy of such a measure would be considered faulty by con- stituents who had become accustomed to both a four-step regional model that directly considered travel times and a federal project evaluation process that for many years had focused on travel-time savings (Interviewee AG, in-person conversation, 8/30/12). Some interviewees expressed discom- fort with sharing TCRP Project H-42 model outputs that are

2-48 The Eugene EmX BRT case suggests that planners of rail and bus fixed-guideway projects may consider the same mea- sures of success. The researchers found that BRT planning lev- eraged many of the same indicator-based methods that were observed in HRT and LRT case studies. As with the other cases, qualitative measures and indicators were often more impor- tant in defining transit plans than quantitative considerations such as ridership. As the initial fixed-guideway investment in a region, the Eugene case shares some features with the Char- lotte South Line. The EmX alignment was selected among sev- eral other viable segments in the system plan because, given its potential to succeed across various measures, it offered the greatest potential to garner political support for additional fixed-guideway investments. The Interstate MAX project in Portland offered an oppor- tunity to investigate a transit project for which project success was redefined under changing planning conditions. A major portion of the Interstate MAX project that was ultimately constructed was projected to be more expensive, slower, have lower ridership, and have more nuisance impacts on the neighborhood than alternatives. However, the qualitative notion of transit-driven community development swayed decision-makers. The project is widely believed to be suc- cessful because it has provided several years of travel benefits for citizens, generated significant community development benefits for the neighborhoods it currently serves, and pre- served opportunities to expand the project north and south as originally envisioned. The Salt Lake City University and Medical Center Exten- sion case is one in which both system planning and proj- ect design were informed by rules of thumb that related to ridership and several other measures of success. Much of the planning for the region’s rail system related to highway capacity constraints, and this project provided a cross-town rail transit connection between major destinations in a corridor that lacked highway links. Relative to other cases, the planning and development of the line was fast-tracked so that operations could commence in time for the 2002 Winter Olympics. Although planning focused on one set of criteria, the projects have proven to be successful across multiple measures of success, suggesting that some indica- tors may effectively address multiple considerations simul- taneously. Finally, the Washington, D.C., Branch Avenue Extension case provided an opportunity to review documents detail- ing 20 years of debate over the route and station locations for a project. Planners of the Green Line had to repeatedly prove their case for the line to Congress, to WMATA’s mem- ber jurisdictions, and to various groups that advocated for alternative alignments or to stop construction altogether. In the end, the line met ridership projections while providing In every case study city, interviewees thought the tool could be helpful for reducing certain workloads. They proposed using the model to narrow the number of project alternatives before they handed proposals to their regional transportation modelers for more robust analyses. Several interviewees sug- gested that they might be able to intuit results after just a few applications and would no longer rely on the tool. Regardless of the usefulness of the method, interviewees agreed that their agency would use any tool that was officially sanctioned by FTA to be used in the New Starts evaluation process (Interviewee AA, in-person conversation, 8/7/12). Given a choice between an FTA-approved spreadsheet tool and an FTA-approved regional model, one planner stated that their agency would likely use whichever model gave them the answer that would win funding (Interviewee AM, in-person conversation, 8/7/12). 6.5 Synopses The Charlotte Lynx South Line case study highlighted the interplay between transportation-related rules of thumb and politically driven strategic thinking, both of which shape transit project planning. The case study suggests that transit project planners consider a wide array of success indicators to predict performance across several measures. Those mea- sures may be more related to indirect transit outcomes, such as land use impacts, than to direct measures of success, such as ridership. This case study suggests that transit planning is a complex art that uses both qualitative indicators and quan- titative forecasts to balance various expectations for a single fixed-guideway transit project. The Dallas North Central Corridor case study provided insights on transit planning in a highway-oriented metro- politan area. Many of the transit project success factors and attendant indicators considered in Dallas related to high- way issues, such as capacity, demand, and expansion costs. This led to the prioritization of project elements that would attract choice riders, thereby helping mitigate highway con- gestion. Although other evaluation criteria and success mea- sures were considered, the project’s overpasses, direct routing, and park-and-ride facilities reflected these highway-oriented concerns. Though the project has underperformed in terms of ridership projections, it is considered a success based on many other qualitative measures. For example, the presence of rail transit is considered a regional economic benefit and a competitive advantage in the global marketplace. It would seem that, in the eyes of many Dallas stakeholders, the most important measure of success for the North Central Corridor extension—and any other DART light rail projects—is that any rail transit was built in unabashedly automobile-centric Dallas, Texas.

2-49 Detailed case study write-ups are provided in Appendix I. The write-ups explain how six distinct transit projects fit into their respective regions’ transit systems and describe the plan- ning processes that led to the ultimate project being chosen from the alternatives proposed during the planning process. They also demonstrate how largely heuristic indicator-based methods have been used to predict the success of transit proj- ects. Regional descriptions of each case study area appear at the end of Appendix I. high-quality transit service to one of the most economically depressed parts of the Washington, D.C., metropolitan area. It was service to a particular location—a station area upheld as an archetypal transit-dependent community—that justi- fied the overall project and the relatively higher cost and less efficient operations of the chosen alternative. The case high- lights the diversity of success measures that can be considered for a single project and how local priorities shaped the defini- tion and interpretation of success measures.

2-50 C H A P T E R 7 The research team combined the results of the model- ing process with feedback from practitioners to create a spreadsheet tool that can be used to predict the transit use impacts of proposed fixed-guideway projects. To build the tool, the researchers selected spreadsheet-friendly versions of the project- and metropolitan-level models, incorporat- ing recommendations from case study participants and the report’s review panel. Selection criteria were that the model must be usefully predictive, the model variables must be easily interpretable, and the data collection process for the model inputs must not be too onerous. A user guide for the spreadsheet tool is provided as Chapter 3 in the hand- book. The material in this chapter provides additional background. The spreadsheet tool serves as an implementation of the models. Its outputs are produced by substituting data on a proposed fixed-guideway transit project into the model equa- tions discussed in Sections 5.2 and 5.3. Users enter data of the same form as the variables comprising the spreadsheet’s underlying models. Where possible, the spreadsheet contains pre-entered data, such as the population of the metropolitan area, so the user can select from a dropdown list. Each input is automatically normalized or otherwise ma - nipulated as necessary, then multiplied by the model coeffi cient. For the project-level model, ridership is simply a sum of the inputs multiplied by their coefficients. At the metropolitan level, incremental PMT is given by subtracting an estimate of PMT on the committed network from PMT on the commit- ted network plus the proposed project. For this calculation, the spreadsheet calculates the model outputs once by multi- plying the coefficients by the data describing the committed network and again by multiplying the coefficients by the sum of the connected network and the proposed project. Users can navigate through the spreadsheet tool with vari- ous buttons. From the inputs page, a button brings the user to the outputs page where a table shows the transit use char- acteristics of the proposed project. In addition to common metrics of success such as ridership, cost per mile, and incre- mental PMT, several charts plot the proposed project in the context of other projects. The comparisons were taken from the researchers’ full database of projects, although not all of the charts portray all of the projects. Some attributes of the comparison projects are reported statistics (such as weekday ridership) whereas others (incremental PMT) have been calculated using the model in the spreadsheet. After examining the predicted transit use, users of the spread- sheet tool can navigate back to the inputs page to adjust the inputs, experimenting with different alignments or growth scenarios. The navigational buttons initiate Microsoft Visual Basic macros which, in addition to navigating between tabs in the spreadsheet, provide the user with visual assistance such as ranking the proposed project in order with the comparison projects, changing the color of the project to make it more visible, or selecting a subset of projects from the database to provide more relevant comparisons. Enabling macros (a set- ting on the user’s Excel interface) is essential for getting the most valuable results from the tool. Spreadsheet Tool: Technical Notes

2-51 C H A P T E R 8 The evaluation method proposed in TCRP Report 167 is not meant to replace existing processes of planning fixed- guideway transit systems, but rather to provide additional information that is consistent for all regions in the United States. The indicators of success presented in this report are only the beginning. If a corridor or project is shown to have good potential for attracting ridership commensurate with its cost, it may be appropriate to conduct more detailed cor- ridor-level planning studies of transit needs and alternative solutions. These studies would typically include the use of travel-demand forecasting models, conceptual engineering, environmental studies, and stakeholder involvement. 8.1 Implementation The implementation of the research recognizes the issues that need to be addressed during the adoption process and the overall concept of research product implementation. The three products of this research are this final research report (presented as Volume 2 in TCRP Report 167), the hand- book (presented as Volume 1 in TCRP Report 167), and the spreadsheet tool (which may be downloaded by accessing the report’s web page at www.trb.org). The research report docu- ments the literature review, the analysis of potential success factors, the spreadsheet tool, and the case studies. The hand- book provides a user-friendly guide to the indicator-based approach in general and to the use of the spreadsheet tool. Publication of TCRP Report 167 electronically through the TCRP website will provide ready access to all of these materi- als for researchers who are interested in how this analysis adds to the understanding of transit success factors. The research results may also be of use to FTA as it develops guidance on its New Starts/Small Starts criteria and potential warrants for evaluating new projects. The handbook will be of interest to transit planning pro- fessionals at transit agencies, MPOs, local jurisdictions, state DOTs, consulting firms, and others interested in conducting a preliminary assessment of a project’s potential. Users will need to access both the handbook and the spreadsheet tool. Successful implementation of this research faces two poten- tial impediments. The first potential impediment is best described as the “black box” nature of the spreadsheet tool. Without an understanding of how the tool works, users may lack confidence in its ability to provide reliable results, and thus may be reluctant to use it. Practitioners are most com- fortable with the tools they know and trust. The second potential impediment is the limitations of the database that underlies the spreadsheet tool. Projects completed since 2008 are not included, and over time the tool may be perceived as more and more dated. In addition, the database used in this research has few CR and BRT projects, and no streetcars or other urban circulator projects. Users who are interested in exploring the potential of these modes may decide to look elsewhere. That said, further research could be undertaken to keep the database up to date, include additional transit tech- nologies, and modify the spreadsheet tool. As this project was being completed, FTA reported that it was developing its own simplified techniques for pre- dicting the benefits of a fixed-guideway system (see Federal Register Vol. 78, Number 6, January 9, 2013), and FTA has since released a report. Given that the ridership projections developed using the FTA-sanctioned methodology will be more readily accepted for funding purposes, practitioners may choose to rely on that methodology rather than the tool developed in this research; however, practitioners may also find it useful to use both methods and compare the results. Conclusion

2-52 Aizenman, N. A Lot Riding on New Stations; Area Residents Hope Metro Brings Growth, Not Crime and Noise. Washington Post, Jan. 13, 2001. Allison, P. D. Missing Data. Sage Productions, Thousand Oaks, CA, pp. 9–11. American Public Transportation Association. 2012 Public Transportation Fact Book, Appendix A: Historical Tables. Washington, D.C., 2012. Amodei, R., and D. Schneck. Fixed Guideway Capital Costs: Heavy Rail and Busway HOV Lane. Federal Transit Administration, Spring- field, VA, 1994. Baird, J. TRAX Debuts Extension to the U. Medical Center, Salt Lake Tribune, Sept. 28, 2003. Banister, D., and J. Berechman. Transport Investment and Economic Development. University College London Press, 2000. Barnes, G. The Importance of Trip Destination in Determining Transit Share. Journal of Public Transportation, Vol. 8, No. 2, 2005. Baum-Snow, N., and M. E. Kahn. The Effects of New Public Projects to Expand Urban Rail Transit. Journal of Public Economics, Vol. 77, No. 2, 2000, pp. 241–263. doi:10.1016/S0047-2727(99)00085-7. Baum-Snow, N., and M. E. Kahn. Effects of Urban Rail Transit Expan- sions: Evidence from Sixteen Cities, 1970–2000. Brookings-Wharton Papers on Urban Affairs, 2005, pp. 147–206. Bay Area Rapid Transit. BART Reports. Available at: http://www.bart. gov/about/reports/index.aspx (accessed March 2011). Bell, R. Metro Celebrates Breaking Ground For Final Leg in PG. Wash- ington Times, Sept. 24, 1995. Bertaud, A., and H. W. Richardson. Transit and Density: Atlanta, the United States and Western Europe. Urban Sprawl in Western Europe and the United States. Ashgate Publishing, Ltd., 2004, Chapter 17. Boarnet, M., and R. Crane. LA Story: A Reality Check for Transit-Based Housing. Journal of the American Planning Association, Vol. 63, No. 2, 1997, pp. 189–204. Boarnet, M., and R. Crane. Public Finance and Transit-Oriented Plan- ning: New Evidence from Southern California. Journal of Planning Education and Research, Vol. 17, No. 3, 1998, p. 206. Boarnet, M., and R. Crane. The Influence of Land Use on Travel Behavior: Specification and Estimation Strategies. Transportation Research Part A: Policy and Practice, Vol. 35 No. 9, 2001, pp. 823–845. doi:10.1016/ S0965-8564(00)00019-7. Booz Allen Hamilton. Light Rail Transit Capital Cost Study Update Final Report, submitted to Federal Transit Administration, Washington, D.C., 2003. Booz Allen Hamilton. TCRP Web-Only Document 31: Managing Capi- tal Costs of Major Federally Funded Public Transportation Projects: Contractor’s Final Report, Transportation Research Board of the National Academies, Washington, D.C., 2005. Available at: http:// www.trb.org/Publications/Public/Blurbs/Managing_Capital_ Costs_of_Major_Federally_Funded_P_158051.aspx. Bowman, J. L., and M. E. Ben-Akiva. Activity-Based Disaggregate Travel Demand Model System with Activity Schedules. Transportation Research Part A: Policy and Practice, Vol. 35, No. 1, 2001, pp. 1–28. doi:10.1016/S0965-8564(99)00043-9. Brown, J., and G. Thompson. The Relationship between Transit Ridership and Urban Decentralisation: Insights from Atlanta. Urban Studies, Vol. 45, Nos. 5&6 (May 2008), pp. 1119–1139. Brown, J., and G. Thompson. Should Transit Serve the CBD or a Diverse Array of Destinations? A Case Study Comparison of Two Transit Systems. Journal of Public Transportation, Vol. 15, No. 1, 2012. Cairns, S., J. Greig, and M. Wachs. Environmental Justice & Transpor- tation: A Citizen’s Handbook. Institute of Transportation Studies, University of California, Berkeley, CA, 2013. Central Lane Metropolitan Planning Organization. Regional Transpor- tation Plan, Lane Council of Governments, 2001. Cervero, R. Transit Pricing Research. Transportation, Vol. 17, No. 2, 1990a, pp. 117–139. doi:10.1007/BF02125332. Cervero, R. The Transit Metropolis: A Global Inquiry, 1st ed., Island Press, 1998. Cervero, R. Alternative Approaches to Modeling the Travel-Demand Impacts of Smart Growth. Journal of the American Planning Asso- ciation, Vol. 72, No. 3, 2006, pp. 285–295. Cervero, R. Transit-Oriented Development’s Ridership Bonus: A Prod- uct of Self-Selection and Public Policies. Environment and Plan- ning, A. Vol. 39, No. 9, 2007, pp. 2068–2085. Cervero, R., G. B. Arrington, J. Smith-Heimer, R. Dunphy, et al. TCRP Report 102: Transit-Oriented Development in the United States: Experiences, Challenges, and Prospects. Transportation Research Board of the National Academies, Washington, D.C., 2004. Cervero, R., and E. Guerra. To T or Not to T: A Ballpark Assessment of the Costs and Benefits of Urban Rail Transportation. Public Works Management & Policy, 2011. Cervero, R., and C. D. Kang. 2011. “Bus Rapid Transit Impacts on Land Uses and Land Values in Seoul, Korea.” Transport Policy, 18 (1), pp. 102–116. Cervero, R., and K. Kockelman. Travel Demand and the 3 Ds: Density, Diversity, and Design. Transportation Research Part D: Transport and Environment, Vol. 2, No. 3, 1997, pp. 199–219. doi:10.1016/ S1361-9209(97)00009-6. References

2-53 Cervero, R., and J. Landis. Development Impacts of Urban Transport: A US Perspective. Transport and Urban Development, 1995, pp. 136–156. Cervero, R., and J. Landis. Twenty Years of the Bay Area Rapid Tran- sit System: Land Use and Development Impacts. Transportation Research Part A: Policy and Practice, Vol. 31, No. 4, 1997, pp. 309– 333. doi:10.1016/S0965-8564(96)00027-4. Cervero, R. and J. Murakami. Effects of Built Environments on Vehicle Miles Traveled: Evidence from 370 U.S. Urbanized Areas. Environ- ment and Planning, A, Vol. 42, No. 2, 2010, pp. 400–418. Charlotte Area Transit System. South Corridor Light Rail Project: Before and After Report, 2012. Chatman, D. Deconstructing Development Density: Quality, Quantity and Price Effects on Household Non-Work Travel. Transportation Research, Part A, Vol. 42, No. 7, 2008, pp. 1008–1030. Chatman, D., and N. Klein. Immigrants and Travel Demand in the United States: Implications for Transportation Policy and Future Research. Public Works Management Policy, Vol. 13, No. 4, 2009, pp. 312–327. doi:10.1177/1087724X09334633. Chatman, D., et al. TCRP Web-Only Document 56: Methodology for Determining the Economic Development Impacts of Transit Proj- ects. Transportation Research Board of the National Academies, Washington, D.C., 2012. City of Portland, Bureau of Technology Services. 2011. Corporate GIS. Available at: http://www.portlandonline.com/bts/index.cfm? c=25779 (accessed March 2011). City of Portland Office of Transportation. 2001. Interstate Corridor Community Involvement. Clower, T., B. Weinstein, and M. Seman. Assessment of the Potential Fis- cal Impacts of Existing and Proposed Transit-Oriented Development in the Dallas Area Rapid Transit Service Area. Center for Economic Development and Research University of North Texas, Denton, TX, 2007. Cook, E. A. Landscape Structure Indices for Assessing Urban Ecological Networks. Landscape and Urban Planning, Vol. 58, 2002, pp. 269–280. Crane, R. On Form versus Function: Will the New Urbanism Reduce Traffic, or Increase It? Journal of Planning Education and Research, Vol. 15, No. 2, 1996, pp. 117–126. doi:10.1177/0739456X9601500204. Daganzo, C. F. Structure of Competitive Transit Networks. Transporta- tion Research Part B, Vol. 44, No. 4, 2010, pp. 434–446. Dallas Area Rapid Transit Authority. 1991. Briefing on Second AA/DEIS Corridor Selection. Dallas Area Rapid Transit Authority. 2006. North Central LRT Extension: Before and After Study. Dallas Area Rapid Transit Authority. 2012. DART Reference Book— April 2012. Data Quick. DataQuick Database. Available at: http://www.dataquick. com/products/database/ (accessed March 2011). Davidson, L. New Airport TRAX Line is in Final Push Toward Comple- tion. Salt Lake Tribune, May 1, 2012. Deakin, E., C. Farrell, J. Mason, and J. Thomas. 2002 Policies and Prac- tices for Cost-Effective Transit Investments: Recent Experiences in the United States. Transportation Research Record: Journal of the Transportation Research Board, 1799(-1),1-9. doi: 10.3141/1799-01. De Neufville, R. Applied Systems Analysis: Engineering Planning and Technology Management, McGraw-Hill Companies, 1990. Evans, J. E. TCRP Report 95: Road Value Pricing: Traveler Response to Transportation System Changes, Chapter 9: Transit Scheduling and Frequency. Transportation Research Board of the National Acad- emies, Washington, D.C., 2004. Ewing, R., and R. Cervero. Travel and the Built Environment—A Meta- Analysis. Journal of the American Planning Association, Vol. 76, No. 3, 2010, p. 265. doi:10.1080/01944361003766766. Feaver, D. Prince George’s County Council Chooses Rosecroft Metro Line. Washington Post, May 10, 1978. Feaver, D. B. What Ever Happened to the Green Line? Washington Post, Oct. 14, 1980. Federal Highway Administration. Highway Statistics Series. Available at: http://www.fhwa.dot.gov/policyinformation/statistics.cfm. Federal Highway Administration. Traffic Congestion and Reliability: Linking Solutions to Problems. http://ops.fhwa.dot.gov/congestion_ report_04/appendix_C.htm Federal Transit Administration. New Starts Ratings: North Central Corridor, 1996. Federal Transit Administration. New Starts Ratings: Dallas, Texas/North Central Corridor, 1998. Federal Transit Administration. Annual Report on New Starts, 2000. Federal Transit Administration. Annual Report on New Starts, 2002. Federal Transit Administration New Starts Funding Allocation: South Corridor LRT Charlotte, North Carolina, FTA NSFA, 2005. Federal Transit Administration South Corridor LRT Charlotte, North Carolina, FTA SCLRT, 2005. Federal Transit Administration. New Starts Before and After Study Report, 2007. Federal Transit Administration. The Predicted and Actual Impacts of New Starts Projects—2007, 2008. Federal Transit Administration. The EmX Franklin Corridor—BRT Project Evaluation, 2009. Federal Transit Administration and Charlotte Area Transit System. Draft Environmental Impact Statement: South Corridor Light Rail Project—Charlotte-Mecklenburg County Light Rail System, 2002. Federal Transit Administration and Charlotte Area Transit System. Final Environmental Impact Statement: South Corridor Light Rail Project—Charlotte-Mecklenburg County Light Rail System, 2003. First American CoreLogic. RealQuest Professional. Available at: http:// pro.realquest.com/home/ (accessed March 2011). Florida Department of Transportation, Transportation Statistics Office. TranStat. Available at: http://www.dot.state.fl.us/planning/statistics/ (accessed March 2011). Flyvbjerg, B. Cost Overruns and Demand Shortfalls in Urban Rail and Other Infrastructure. Transportation Planning and Technology, Vol. 30, No. 1, 2007, pp. 9–30. Flyvbjerg, B., M. K. Skamris Holm, and S. L. Buhl. Underestimating Costs in Public Works Projects: Error or Lie? Journal of the Ameri- can Planning Association, Vol. 68, No. 3, 2002, pp. 279–295. Flyvbjerg, B., N. Bruzelius, and W. Rothengatter. Megaprojects and Risk: An Anatomy of Ambition. Cambridge University Press, 2003. Forman, R. T. and M. Godron. Landscape Ecology. Wiley, New York, 1986, pp. 417–419. Frank, L., M. Bradley, S. Kavage, J. Chapman, and T. K. Lawton. Urban Form, Travel Time, and Cost Relationships with Tour Complexity and Mode Choice. Transportation, Vol. 35, No. 1, 2008, pp. 37–54. doi:10.1007/s11116-007-9136-6. Garrett, M., and B. Taylor. Reconsidering Social Equity in Public Tran- sit. Berkeley Planning Journal, Vol. 13, 1999, pp. 6–27. Gercek, H., B. Karpak, and T. Kilincaslan. A Multiple Criteria Approach for the Evaluation of the Rail Transit Networks in Istanbul. Trans- portation: Planning, Policy, Research, Practice, Vol. 31, Issue 2, 2004, pp. 203–228. Glaeser, E. L., M. E. Kahn, and J. Rappaport. Why Do the Poor Live in Cities? The Role of Public Transportation. Journal of Urban Eco- nomics, Vol. 63, No. 1, 2008, pp. 1–24. doi:10.1016/j.jue.2006.12.004. Gomez-Ibanez, J. A. Big-City Transit Ridership, Deficits, and Politics: Avoiding Reality in Boston. Journal of the American Planning Asso- ciation, Vol. 62 No. 1, 1996, p. 30. doi:10.1080/01944369608975669.

2-54 Guerra, E. Valuing Rail Transit: Comparing Capital and Operating Costs to Consumer Benefits. Transportation Research Record: Journal of the Transportation Research Board, No. 2219, Transportation Research Board of the National Academies, Washington, D.C., 2011. Guerra, E., and R. Cervero. Cost of a Ride: The Effects of Densities on Fixed-Guideway Transit Ridership and Capital Costs. University of California Transportation Center Research Report, University of California, Berkeley, 2010. Guerra, E. and R. Cervero. Cost of a Ride. Journal of the American Plan- ning Association, Vol. 77, No. 3, 2011, pp. 267–290. Hagget, P., Cliff, A. D., and Fry, A. Locational Analysis in Human Geog- raphy, 2d Ed. Wiley, New York, 1977, p. 454. Harford, J. D. Congestion, Pollution, and Benefit-to-Cost Ratios of US Public Transit Systems. Transportation Research Part D: Transport and Environment, Vol. 11, No. 1, 2006, pp. 45–58. doi:10.1016/ j.trd.2005.09.001. Harmatuck, D. Light Rail Cost Functions and Technical Inefficiency. Transportation Research Record: Journal of the Transportation Research Board, No. 2042. Transportation Research Board of the National Academies, Washington, D.C., 2008, pp. 58–70. doi:10.3141/2042-07. Hess, P. M. Measures of Connectivity. Places, Vol. 11, No. 2, 1997, pp. 58–65. Hess, D. B., and T. M. Almeida. Impact of Proximity to Light Rail Rapid Transit on Station-Area Property Values in Buffalo, New York. Urban Studies, Vol. 44, No. 5-6 (May 1, 2007), pp. 1041–1068. doi:10.1080/00420980701256005. Holmgren, J. Meta-Analysis of Public Transport Demand. Transpor- tation Research Part A: Policy and Practice, Vol. 41, No. 10, 2007, pp. 1021–1035. doi:10.1016/j.tra.2007.06.003. Kain, J. Choosing the Wrong Technology: Or, How to Spend Billions and Reduce Transit Use. Journal of Advanced Transportation, Vol. 21, No. 3, 1988, pp. 197–213. doi:10.1002/atr.5670210303. Keeler, T., and K. Small, Associates. The Full Costs of Urban Transport, Part III: Automobile Costs and Final Intermodal Comparisons. Monograph. University of California, Berkeley: Institute of Urban and Regional Development, 1975. Kittelson & Associates Inc., KFH Group, Inc., Parsons Brinckerhoff, Inc., and K. Hunter-Zaworski. TCRP Report 100:Transit Capacity and Quality of Service Manual—2d Ed.: Part 3 Quality of Service Contents. Transportation Research Board of the National Acad- emies, Washington, D.C., 2003. Available at: http://onlinepubs.trb. org/onlinepubs/tcrp/tcrp100/part%203.pdf. Kohn, H. Factors Affecting Urban Transit Ridership. Statistics Canada, 1999. Kuby, M., A. Barranda, and C. Upchurch. Factors Influencing Light-Rail Station Boardings in the United States. Transportation Research Part A, Vol. 38, 2004, pp. 223–247. Lane Transit District. EmX promotional brochure, date unknown. Layton, L. “Green Line Riders Are Feeling Crunched.” Washington Post, Jan. 19, 2001. Layton, L. Metro Seeks To Unclog Green Line. Washington Post, Jan. 25, 2001. Lee, D. B. Requiem for Large-Scale Models. Journal of the American Institute of Planners, Vol. 39, 1973, pp. 163–178. Levinson, D. M., and A. Kumar. Density and the Journey to Work. Growth & Change, Vol. 28, No. 2, 1997, p. 147. Lewis-Workman, S., and D. Brod. Measuring the Neighborhood Ben- efits of Rail Transit Accessibility. Transportation Research Record: Journal of the Transportation Research Board, No. 1576, Transporta- tion Research Board of the National Academies, Washington D.C., 1997, pp. 147–153. Lewis-Workman, S., B. White, S. McVey, and F. Spielberg. The Predicted and Actual Impacts of New Starts Projects—Capital Costs and Ridership. Technical Report, Federal Transit Administration and Vanasse Hangen Brustlin, Inc., 2008. Litman, T. Evaluating Public Transit Benefits and Costs. Victoria Trans- port Policy Institute, 2006. Liu, Z. Determinants of Public Transit Ridership: Analysis of Post World War II Trends and Evaluation of Alternative Networks. Harvard University, Cambridge, MA, 1993. Lowe, J. C. and S. Moryadas. The Geography of Movement. Houghton Mifflin, Boston, MA, 1975, p. 442. Massachusetts Bay Transportation Authority. Information at: http:// www.mbta.com/ (accessed March 2011). Maze, T. H. and O. Smadi. GASB Statement 34: The On-ramp to Transportation Asset Management, or a Detour Leading Busi- ness as Usual? Transportation Quarterly, Vol. 57, No. 4, 2003, pp. 23–29. McCollum, B. E., and R. H. Pratt. TCRP Report 95: Traveler Response to Transportation System Changes. Chapter 12—Transit Pricing and Fares. Transportation Research Board of the National Academies, Washington, D.C., 2004. McMillen, D. P., and J. McDonald. Reaction of House Prices to a New Rapid Transit Line: Chicago’s Midway Line, 1983–1999. Real Estate Economics, Vol. 32, No. 3, 2004, pp. 463–486. doi:10.1111/j.1080- 8620.2004.00099.x. Meyer, J., J. Kain, and H. Wohl. The Urban Transportation Problem. Harvard University Press, Cambridge, MA, 1965. Metropolitan Transportation Commission. Resolution 3434, as revised 4/26/06. (MTC Res3434), 2006. Miller, J. and W. Ibbs. Toward a New Paradigm: Simultaneous Use of Multiple Project Delivery Methods. Journal of Management in Engi- neering, May/June 2000, pp. 58–66. Mohring, H. Land Values and the Measurement of Highway Benefits. The Journal of Political Economy, Vol. 69, No. 3, 1961, pp. 236–249. Mohring, H. Optimization and Scale Economies in Urban Bus Trans- portation. The American Economic Review, Vol. 62, No. 4, Septem- ber 1972, pp. 591–604. Moilanen, A. and M. Nieminen. Simple Connectivity Measures in Spatial Ecology. Ecology, Vol. 83, 2002, pp. 1131–1145. Musgrave, R. A. The Theory of Public Finance: A Study in Public Econ- omy, 1959. Nelson, A. C. Transit Stations and Commercial Property Values: A Case Study with Policy and Land-Use Implications. Journal of Public Transportation, Vol. 2, No. 3, 1999, pp. 77–93. Neumann, L. and M. Markow. Performance-based Planning and Asset Management. Public Works Management and Policy, Vol. 8, No. 3, 2004, pp. 156–161. Newman, P., and J. Kenworthy. Urban Design to Reduce Automobile Dependence. Opolis, Vol. 2, No. 1, 2006, pp. 35–52. North Central Texas Council of Governments Transportation Depart- ment. Rail Station Access: Bicycle and Pedestrian Needs Assessment, 2011. Available at: http://www.nctcog.org/trans/sustdev/bikeped/ access_to_rail/atrgis.asp (accessed March 2011). O’Toole, R. Defining Success: The Case Against Rail Transit. Policy Anal- ysis, No. 663, 2010. Available at: http://www.cato.org/pub_display. php?pub_id=11608. Okun, A. Equality and Efficiency, the Big Tradeoff. Brookings Institution Press, 1975. Parry, I. W. H., and K. Small. Should Urban Transit Subsidies Be Reduced? American Economic Review, Vol. 99, 2009, pp. 700–724. doi:10.1257/aer.99.3.700.

2-55 Small, K. Project Evaluation, Chapter 5. Essays in Transportation Eco- nomics and Policy: A Handbook in Honor of John R. Brookings Insti- tution Press, Washington, D.C., 1999, pp. 137–177. Small, K. and E. Verhoef. The Economics of Urban Transportation. Taylor & Francis Group, 2007. Spanberg, E. Charlotte Wins $580M from Feds for Light-rail Extension, Charlotte Business Journal, Oct. 16, 2012. Stokes, R. J., J. MacDonald, and G. Ridgeway. Estimating the Effects of Light Rail Transit on Health Care Costs. Health & Place, Vol. 14, No. 1, 2008, pp. 45–58. doi:10.1016/j.healthplace.2007.04.002. Sugihara, G. and R. M. May. Applications of Fractals in Ecology. Trends in Ecology and Evolution, Vol. 5, No. 3, 1990, pp. 79–86. Taaffe, E. J., H. L. Gauthier, and M. E. O’Kelly. Geography of Transporta- tion, 2d Ed. Prentice-Hall, New Jersey, 1996. Taylor, B. D. The Geography of Urban Transportation Finance. The Geography of Urban Transportation, 3d. Ed., eds. Susan Hanson and Genevieve Giuliano, New York: The Guilford Press, 2004, pp. 294–331. Taylor, B., E. J. Kim, and J. E. Gahbauer. The Thin Red Line: A Case Study of Political Influence on Transportation Planning Practice. Journal of Planning Education and Research, Vol. 29, No. 2, 2009, pp. 173–193. doi:10.1177/0739456X09344718. Taylor, B., D. Miller, H. Iseki, and C. Fink. Nature and/or Nurture? Ana- lyzing the Determinants of Transit Ridership across US Urbanized Areas. Transportation Research Part A: Policy and Practice, Vol. 43, No. 1, 2009, pp. 60–77. doi:10.1016/j.tra.2008.06.007. Theobaldi, D. M., K. Ramseyii, and J. Thomasii. A National Dataset for Characterizing Location Sustainability and Urban Form. Draft Technical Report, February 2011. Thompson, G., and J. Brown. Explaining Variation in Transit Ridership in U.S. Metropolitan Areas Between 1990 and 2000: Multivariate Analysis. Transportation Research Record: Journal of the Transpor- tation Research Board, No. 1986. Transportation Research Board of the National Academies, Washington, D.C., 2006, pp. 172–181. doi:10.3141/1986-23. Thompson, G., and J. Brown. Evaluation of Land Use and Transporta- tion Strategies to Increase Suburban Transit Ridership in the Short Term. The Public Transit Office, Florida Department of Transpor- tation, 2010. Thompson, G., and J. Brown. Making a Successful LRT-Based Regional Transit System: Lessons from Five New Start Cities. Journal of Public Transportation, Vol. 15, No. 2, 2012. Thompson, G., J. Brown, T. Bhattacharya, and M. Jaroszynski. Under- standing Transit Ridership Demand for a Multi-Destination, Multi- modal Transit Network in an American Metropolitan Area: Lessons for Increasing Choice Ridership While Maintaining Transit Depen- dent Ridership. Mineta Transportation Institute, Report 11-06, January 2012. Tischendorf, L., and L. Fahrig. How Should We Measure Landscape Connectivity? Landscape Ecology, Vol. 15, 2000, pp. 633–641. Tomer, A. Where the Jobs Are: Employer Access to Labor by Transit. Metro politan Infrastructure Initiative Series and Metropolitan Opportunity Series. Brookings Institution, Washington, D.C., July 2012. Tomer, A, E. Kneebone, R. Puentes and A. Berube. Missed Opportunity: Transit and Jobs in Metropolitan America. Metropolitan Infrastruc- ture Initiative Series and Metropolitan Opportunity Series, Brookings Institution, Washington, D.C., May 2011. Transportation Institute. Annual Urban Mobility Report. Available at: http://mobility.tamu.edu/ums/. Parsons Brinckerhoff Quade & Douglas, Inc. I-15/State Street Corri- dor Study—Transit, Salt Lake County, Utah: Final Environmental Impact Statement. Federal Transit Administration and Utah Tran- sit Authority, 1994. Parsons Transportation Group. University-Downtown-Airport Major Investment Study: Draft Environmental Impact Statement. Wasatch Front Regional Council, July 1997. Parsons Transportation Group. Airport to University West-East Light Rail Project: Final Environmental Impact Statement. Utah Transit Authority, 1999. Peat, Marwick, Mitchell, & Co. Metrorail Alternatives Analysis: F Route. Metropolitan Washington Council of Governments, 1977. Pendall, R., R. Puentes, and J. Martin. From Traditional to Reformed: A Review of the Land Use Regulations in the Nation’s 50 largest Metropolitan Areas. Brookings, 2006. Retrieved from http://www. brookings.edu/research/reports/2006/08/metropolitanpolicy- pendall. Pickrell, D. Urban Rail in America: A Review of Procedures and Recom- mendations from the Regional Plan Association Study. Transportation Systems Center, U.S. Department of Transportation, Cambridge, MA, 1985. Pickrell, D. Urban Rail Transit Projects: Forecast Versus Actual Ridership and Costs. Urban Mass Transportation Administration, Washington, D.C., 1990. Pickrell, D. A Desire Named Streetcar: Fantasy and Fact in Rail Transit Planning. Journal of the American Planning Association, Vol. 58, No. 2, 1992, pp. 158–176. Pickrell, D. Transportation and Land Use. Transportation Economics and Policy Handbook, 1999. Project Management Group (PMG). South/North Transit Corridor Study—Briefing Document: Tier I Technical Summary Report, 1994. Project Management Group. South/North Transit Corridor Study— Briefing Document: Design Option Narrowing, 1995. Project Management Group. South/North Transit Corridor Study— Design Option Narrowing Final Report, 1996. Portland Metro. Segments and Design Options, 1995. Portland Metro. South/North Corridor Project: Draft Environmental Impact Statement, 1998. Portland Metro. Regional High Capacity Transit System Plan 2035: Public Involvement Outreach Summary and Attachments. Retrieved from http://library.oregonmetro.gov/files/hct_pi_outreach_summaryno_ attachments.pdf Pushkarev, B., and J. Zupan. Public Transportation and Land Use Policy. Indiana University Press, Bloomington, IN, 1977. Pushkarev, B., J. Zupan, and R. S. Cumella. Urban Rail in America: An Exploration of Criteria for Fixed Guideway Transit. Indiana Univer- sity Press, Bloomington, IN, 1982. Randall, T. A., and B. W. Baetz. Evaluating Pedestrian Connectivity for Suburban Sustainability. Journal of Urban Planning and Develop- ment, Vol. 127, No. 1, 2001, pp. 1–15. Rodrigue, J., C. Comtois, and B. Slack. Geography of Transportation Systems, 2d Ed. Routledge, New York, 2009. San Francisco Bay Area Metropolitan Transportation Commission. GIS Data: Category 2. Available at: http://www.mtc.ca.gov/maps_and_ data/GIS/data.htm (accessed March 2011). Schrag, Z. M. The Great Society Subway: A History of the Washington Metro. The Johns Hopkins University Press, 2006. Shapiro, M. Group Seeks to Delay Rosecroft Line. Washington Post, Feb. 27, 1980. Small, K. Urban Transportation Economics. Hardwood Academic Pub- lishers, Chur, Switzerland, 1992.

2-56 Walk Score. Transit Score Methodology. Available at: http://www.walkscore. com/transit-score-methodology.shtml (accessed December 2012). Wallace, D. A., I. McHarg, W. H. Roberts, and T. A. Todd. Preliminary Evaluation Report: Branch (F) Route. The Washington Metropoli- tan Area Transit Authority (WMATA), 1976. Washington State Department of Transportation. Congestion Report, 2011. Available at: http://www.wsdot.wa.gov/Accountability/Congestion/ (accessed March 2011). Washington Metropolitan Area Transit Authority (WMATA). F Route Comparative Alignment Study, 1984. Washington Metropolitan Area Transit Authority (WMATA). Summary: Final EIS and Section 4(f) Statement, Washington Regional Rapid Rail Transit System, Outer Branch Avenue Segment, 1993. Washington Metropolitan Area Transit Authority (WMATA). Development-Related Ridership Survey, 2005. Washington Metropolitan Area Transit Authority (WMATA). Draft Supplemental Environmental Impact Statement—Washington Regional Rapid Rail Transit System—Green Line (F) Route Outer Branch Avenue Segment, Sections F-6 through F-11, 1992. Weinberger, R. Light Rail Proximity: Benefit or Detriment in the Case of Santa Clara County, California? Transportation Research Record: Journal of the Transportation Research Board, No. 1747, January 1, 2001, pp. 104–113. doi:10.3141/1747-13. Winston, C., and V. Maheshri. On the Social Desirability of Urban Rail Transit Systems. Journal of Urban Economics Vol. 62, No. 2, 2007, pp. 362–382. doi:10.1016/j.jue.2006.07.002. Yeh, C. H., H. Deng, and Y. H. Chang. Fuzzy Multicriteria Analysis for Performance Evaluation of Bus Companies. European Journal of Operational Research, Vol. 126, Issue 3, 2000, pp. 459–473. Yin, R. K. Case Study Research: Design and Methods, 4th ed. Sage Pub- lications, Inc., 2008. Zillow, Inc. Zillow.com. Available at: http://www.zillow.com/ (accessed March 2011). Zupan, J., and Cervero, R. Transit and Urban Form. Vol. 1, Part 2: Com- muter and Light Rail Transit Corridors: The Land Use Connection. Washington D.C.: National Academies Press, 1996. Urban Mass Transportation Administration (UMTA). Dallas North Central Subarea Transit Alternatives Analysis: Draft Environmental Impact Statement, 1982. U.S. Census Bureau. Decennial Census 2000 Summary File 1 and 3. U.S. Census Bureau. American Community Survey, 1-Year Estimates, 2005–2009. U.S. Department of Transportation. Final Environmental Impact State- ment: Metropolitan Washington Regional Rapid Rail Transit System, 1975. U.S. Department of Transportation. Bicycle and Pedestrian Data: Sources, Needs, & Gaps. BTS00-02. Bureau of Transportation Statistics, Wash- ington, D.C., 2000. U.S. Department of Transportation. Table 1-70: Travel Time Index in the National Transportation Statistics. Bureau of Transportation Statistics, Washington, D.C., 2000. Available at: http://www.bts. gov/publications/national_transportation_statistics/html/table_ 01_70.html. U.S. Department of Transportation and Washington Metropoli- tan Area Transit Authority (WMATA). Draft EIS: Washington Metrorail System, Branch/Rosecroft (F) Route. Urban Mass Transportation Administration and WMATA, Washington, D.C., 1979. U.S. General Accounting Office. Bus Rapid Transit Shows Prom- ise. Report to Congressional Requesters, U.S. General Account- ing Office, 2001. Available at: http://www.gao.gov/new.items/ d01984.pdf. Utah Transit Authority. 2005. Weber County to Salt Lake City Com- muter Rail Project: FEIS. Voith, R. Changing Capitalization of CBD-Oriented Transportation Systems: Evidence from Philadelphia, 1970–1988. Journal of Urban Economics Vol. 33, No. 3, May 1993, pp. 361–376. doi:10.1006/juec. 1993.1021. Waddell, P. Forecasting Land Use Activities (1): The Evolving State of the Practice. U.S. Department of Transportation Travel Model Improvement Program (TMIP) website. Available at: http://tmip. fhwa.dot.gov/webinars/flua1 (accessed March 2011).

A p p e n d i c e s Appendices A–J were prepared in conjunction with the final report. The TCRP Project H-42 Draft Spreadsheet Tool: Estimated Ridership and Cost of Fixed- Guideway Transit Projects provides the analytical model in Excel format. The spreadsheet tool is not published herein, but is available separately for download from the report web page at www. trb.org by searching for “TCRP Report 167”. Appendices A–J and Related Material

A-1 Appendix A Data Source Review A-1 A.1 Data A-2 A.2 Internal Attributes A-4 A.3 Extensive Attributes A-9 A.4 Unconventional Data Sources A-12 A.5 External Attributes B-1 Appendix B Data Collection and Construction of Variables B-2 B.1 Measures of Transit Project Success B-3 B.2 Predictors of Transit Project Success: Metropolitan Area B-7 B.3 Predictors of Transit Project Success: Project-Level B-8 B.4 Additional Variables Considered c-1 Appendix C All Fixed-Guideway Transit Projects in the United States d-1 Appendix D Network Measures e-1 Appendix E Variables List F-1 Appendix F Fixed-Guideway Projects Included in Analysis G-1 Appendix G Model Technical Information H-1 Appendix H Focus Groups, Phase 2: Topic Responses i-1 Appendix I Detailed Case Study Write-Ups and Regional Profiles I-1 I.1 South Line (Charlotte, NC) I-10 I.2 North Central Corridor (Dallas, TX) I-18 I.3 EmX Phase I Bus Rapid Transit (Lane County, Oregon) I-24 I.4 Interstate MAX (Portland, OR) I-34 I.5 University & Medical Center Extensions (Salt Lake City, UT) I-41 I.6 Branch Avenue Extension (Washington D.C., Prince George’s County, MD) I-51 I.7 Regional Contexts J-1 Appendix J Data Sources C O N T E N T S

A-1 APPENDIX A: Data Source Review A.1 Data The measures and predictors of transit success addressed in the literature cover a range of transportation issues. Using these measures to evaluate transit systems requires a variety of quantitative datasets covering regions where fixed-guideway transit projects are proposed and delivered. Nonetheless, there is little study that systematically and comprehensively links the measures and predictors of transit success to the many secondary data sources. This lack of integration among various national, regional, and local data sources, in addition to the lack of quantitative data on unconventional transit technologies and policies, prevents federal, state, metropolitan, and local decision makers from incorporating more innovative measures and predictors of transit success into their project evaluation practices. This section thus attempts to match the measures and predictors of transit success that are described in the literature review with readily available data sources in the United States and to identify some measures and predictors that are not sufficiently available in existing U.S. nationwide databases. There are certain challenges to developing one comprehensive database for measures and predictors of transit success. Foremost are inconsistencies across different data sources and the general lack of information on newer transit technologies and policies. A variety of readily available databases on transportation, land use, economic elements, and social characteristics are managed by several public, private, non-profit, and academic entities for different purposes. This review lays out the similarities and differences in how these nationwide data sources define significant variables (e.g., transit capital, operation, maintenance, disposal, or life cycle costs and transit agency or private household expenditures), measure the scope of analysis (e.g., state, transit agency, metropolitan area, census tract, census block or household), measure survey periods (e.g., short-, mid- or long-term), and the frequency at which information is updated (e.g., decennially, annually, quarterly, or monthly). In addition, many of the readily available nationwide databases on fixed- guideway transit systems in the United States do not yet satisfactorily cover contemporary transportation attributes, such as geographic scope/network, intermodal features, bus rapid transit (BRT) systems, parking, and urban design characteristics, across national, regional, local, and human scales. To fill the knowledge gaps, some local data sources and alternative analysis approaches are listed to complement each of these under-documented transportation attributes. One of the most difficult practices for transportation researchers and decision makers is to conceptualize in a logical order all of the various datasets and their broad ranges of measures and predictors of transit success. To aid in this process, this review re-organizes the secondary data sources and discusses the challenges to computing measures and predictors of transit success. The review breaks down the measures and predictors into four categories: (i) internal (cost and system/finance data); (ii) extensive (social, geographic/network and intermodal data); (iii) unconventional (BRT, parking and urban design data); and (iv) external (urban development data such as residential and business location, property transaction, land use integration, and urban

A-2 simulation). A summary table of the key data sources reviewed in this section is attached in Appendix J. A.2 Internal Attributes Fixed-guideway transit projects have long been evaluated primarily by the cost-performance data managed and reported by traditional operators. Computations of such internal performance measures and predictors call for appropriate datasets on transit costs, supplies and demands, and funding sources. A.2.1 Cost Data The costs associated with fixed-guideway transit systems can be differentiated among those held by internal, extensive, and external stakeholders. Internal cost components, which are reported by large fixed-guideway transit agencies, are relatively clear to identify and address both measures and predictors of transit success. On the other hand, extensive and external cost components, which are covered by small transit operators, individual households, specific organizations, or local jurisdictions, are generally challenging to define as measures of transit success in a broader social context. Here we discuss the availability and limitations of datasets that describe internal cost components. These can be subdivided into transit expenditures of capital, operations, maintenance and disposal for the short-term phases and the long-term life cycle of a fixed-guideway transit project. A number of readily available nationwide data sources cover U.S. transit agencies’ capital and operation expenditures for short-term phases. At the aggregate agency or metropolitan level of analysis, the National Transit Database (NTD) and the APTA Public Transportation Factbook (PTFB) both offer summary-level information on annual transit agency expenses since 1996 and 2003, respectively. Both of these sources provide statistics that allow data on transit agency internal expenditure to be normalized by transit facility and service factors, including rail track miles, number of vehicles, service hours, total passengers, or service areas. Since annual data on transit fare prices or farebox revenues are also included in the NTD/PTFB tables, the magnitude of cross- subsidies can be roughly estimated as the net difference between annual-based transit fare revenues and total transit expenditures at the aggregate transit agency and metropolitan-area levels. While there is much information on capital and operating costs at the aggregate level, the need for disaggregate cost data rapidly increases as transportation researchers and transit managers recognize the essential roles that space, direction, and time play in determining transit agencies’ expenditures. This effect can be attributed to the phenomenon of peaking, which Taylor (2004) identifies in his following conclusions: (i) providing both demand-driven and policy-driven network patterns tends to be expensive (spatial peaking); (ii) the marginal costs of adding transit services in peak directions to/from downtown and other employment centers tend to be high (directional peaking); and (iii) the marginal costs between morning, midday, evening, and midnight hours on weekdays and weekends are different (temporal peaking). Despite the fact that space, direction and time characteristics significantly determine the internal performance of transit systems and transit finance programs, the lack of nationwide databases on disaggregate expenditures impedes the use of more accurate internal measures and predictors of success in transit costs. The long-term maintenance, disposal, and life cycle costs of a fixed-guideway transit project have historically been overlooked and are rarely reported by traditional public transit agencies. The documentation of these non-traditional cost accounts or “asset management records” has recently

A-3 become more crucial, however, as public-private partnerships are increasingly utilized to deliver mega-infrastructure projects and avoid cost overrun problems in the United States (Maze and Smadi 2003; Neumann and Markow 2004; Miller and Ibbs 2000; Flyvbjerg 2007; Flyvbjerg et al. 2003). One important piece of information related to these lifetime cost accounts is social and business discount rates. These data, which can be found in sources like the Board of the Governor of the Federal Reserve System’s Statistics & Historical Data on Interest Rates (1954), provide information on the long-term measures and predictors of transit success in internal costs. A.2.2 System and Financial Data By and large, transportation professionals and public decision makers in the United States consider success based on the internal measures of transit system performance and financial program performance (Taylor 2004). To measure transit system performance, the internal costs paid by transit agencies are normalized by transit service supply and passenger demand variables. Using finance programs to measure transit success requires information on the internal structures of transit revenues and cross-subsidies from several public funding resources (e.g., fuel taxes, sales taxes, property taxes, land and air-right sales, toll revenues, and obligation bonds). The data necessary for both of these performance measures are discussed in this section. Transit service supply, which is used to measure transit system performance, can be characterized in a number of ways, such as facility and vehicle counts; travel times and speeds; service capacity, frequency and hours; reliability and comfort; fare policies and technologies; network patterns; and market coverage (TCRP Report 100: Kittelson & Associates Inc. et al. 2003). However, many of these service attributes are hard to quantify in a comparative and in-depth way due largely to the paucity of nationwide disaggregate datasets. Two of the most common national databases on transit supply attributes, both tracked at the aggregate agency and metropolitan levels, are the APTA Public Transportation Factbook (PTFB) and the National Transit Database (NTD). These sources include information on vehicle availability, track and service lengths, transit vehicle speeds, fares, and service areas. Disaggregate datasets on service supply attributes (by space, direction, and time) can be extracted from online information systems, annual operation reports, and internal survey documents provided by individual transit agencies, but the service attributes and periods covered are often inconsistent across different transit agencies. The Massachusetts Bay Transportation Authority (MBTA) and Bay Area Rapid Transit (BART) are two examples of transit agencies that make extensive information available, including data on service reliability, passenger environment indicators, train cleanliness, customer complaints, crime, etc. The PTFB and NTD tables contain only aggregate data on passenger demand characteristics (e.g., unlinked passenger trips, trip lengths, and passenger miles traveled), whereas disaggregate ridership patterns at corridor and station levels must be obtained from the individual agency reports published annually or monthly. Because of this discontinuity among national and local data sources, it is difficult to compile and compare peaking transit demands by zone, direction, and time, which critically influence the internal structures of marginal expenditures, fare revenues, and cross- subsidies. To study the internal structures of fare revenues and cross-subsidies, the NTD table contains many transit finance variables: public funds used to pay back interest and principal on bonds and loans; capital program funds; carryover amount to next year; state and local government contributed services; passenger fares earned by mode and type of service; gasoline tax amount and percentage; the general revenues of the government entity; revenues earned from high occupancy toll lanes; investment revenue and non-transportation funds; park-and-ride parking revenue; and income,

A-4 property, and sales tax amounts and percentages. These fare revenue and cross-subsidy attributes are very important for decision makers to understand because public funding resources not only help reduce the financial deficits of transit agencies but also extensively redistribute the internal costs of fixed-guideway transit systems between internal transit users and extensive social stakeholders based on the benefit principle and the ability-to-pay principle (Taylor 2004; Musgrave 1959). A.3 Extensive Attributes The internal measures and predictors of fixed-guideway transit success suggested in the literature are usually associated with equity concerns, urban geographic patterns, or regional network performances. Accounting for such extensive attributes, existing databases on social, geographic information system/network, and intermodal characteristics need to be well understood. A.3.1 Socioeconomic Data In practice, public funding decisions are guided not only by the internal measures and predictors of success in transit systems and finance programs, but also by the extensive debates over four types of social equity: individual equity, environmental equity (or justice), group equity, and geographic equity (Taylor 2004; Cairns et al. 2013). As reviewed, fixed-guideway transit projects are likely to redistribute both user benefits and social costs among different individuals and groups in certain jurisdictions and geographies. Therefore, there is a growing need for the organization and integration of nationwide secondary data on the measures of transit success through social welfare attributes: (i) transportation and housing affordability; (ii) public health and safety; (iii) socioeconomic diversity; and (iv) geographic accessibility. According to classic theories, fixed-guideway transit investments change individual and household transportation and housing affordability in regional and local spaces. Individual and household information on transportation and housing costs, representing measures of transit success in affordability, can be gained from several U.S. government surveys and data packages: Decennial Population and Housing Census; American Community Survey (ACS); Census Transportation Planning Package (CTPP); National Household Travel Survey (NHTS) and Nationwide Personal Transportation Survey (NPTS); and American Housing Survey (AHS). These nationwide data sources complement one another with a range of datasets about personal earnings, household incomes and expenditures, housing prices and rents, transportation expenditures and commuting times for different years at several geographic levels. Some nationwide private data services, such as ESRI Updated Demographics and GeoLytics 2001-2008 Demographic Data, cover similar household income, housing expenditure and commuting cost variables to fill the gaps in the above public databases. While most of the housing variables in these public and private data sources are neither temporally nor geographically standardized, the U.S. House Price Index (HPI) data, with a special focus on housing affordability, are annually and regionally comparable over the last few decades at the census region, state, and metropolitan statistical area (MSA) levels. The Center for Neighborhood Technology (CNT)’s Housing +Transportation (H+T) Affordability Index website further provides a nationwide interactive map with computed 2000 and 2008 affordability variables (such as transportation and housing costs standardized by income, gasoline and housing expenditures, transit service indices, travel times, and household incomes) at the region, county, city, and census block group levels.

A-5 Another social welfare measure of success is a fixed-guideway project’s ability to decrease the negative effects on public health and safety of an auto-dependent society. In the United States, various databases on public health and safety are organized by the Centers for Disease Control and Prevention (CDC) at the national or regional aggregate level. The CDC’s Behavioral Risk Factor Surveillance System (BRFSS) began an annual survey in 1984 to assess health-related risk factors by calculating health condition, physical behavior, and social status variables of selected metropolitan/micropolitan areas (SMART). The National Health Interview Survey (NHIS) is another aggregate data source provided by the CDC’s National Center for Health Statistics (NCHS) since 1963, covering injury and poison condition variables by family, household, person, and age categories. In addition, the CDC maintains the National Vital Statistics System (NVSS) that annually reports all deaths by cause (e.g., motor vehicle) and circumstance (including fall and collision with motor vehicle, animal, bicycle, pedestrian, or fixed object), as well as the National Hospital Ambulatory Medical Care Survey (NHAMCS) that contains a sample of injuries by cause. The National Highway Traffic Safety Administration (NHTSA) also manages the Fatality Analysis Reporting System (FARS), which contains data on more than 100 accident, vehicle, and person variables since 1975, and the National Automotive Sampling System: General Estimates System (NASS GES), which covers a nationally representative sample of police-reported motor vehicle accidents of all types since 1988. The social costs mentioned above are unevenly reduced or redistributed by fixed-guideway transit projects among a variety of socioeconomic groups and local jurisdictions along regional transportation networks. Unfortunately, these referenced nationwide secondary databases on public health and safety are not directly linked to other databases on transit systems, socioeconomic characteristics, and micro-geographic boundaries, making it difficult to evaluate fixed-guideway projects on the basis of success with regard to environmental justice. Socioeconomic diversity across the United States can be measured by a variety of characteristics in readily available government sources, as well as in customizable private data services, for different survey periods. However, the unit of analysis used to measure socioeconomic diversity must be carefully chosen from among various options, such as income, age, gender, educational status, race/ethnicity, religion, occupation, and place of residence. These groups share the user benefits and social costs of fixed-guideway transit projects within a certain jurisdictional/geographic boundary. None of the nationwide databases reviewed in this report calculates any socioeconomic mixture indices at a given jurisdictional level. In recent databases (e.g., National Dataset for Location Sustainability and Urban Form), socioeconomic variables are instead associated with transit network, regional employment, and local service accessibility on multiple geographic information system (GIS) tabulations. A.3.2 GIS and Networking Data Geographic information systems have grown to play an important role in transit research. This is because transit systems operate in various economic and social geographies, and their performance is strongly related to spatial characteristics. GIS software uniquely enables the analysis of the spatial relationships that occur between transit facilities and their surrounding communities. Many popular indicators of successful fixed-guideway transit can be derived from manipulation of GIS shapefile data sets. Most prominently for transit analysis, GIS can be used to estimate the amount of population or employment that is located within a radius or street network “access shed” of a transit facility. It can also be used to measure accessibility to jobs, retail, and other destinations by different modes.

A-6 Although GIS systems enable a broad array of spatial data processing and analysis techniques, it is important to note that products derived through GIS applications are only as useful as the input data and the theoretical approach to understanding spatial relationships. In particular, GIS representations of population or employment are spatial averages of survey, forecast, or estimate data. Not only is the accuracy of the data often questionable, but it often represents finite discrete attributes in a spatially flattened context. Population or other land use characteristics can be spatially clustered or broadly dispersed within a geographic unit of analysis, but GIS techniques are insensitive to these distinctions. The geometry of transportation analysis zones (TAZ)—a common unit of analysis in regional transportation models—is particularly insensitive to the pedestrian scale features of a neighborhood that could facilitate pedestrian activity and transit use or render it infeasible. GIS shapefiles are produced both by public governments or agencies and by private parties. Many useful shapefile datasets can be downloaded for free from state or regional geographic data clearinghouses. Possibly the most commonly used shapefiles are the U.S. Census TIGER/Line shapefiles. These shapefiles, available for download from the U.S. Census Bureau's website, feature political boundaries, census tabulation boundaries, and basic street networks. Numerous private street network datasets are available, but they are usually expensive. Fixed-guideway transit shapefiles are available as part of the National Transportation Atlas Database (NTAD). NTAD shapefiles can be downloaded for free from the Internet and feature various transportation-related shapefiles on a nationwide basis, but not all transit systems are included in NTAD. Many metropolitan planning organizations (MPOs), DOTs, and transit agencies provide more complete transit network shapefiles, possibly including bus route information. Another option for geographically representing transit station locations is the manipulation of General Transit Feed Specification (GTFS) data feeds. The GTFS format was developed by Google for transit agencies to make route and schedule information publicly available on Google Maps. The text-based GTFS format is now becoming a standard data format for expressing transit service information. Available GTFS data differ by transit agency, but they typically include latitude and longitude location fields that can be used to geocode station locations in GIS applications. Schedule and route information in GTFS format can also be used to obtain data on transit service characteristics. Another useful resource for publicly available geospatial data is a national GIS portal supported by the Geospatial One-Stop E-Government initiative: www.geodata.gov. The characteristics of, and differences between, various transit networks are of unique interest. As transit networks expand, the expanded accessibility offered by a system as a whole is expected to improve the desirability of all stations and thus increase ridership across existing stations. Despite the significance of network characteristics, methods of directly measuring such characteristics are not widespread. This section presents some tools and challenges related to preparing data to represent network characteristics of fixed-guideway transit systems. Several issues that complicate transit network characteristics measurement are immediately apparent. First, the format in which most fixed-guideway transit system geospatial information is made publicly available is not conducive to easy network analysis. This would require a complete and integrated set of route and station GIS shapefiles. In practice, when shapefiles of transit networks are readily available, stations are not usually placed as nodes at link vertexes and links often do not relate to connections between nodes. GIS data for fixed-guideway transit systems vary in format, but often they do not include a unique feature for each link between two stations. Often, a single polyline represents an entire transit corridor, or multiple polyline features start and end at curves and other geographic features not related to station location. As a result, some amount of

A-7 manual editing and data manipulation is often necessary before network analysis can be conducted with GIS shapefiles. There are different possible measures of network size or density. The simplest is to count the total number of stations or route-miles. This method represents network size but does not give any indication of a network’s connectivity or accessibility between origins and destinations. Link-to- node ratios relate system-wide station counts to interconnectivity between stations. This is a measure of network density. The rationale for this measure is that good connectivity can be defined as routes offering more linkages between any given fixed number of stations. But it does not distinguish some systems that vary greatly in connectivity; for example, the link-to-node ratio of a single-line system might differ only slightly from a multi-route system that converges on a multiple- line transfer hub in the central business district. Due to GIS issues discussed above, link-to-node ratios are often easiest to calculate by hand through observation of a transit route map. Another option for quantifying network scale and scope is measuring the population, employment, and activity center characteristics of places that can be accessed by the transit system. At a system-wide level this can be done using GIS to create service area zones around stations and then calculating the population and other features of the areas within the service area. Service area populations can be calculated by associating census or third-party data with Census TIGER/Line shapefiles, as discussed in the GIS section above. For localized transit demand modeling, a similar approach can be performed for each station where land uses at other stations that are accessible by the fixed-guideway transit are calculated for each individual station. Numerous other methods are available for analyzing network characteristics of transit systems, but many in reality only apply to networks that provide coverage across broad service areas. Daganzo (2010) introduced methodologies for determining ideal network layout based on population, travel demand, and service area geography conditions. Under Daganzo’s framework, transit systems can form radial networks, grid networks, or a hybrid central grid and radial periphery system. The underlying characteristics of a region can be analyzed to determine which system type would optimally serve a specific service region. In short, the best network is the one that best matches the distribution origins and destinations in a given city. A.3.3 Intermodal Characteristic Data Fixed-guideway transit lines are often a part of wider national, state, regional, and local networks consisting of other transportation modes. This implies that the congestion levels, market shares, nodal facilities, and feeder services for a given fixed-guideway service not only influence that service’s internal measures of success, but they also change the extensive measures of success for other modes throughout the network. Such intermodal characteristics need to be systematically and comprehensively analyzed as measures and predictors of transit success by federal, state, regional, and local transportation decision makers. To facilitate this process, we organize a number of nationwide databases into the following four categories: (i) urban mobility on regional roadway systems; (ii) modal competitiveness in national, state, regional and local transportation markets; (iii) intermodal connectivity among regional airport, waterway, and fixed-guideway transit systems; and (iv) local access availability around urban transit centers and along suburban transit corridors. Urban mobility on regional roadway systems is annually reported by the Texas Transportation Institute (TTI) and Federal Highway Administration (FHWA). The TTI’s Urban Mobility Report (UMR) provides aggregate estimates of several traffic mobility and congestion variables (e.g., daily vehicle miles traveled, annual travel delay, travel time index, and overall congestion cost) for approximately 100 select urban areas. For some urban areas, however, UMR data is inconsistent

A-8 with secondary data from state DOTs and metropolitan planning organizations. The other annual resource on urban mobility, the FHWA’s Highway Statistics, covers a larger number of urbanized areas (more than 400 in 2009) and includes aggregate estimates of daily vehicle-miles traveled by hierarchical roadway systems (including Interstate, other freeway, and expressway, other principal arterial, minor arterial, and collector). One disadvantage of this database is that, due in large part to changes in the definition of U.S. urbanized areas, the Highway Statistics’ panel data on aggregate estimates of daily vehicle-miles traveled are inconsistent from 1992 through 2009. In order to thoroughly investigate urban mobility characteristics at local roadway network and corridor levels, transportation analysts and decision makers need to rely on state-based or metropolitan organization-based disaggregate data sources (e.g., the California DOT’s PeMS, the Florida DOT’s TranStat, and the Washington DOT’s Congestion Report). Modal shares for commute trips are covered by the U.S. decennial census and the annual community and housing surveys for different statistics periods at different geographic levels. Primarily, the U.S. DOT’s Census Transportation Planning Package (CTPP) contains data on travel mode to work, allocating commuters among 18 different means of transportation to work categories in the 1990 and 2000 decennial censuses at the state, MSA, county, census tract, and transportation analysis zone (TAZ) levels. Secondarily, CTPP products based on the most recent American Community Survey (ACS) classify workers into 10 categories of means of transportation to work based on the 3-year (2006-2008) and 5-year (2006-2010) tabulations at the state, MSA, county, place, and public use microdata area (PUMA) levels. In addition, the U.S. Census Bureau annually conducts the American Housing Survey (AHS) to obtain up-to-date housing and household characteristics, including data on the numbers of housing units in 12 categories of principal means of transportation to work from 1973 to the present year on the MSA, county, central city/suburban status, and census tract scales. These secondary data sources, however, do not account for non- commute, daily chain, and long-distance trips. This considerably limits the understanding of intermodal competition in the national, state, regional, and local transportation markets where fixed- guideway transit projects are proposed. In the 2001-2002 and 2009 National Household Travel Surveys (NHTS), the U.S. DOT Bureau of Transportation Statistics (BTS) legislation addressed this deficiency. It specifically targeted data on the volumes and patterns of both long-distance and local- based travel for multiple non-commute purposes for 25 different modes of transportation at the census tract, block group, and household levels in order to analyze and evaluate the nation’s new capital investments. Intermodal connectivity indices among different passenger transportation systems can be computed by using the GIS point and line shapefiles included in the BTS/RITA’s National Transportation Atlas Database (NTAD). Point shapefiles contain locations such as airports, Amtrak stations, fixed-guideway transit stations and intermodal terminal facilities; line shapefiles contain networks such as national railways, fixed-guideway transit lines, and navigable waterways. The nationwide NTAD database has been issued annually since 1996 (it is currently available for 2008, 2009, and 2010); however, NTAD has failed to consistently update its point and line shapefiles as new capital investments occur. Complementary to NTAD, the BTS/RITA’s Intermodal Passenger Connectivity Database provides a nationwide table of passenger transportation terminals with data on the availability of intercity and commuter rail, air, and ferry services. This database has more frequently updated its files with new facility and service information since mid-2006, and heavy and light rail transit stations were added to this database in 2011. The scope of the Intermodal Passenger Connectivity Database is wide, including intercity buses, code-share buses, and

A-9 supplemental service buses for intercity rail and air carriers, intercity ferries, and transit or local ferries, but it does not attempt to cover every possible transit bus stop in every street block. Local information on feeder transit systems can be obtained in General Transit Feed Specification (GTFS) data for many local systems. Various transit operators in North America occasionally update their local service characteristics in the GTFS format (e.g., stations, stops, routes, transfers, runs, hours, frequency, and fares), but there is no comprehensively integrated nationwide database on local access availability around regional transit centers and corridors. Key data challenges arise around bus rapid transit (BRT) systems, parking facilities and policies, and pedestrian/bicycle facilities and amenities. A.4 Unconventional Data Sources Contemporary funding programs and academic studies have stressed the increasing importance of incorporating new factors and system types when evaluating fixed-guideway transit projects. We discuss BRT, parking, and urban design below. A.4.1 BRT Data In recent years, U.S. transit agencies have become increasingly interested in BRT systems. BRT systems are considered an attractive option for improving transit service due to perceived affordability and flexibility. There are numerous international examples, particularly in Latin America, where BRT systems offer high levels of service at costs far below those of rail-based alternative modes. As a result, U.S. agencies see BRT as a potential tool for improving transit service in environments where capital for transit expansion is limited. Despite the growth of interest in BRT systems, obtaining data to study their potential success in U.S. environments is challenging. The two primary data constraints are the limited number of BRT systems operating in the United States and the recent nature of the systems that are in place. Due to these conditions, transit agencies considering the possibility of developing BRT systems have few comparable applications and limited historical data that can be analyzed. Meanwhile, although BRT systems in Bogota, Columbia; Curitiba, Brazil; Guangzhou, China; and numerous other global cities are widely considered successful, these cities often feature social, political, cultural, and economic differences from the United States that are so dramatic that any direct comparison would be problematic. Generally speaking, domestic experiences are available for the study of short-term responses to some forms of BRT in the United States, but long-term United States experience does not yet exist. The prediction of long-term BRT success may therefore require some form of international comparative analysis. Besides the limited availability of data, there are also compatibility issues across BRT systems that are in place in the United States. Firstly, the concept of BRT is poorly defined. BRT is actually a toolbox of features that can be used to improve service and increase commercial operating speeds. A fully featured BRT system is one that includes most or all of the BRT toolbox features. The literature on BRT differentiates between “full BRT” and “BRT-light.” Although many U.S. transit agencies claim to operate BRT services, most examples are actually limited to bus routes with relatively higher service frequencies and longer stop spacing than regular bus services that operate in mixed traffic conditions. Of bus systems that do qualify as “full BRT” or “BRT-light,” there are significant differences in features that make direct comparisons problematic. By definition, a bus system qualifies as BRT if and only if a separated right-of-way is present on at least part of its corridor. BRT systems that currently exist in the United States are listed below:

A-10 • LA MTA Orange Line • Boston Silver Line • Eugene, Oregon, EmX Line • Cleveland Health Line • NY MTA NYCT Fordham Road, 34th St., and 1st/2nd Aves Lines • Pittsburgh The relatively recent arrival of BRT applications in the United States precludes the possibility of longitudinal analysis of its long-term measures of success. However, short-term before-and-after analysis can be performed for most extant systems using data on service features and ridership at one time point before implementation and another following commencement of BRT operations. While land use patterns cannot be expected to adjust to service changes in a period ranging from mere months to a couple of years, short-term ridership changes can indicate the degree to which BRT service improves transit competitiveness for existing travel demand patterns. Since one of the main driving factors behind BRT in the United States is that it is perceived as a more affordable alternative to rail transit, it is essential to gain a better understanding of its comparable costs and competitiveness. BRT capital and operational costs can be compared to rail systems on a mileage or passenger basis. This requires individual consideration of recent rail and BRT projects in the United States, a topic that recent literature has begun to address. Regarding competitiveness with rail, there are several metrics that should be considered. Firstly, a key concern of BRT skeptics is that BRT is less attractive to U.S. travelers than rail-based modes. This issue can be addressed through ridership studies comparing BRT and light rail systems in corridors of similar demographics. Other issues of concern include operating speeds, service reliability, required real estate footprint, and capacity constraints. Due to a lack of available information necessary to answer the above questions, some comparative international analysis may be useful in predicting BRT success in the United States. Although Latin American cities have advanced BRT systems with long operational histories, the major differences between the United States and Latin America with respect to socioeconomic attributes, land densities, car ownership rates, culture and other factors complicate our ability to effectively apply Latin American lessons to U.S. systems. In order to better predict BRT desirability, operational limitations, user acceptance, capital costs, and operational expenses it would be ideal to consider examples in societies that are structurally similar to the United States. Canada is one such example, with comparable demographics, economic development and culture, in addition to a relatively extensive history of BRT experimentation. Ottawa, Canada, has over 30 years of experience with a full BRT system. BRT routes have also been in operation in Calgary since 2004, although one line is currently being converted to light rail. The cost structure, ridership, and long-term land use impacts of these systems could be compared to Canada’s recent urban rail systems to inform expectations of BRT success in the United States. A.4.2 Parking Data Parking is considered an important factor in the success of fixed-guideway transit systems in multiple ways. Firstly, the availability and cost of parking is an element of overall user costs experienced by drivers of private vehicles. In dense urban centers the cost of parking can amount to a significant share of the overall cost and convenience of auto travel. These costs are incorporated into travel mode decisions that ultimately determine transit demand. When parking is located at travel destinations, it can be analyzed as part of the infrastructure supporting a mode that competes with fixed-guideway transit.

A-11 Parking supply can also serve as an important element of intermodal transit stations. Suburban fixed-guideway transit stations are often particularly dependent on parking capacity to facilitate commuter station access. While parking can act as a competing mode at trip destinations, the lack of parking availability at transit stations can instead constrain ridership. Parking is not only a facility that can be present at a transit or activity node, but it also occupies physical space and has significant urban design attributes. Principally, the use of land for automobile storage can crowd out other activities, particularly in downtowns or destination areas where parking facilities occupy land that could otherwise serve different purposes. The result is a reduced density of other land uses that are supportive of and supported by transit service. Parking facilities are also often visually unappealing and may reduce the attractiveness of pedestrian environments. Meanwhile, at transit facilities where parking may be present in order to facilitate access, the use of land for parking lots or structures reduces the availability of land for transit- oriented development. The supply of parking is often more complex than a private market reaction to demand. Most communities in the United States regulate the supply of parking and mandate some bundling of parking supply with other land uses. In this regard, local knowledge is critical for understanding the dynamics behind parking supply, ownership, and pricing in any given urban environment. Despite being an important element in mode choice decisions, it is very difficult to acquire broad, aggregate information on parking availability and pricing. In fact, most municipalities have no inventory of parking capacity outside of their central business district (CBD) parking supply. Even at the local level, accurate information of parking supply is very difficult to find. Fortunately, in the case of park-and-ride facilities operated by transit agencies, parking capacity and pricing information can usually be collected from the agency itself with modest expense of effort. For a unit of analysis at the metropolitan level, some private studies are also available to provide order of magnitude approximations of CBD parking prices. Two examples of these types of sources are Colliers International’s North America Central Business District Parking Rate Survey and National Parking Association’s Parking in America Report. If a more detailed study is necessary, several websites provide parking prices at various parking garages (e.g., www.bestparking.com). None of these data sources provide information about the supply, costs, or availability of on-street parking, the personal use of private parking, or the temporal distribution of demand, but the private parking lot prices that they do present can give an indication of the interaction of demand with land values (given local zoning ordinances associated with parking provision). In preparation for the implementation of an information-technology driven adjustable rate parking system, the City of San Francisco performed an inventory of on-street and publicly owned parking garage capacity. The data generated from this SFPark program will prove invaluable for future analyses of parking supply and prices as predictors of transit success. A.4.3 Urban Design Data There is increased emphasis on urban design elements around fixed-guideway transit stations, as planners and policymakers recognize the importance of transit-supportive land uses and place- making efforts in promoting transit ridership, pedestrian/bicycle travels, public health and safety, community livability, social interactions, and economic innovations. Nevertheless, incorporating urban design criteria as predictors of transit success into fixed-guideway transit project evaluation is often hampered by a lack of nationwide data on human-scale built environments. Better information on the built environment around fixed-guideway transit stations might include types of public space;

A-12 street facilities; street locations, lengths, widths and physical conditions; street amenities; topography; and intersection/network characteristics. The most common approach to measuring urban design characteristics is to compute the connectivity of local street networks within one-quarter- and/or one-half-mile of a fixed-guideway transit station using the U.S. Census Bureau’s TIGER/Line GIS shapefiles. Despite its geographical comprehensiveness, analytical ease and practical usefulness for transit project evaluation at the local street level, the TIGER/Line shapefile application does contain limitations. A secondary data review conducted by the U.S. DOT Bureau of Transportation Statistics (BTS) in 2000 addressed two of these drawbacks: (i) the file does not contain any facility attributes (such as street widths, number of lanes, and presence of sidewalks); and (ii) the file does not contain pedestrian or bicycle connections that are not part of the street network (such as alleys, walkways, or pathways). In short, further details of transit-supportive design characteristics need to be frequently and accurately updated on the nationwide street-level GIS map. Indeed, progressive cities, counties, and metropolitan planning organizations independently establish and maintain their own GIS databases to describe the unrecorded bicycle and pedestrian amenities of fixed-guideway transit station areas, including sidewalk/bikeway continuity, street connectivity, topography, and other urban design elements (e.g., the city of Portland’s Corporate GIS, the North Central Texas’s Rail Station Access, and the San Francisco MTC’s GIS Data Category 2). In early 2011, the National Dataset for Location Sustainability and Urban Form became readily available through the Natural Resource Ecology Laboratory at the Colorado State University. Based on a road shapefile from the U.S. Census TIGER/Line 2009, this nationwide GIS database computed the weighted number of intersections within one-quarter-mile of each census block group as an urban design factor. In the technical report of this dataset, Theobaldi et al. (2011) noted that a methodological challenge to calculating this design variable is that either adjacent land uses (e.g., a park) or transportation corridors (e.g., railway line or Interstate highway) often limit pedestrian access, yet such physical barriers are not accounted for in the variable. Ideally a number of site- specific information would be included for each of these variables, such as sidewalk completeness, directness of pedestrian routes, and bicycle pathways. However, these local built environment attributes are hard to cover and update on a national scale and in a timely manner. Facing this insufficiency of nationwide databases, the BTS report (2000) recommended: (i) standardizing formats and definitions of urban design characteristics to improve data comparability among local agencies and geographic areas; (ii) facilitating discussions among various data user groups to identify key urban design characteristics and provide guidance to state and local agencies responsible for collecting and maintaining data; and (iii) applying new technologies for database development, such as aerial photography and satellite imagery techniques, to improve the cost effectiveness of local-level data collection and management. A.5 External Attributes The literature also asserts that factors outside of the urban transportation systems themselves, such as land use impacts and urban density thresholds, should be analyzed as secondary measures and predictors of long-term transit success. Such external indicators can be obtained from multiple databases and studies on urban location shifts and economic development patterns in cross- industrial and micro-geographic realms.

A-13 A.5.1 Urban Development Data Over the last decade, the demand for disaggregate data on urban development patterns has been growing in the United States, as the ability of transit-oriented development (TOD) to increase transit ridership, discourage urban sprawl, and promote economic development by densely locating a variety of property packages, business clusters, and residential communities around urban transit centers and along suburban transit corridors has been more importantly assessed as both the short- term predictor of success and the long-term measure of success by federal, state, and local decision makers. U.S. databases on urban development patterns are classified into: (i) residential location; (ii) business location; (iii) property transaction; (iv) multiple characteristic integration; and (v) urban simulation. By tradition, the U.S. Decennial Population and Housing Census has long been the chief public data source to analyze long-term residential location patterns around national, state, regional, and local transportation systems on different geographic scales (e.g., states, metropolitan statistical areas, counties, urbanized areas, ZIP codes, tracts, and block groups). In recent years, however, some supplemental secondary databases have become available to cover short-term changes in residential location patterns between the decennial census years. The GeoLytics Demographic Data, for example, currently deals with disaggregate estimates of population, housing, and household and labor location characteristics on micro-geographic scales (e.g., ZIP codes, census tracts, and block groups) from 2001 through 2008. The ESRI Updated Demographics database annually offers disaggregate estimates of more than 2,000 population, household, and original industrial and occupation variables in a variety of U.S. jurisdictions and geographies as custom-order commercial products. Additionally, the U.S. Census Bureau’s Longitudinal Employer-Household Dynamics (LEHD) database makes publicly available the residential location characteristics of workers by age, earning and standardized industrial categories at the census block-level from 2002 to 2008, excluding some state areas (New Hampshire, Massachusetts, Virgin Islands, and Puerto Rico). Similarly, disaggregate databases on short-term business location characteristics are now publicly available. The U.S. Economic Census Economy-Wide Key Statistics (EWKS) supplies the number of establishments and employees by the North American Industry Classification System (NAICS) code, value of sales, shipments, receipts, revenue, and annual payroll, at the ZIP code level in 1997, 2002, and 2007. On the basis of annual economic surveys, the U.S. Census Bureau’s ZIP Code Business Patterns (ZBP) series from 1994 through 2008 reports the number of establishments by employee size and total annual and quarter payroll figures. Notably, applying the long-term panel data in the ZBP series to examine cross-industrial composition changes is complicated by the fact that the classification system used to categorize establishments changed from the Standard Industrial Code (SIC) system to NAICS in 1997. While the LEHD database does keep the consistency of time-series data on the business location patterns of workers by age, earning, and industrial (NAICS) categories at the census block-level from 2002 to 2008, it excludes four state areas and a number of important business performance variables. Some private vendors offer time- series data on firm-level business inputs and outputs by industrial type, from which dynamic changes in business productivity around transit stations can be micro-geographically calculated as a measure of transit success in agglomeration economies. Walls & Associates, for instance, maintains the National Establishment Time-Series (NETS) Database, including over 36.5 million establishments with time-series information about their industries, locations, headquarters, ownerships, employment sizes, and annual sales over the period 1990 to 2009. Hedonic price analysis around stations is more a common approach to measuring the capitalization impacts of fixed-guideway transit projects on both business and residential activities.

A-14 Generally, net price increases are estimated as the accessibility/agglomeration benefits generated by transit investments, and they are expected to help recover the upfront capital costs of fixed- guideway transit systems. Property transaction records in the United States are provided by several private entities. One of the largest national databases is the First American CoreLogic’s RealQuest Professional. This online database system covers 97% of all U.S. real estate transactions, offers a geographic radius search tool to find target property records by proximity to transit stations and related facilities, and customizes up to 25,000 records in multiple downloadable data formats. DataQuick is another popular private database that contains more than 105 million assessor parcels in 2,300 jurisdictions, 85% of all properties in the top 100 MSAs, and 250 million historical recordings in 1,800 jurisdictions with geocodes, household level demographics, behavior, and lifestyle attributes. Also, Zillow, Inc. operates its newly established Zillow.com, an online real estate information system that provides very short-term data on home sales transactions for free. In the public sector, the Office of Federal Housing Enterprise Oversight (OFHEO) quarterly estimates the House Price Indexes (HPIs) for single-family detached properties using data on conventional conforming mortgage transactions obtained from the Federal Home Loan Mortgage Corporation (Freddie Mac) and the Federal National Mortgage Association (Fannie Mae). The Federal Housing Finance Agency’s (FHFA) website presents quarterly HPI data for long-term trends from 1975 to the present year at the census division, state, and MSA levels. Looking at short-term trends, the U.S. Department of Housing and Urban Development’s (HUD) Aggregated USPS Administrative Data On Address Vacancies website releases on a quarterly basis publicly available nationwide data on the total count and average number of days of vacant addresses, broken out by residential, business, and other property type, from 2005 to the present year at the census tract level. In general, these U.S. property transaction records are poorly integrated with other nationwide databases on fixed- guideway transit systems and urban development patterns, although dynamic trends in real estate markets importantly indicate the short-term predictors and long-term measures of success of fixed- guideway transit investments in transit-oriented developments. Given the above secondary data sources, some non-profit and academic research institutes have recently been developing readily integrated nationwide databases on multiple urban development characteristics around fixed-guideway transit stations. The Center for Transit-Oriented Development (CTOD) and Center for Neighborhood Technology (CNT) conduct the National TOD Database project, organizing over 40,000 observations of population, household, housing, employment, and travel variables at the transit zone (one-quarter- or one-half-mile buffer around existing and proposed stations in 47 metropolitan areas), the transit shed (the spatial aggregate of transit zones), and the transit region (aligns with the MSA boundary) levels based on the 2000 Decennial Population and Housing Census, the Census Transportation Planning Package (CTPP), and the LEHD data. The Natural Resource Ecology Laboratory at the Colorado State University delivers the National Dataset for Location Sustainability and Urban Form, whose 2009 census tract and block group data includes: residential-employment balance, density of people, housing, and jobs; diversity of land uses; accessibility to destinations; distance to transit stations (the 5Ds factors); and (originally defined) smart location indices (SLIs). Both of these integrated nationwide databases are useful for measuring the short-term predictors of transit success on the basis of recent location characteristics, but since they are new they do not have panel data for an extended period of time and cannot aid in the examination of long-term measures of success in urban development changes. Although U.S. decision makers have long faced a lack of longitudinal data on fixed-guideway transit investments and urban development impacts, several urban simulation models, such as

A-15 UrbanSim, PECAS, ILUTE, TRANUS, MEPLAN, and DRAM/EMPAL, are helping to shed light on predictors of long-term travel behavior and land use impact success. In their early stages, these large-scale simulation models had many deficiencies. They reflected too much real-world complexity, provided too coarse predictions to policymakers, required excessive data and money, imputed individual behaviors from aggregate data, and depended on unrealistic iterative processes (Lee 1973). The recent dynamic microsimulation models, however, have moved toward more realistic activity-based travel behaviors and more practical lot-level land uses based on understandable location theories, cost-efficient computations, and path-dependent interactions (Waddell 2011). Unfortunately, these technical improvements have made microsimulation models much more difficult for decision makers to use.

B-1 APPENDIX B: Data Collection and Construction of Variables This appendix provides a detailed description of the data and data sources used as measures and indicators or transit success at the project and metropolitan levels of analysis. We compiled data on fixed-guideway transit projects and metropolitan areas across the United States, including station- level, project-level, and regional-level information on ridership levels, agency operating costs, demographics, employment and population density, gross domestic product (GDP), gas prices, parking availability and pricing, regulatory restrictiveness in land uses, neighborhood walkability, rail and highway networks, and transit service characteristics. In some cases we measured these characteristics ourselves and in others we used secondary data sources. Below are the data utilized for our analysis (with data source, years collected, and geographic level): • Transit system ridership (National Transit Database [NTD], 1997-2009, urbanized area) Transit system operating costs (NTD, 1997-2009, urbanized area) • Transit system capital costs (Guerra and Cervero (2011), 2011, project) • Population, household, income and employment demographics (Census SF1 and SF3, 2000, metropolitan area/county/block group/block) • Population, household, income, and employment demographics (American Community Survey [ACS] 1-Year, 2005-2009, metropolitan area) • Employment demographics (Longitudinal Employer-Household Dynamics [LEHD], 2002- 2008, block) • Total unemployment (U.S. Bureau of Labor Statistics [BLS], 1990-2009, county) • Total jobs (U.S. Bureau of Economic Analysis [BEA], 1969-2009, metro area) • Total personal income (BEA, 1969-2009, metro area) • Total GDP (BEA, metro area) • Consumer expenditures (BLS Consumer Expenditure Survey [CES], 2005-2009, metro area) • Fuel cost (National Household Travel Survey [NHTS], 2009, county) • Retail gasoline price (GasBuddy.com, 2000-2011, county) • Average downtown parking price (Colliers International, 2009-2011, city) • Off-street private parking prices (Parking In Motion Inc., 2011, rooftop geocodes) • Highway congestion (Federal Highway Administration [FHWA], 1995-2009 and Texas Transportation Institute [TTI], 1995-2009, urbanized/urban area) • Land use regulatory restrictiveness ((Pendall et al. 2006), municipality) • Neighborhood walkability (walkscore.com, 2012, transit station) • Weather (National Climatic Data Center [NCDC], 2012, metro area) • Entropy indexes (various data from above; explained below) • Route-miles, number of stations, opening year, mode (various sources, 2011-2012, project) • Park-and-ride spaces, bus line connections (various sources, 2011-2012, project)

B-2 • Peak AM hour service frequency, average speed (various sources, 2011-2012, project) • Track grade (FTA Capital Cost Database and Google Earth, 2010, project) The multiple units of analysis employed for our study (metropolitan area, project, and station area) made it particularly challenging to collect and compile all of the necessary data. The different levels of data required that we take into account both the local information about a transit project and the regional information about the metropolitan area served by that project. We considered each potential success measure and predictor twice, at the project level and again at the metropolitan level. This challenge factored into both our data collection processes and our spatial data analyses. For example, we collected two sets of parking data from two different sources—the first on citywide parking prices and the second on localized rates around transit project stations. Additionally, we created catchment areas around each station for use at both the local project level of analysis and in aggregation up to the metropolitan level of analysis (Figure B-1.) Figure B-1: Example Catchment Areas Around Urban Rail and Commuter Rail Stations The extended time frame of our analysis also introduced complications into the process of data collection and compilation. For our panel dataset, we collected metropolitan-level information across seven years. The creation of annual catchment areas was particularly work-intensive, requiring the creation of new service area boundaries for each year that new stations opened. The resulting data set is possibly the most complete existing for urban rail transit stations and networks in the United States, covering 3,263 transit stations in 44 metropolitan areas across the country, with network links, consistent station and metropolis identifiers, system type and transfer dummies, and station opening years. To these data we spatially joined station-level, project-level, and metropolitan area-level information on demographics, employment, costs of driving and parking, transit service characteristics, and other variables. The majority of spatial data on transit lines and stations came from the NTAD, but we identified a number of gaps in that source’s LRT and HRT networks using a complete list of transit lines and stations provided by NTD. We filled in the missing spatial information using Google Earth, transit agency maps, the Center for Transit-Oriented Development (CTOD) station database, and the website urbanrail.net. B.1 Measures of Transit Project Success To measure transit project success, we collected ridership and cost data on the transit systems in our 18 metropolitan areas of study and on the 55 individual transit projects we analyze.

B-3 In a UCTC-funded project, Guerra and Cervero (2011) probed the relationship between job and population densities around rail stations and various cost-effectiveness measures. Data on ridership and capital costs come primarily from the information compiled by Guerra and Cervero, with some additional data collection performed for this project. For the transit system-level analysis, conducted for metropolitan areas, we collected total person miles traveled (PMT) from the National Transit Database (NTD), broken down by agency, mode, and operator from 1996 to 2009. We summed this information by metropolitan area and mode to create ridership statistics (measured in passenger miles traveled) for every fixed-guideway urban rail transit system across all relevant metropolitan areas in the United States. Since NTD data are broken down geographically by Urbanized Area (UZA), and not by census-defined metropolitan area, we mapped each UZA to its primary metro area to enable the metropolitan area-level analysis. U.S. Census Urban Areas served as the spatial link between UZA and metro area. Each UZA name was first matched to its affiliated urban area and then assigned to the primary metropolitan area in which the urban area falls. All transit agency operations are located within one metropolitan area, with the exception of NJ Transit, which operates in both the New York and Philadelphia metro areas. We chose to allocate NJ Transit’s ridership and cost information to the New York metropolitan area (New York-Northern New Jersey-Long Island, NY-NJ-PA) over the Philadelphia metropolitan area (Philadelphia-Camden-Wilmington, PA-NJ-DE-MD), as presumably larger shares of NJ Transit trips are oriented to the New York metropolitan area over the Philadelphia metropolitan area. Most project-level capital cost and ridership information data were from the Government Accountability Office (GAO), the Federal Transit Administration (FTA) (1994-2008), and individual transit agencies as part of previous work conducted by two members of our research team (Guerra and Cervero 2011). Building on their database of projects, we supplemented missing ridership information for over 10 transit lines, including newly built LRT and BRT systems, all CR systems, and any stations that were recently constructed as part of extensions or new lines. We collected average weekday boardings/alightings at the station level from individual agency websites and various other online sources, and we then assigned each station to the project of which it is a part. B.2 Predictors of Transit Project Success: Metropolitan Area To determine the factors that best predict a transit project’s success, we collected information on demographics and employment, system rail and highway network connectivity, costs of driving and transit use, regulatory restrictiveness in land uses, and transit service characteristics. B.2.1 Regional Demographics and Employment We compiled regional demographic information for the nationwide analysis of metropolitan areas from the U.S. Census 2000 and the 1-Year American Community Survey (2005-2009). We collected ACS data by metropolitan area and census data by county, which we then aggregated to the metropolitan area through either a summation or weighted average. Our selected data included population characteristics (race, median age), housing unit characteristics (occupancy, tenure, median rent and value), economic characteristics (median household income, per capita income, percentage of population below poverty line), and work force characteristics (workers per household, commute mode to work, vehicles per household). We also collected metropolitan-area economic data from the U.S. Bureau of Labor Statistics and the U.S. Bureau of Economic Analysis,

B-4 including job counts, unemployment figures, personal income levels, and GDP from 2000 through 2009. B.2.2 Catchment Area Demographics and Employment To determine the demographic characteristics around each transit project, we created “catchment areas” surrounding each project station and spatially applied to them Census 2000 and LEHD block-level data that fell within the designated area. For our analysis, we aggregated the station catchment area information to the metropolitan level and the project level (Figure B-2). Figure B-2: Catchment Area Creation Process The first step in creating catchment measures was to delineate catchment areas spatially. We assigned each station to its respective block/block group using the geographic areas defined by ESRI Census 2000 TIGER/Line Data. Around each station we created straight-line-distance buffers of 0.25, 0.5, and 1 mile for urban rail systems and 0.5, 1, and 3 miles for commuter rail systems. We intersected these buffers with Thiessen polygons constructed around each station to ensure that each station’s catchment area was mutually exclusive of the catchment areas around neighboring stations. We then clipped census blocks or block groups to each buffer to create shapefiles representing the portions of each block or block group within the catchment areas. We repeated this process for each year of station data available (2000-2009), since the opening of a new station in a given year sometimes changed the size and shape of the catchment areas.

B-5 Once the catchment areas were complete, we assigned Census 2000 block and block group data to each station catchment area, including residential demographics such as age, race, and commute mode/duration and household information such as size, occupancy, tenure, income, and automobile ownership, based on the land area share of each block falling within the buffer. We similarly incorporated LEHD block-level data on job counts by employment location from 2002-2008, broken down by industry and income group. To incorporate this information, we first calculated the fraction of land area of each block/block group falling within a given catchment area. Some demographic indicators were only available at the block group level. Rather than re-creating catchment areas using block group shapefiles, we aggregated the existing census block catchment area shapefiles up to the block group level. In these cases, the clipped area (within the catchment area) of each block within the block group was added up and then divided by the land area (as reported by the census) of the containing block group. Demographic information was then multiplied by this fraction in a similar fashion. We assigned census data to a catchment area based on that land area ratio. If an entire block/block group is within the bounds of a catchment area, the land area fraction would be equal to one and the full census count for a given demographic variable would be allocated to that catchment area. If only a portion of a block/block group falls within a catchment area, we applied the land area fraction and allocated only that percentage of the census count to the catchment area. We assigned a non-count census variable (e.g., median age) to a catchment area by taking weighted averages based on the catchment’s population size. Finally, we aggregated the characteristics of each station catchment area up to the regional level for our nationwide analysis of metropolitan areas through either a summation or a calculated average (in some cases weighted by population or households). B.2.3 Rail Network Measures Capturing the importance of the layout of the nodes and links of a transit network in determining its success, we created rail network connectivity measures by metropolitan area for all transit projects in our database. We conducted connectivity analyses in each of our metropolitan areas for every year between 2000 and 2010 in order to take into account annual changes in the system due to new line investments in a given year. The output measures are based on graph theory and spatial analysis tools. They quantify different network characteristics, allowing comparisons between networks and within networks over time. More details on the network index calculations are provided in Appendix D. We also utilized GIS to calculate additional network characteristics such as link and node density, network diameter, and nearest neighbor distance. B.2.4 Job Accessibility Measures There are two major approaches to analyzing job accessibility achieved through urban transportation networks: gravity-based and opportunity-based (isochrone). We applied a gravity- based measure to each transit station in 25 of our metropolitan areas, using LEHD employment location data and GIS-based network distance calculations between all stations in a transit project, and aggregated the results to the metropolitan area. Gravity-based job accessibility was measured as follows:

B-6 where i is station i, j is station j in the same metropolitan area, Ei is the number of employment in the catchment of station i (zones 1, 2, and 3), Ej is the number of employment in the catchment of station j (zones 1, 2, and 3), f(Dij) is an impedance function (linear, squared, exponential, powered 0.25, 0.50 and 0.75), zone 1 is 0.25 mi. for urban rail and 0.50 mi. for commuter rail, zone 2 is 0.50 mi. for urban rail and 1.00 mi. for commuter rail, and zone 3 is 1.00 mi. for urban rail and 3.00 mi. for commuter rail. To incorporate roadway-based job accessibility into our analyses we utilized existing data on opportunity-based job accessibility (Cervero and Murakami 2010). We geographically related to our transit station points the access values on nationwide 500-meter-grid-cells (i.e., total basic jobs accessible within 30 minutes through Interstate, freeway, and local arterial systems in 2003) and aggregated to the metropolitan area, weighted by population density. B.2.5 Auto Cost Measures To capture the costs of transportation alternatives in a region, in particular the time and monetary costs of the automobile, we compiled data on consumer transportation expenditures, gas prices, parking prices, and congestion. The U.S. Bureau of Labor Statistics provides broadly categorized information on consumer expenditures for selected Core-Based Statistical Areas (CBSAs) from 2005 to 2009, including a breakdown of transportation expenditure into net outlay for vehicle purchases, gasoline/motor oil, other vehicle expenses, and public transportation. We compiled measures of the market prices of gas and parking from various sources. First, we purchased average retail gasoline price data from GasBuddy.com for the years 2000-2011, which we aggregated up to the metropolitan area from the county level using a weighted average by Census 2000 population. We also calculated regional fuel cost using the 2009 National Household Travel Survey (NHTS), aggregated from the county to metropolitan level using a weighted average by population. From Colliers International we gathered data on average city parking prices between 2009 and 2011, and we aggregated those to the metropolitan area by separately averaging “primary” and “secondary” cities within the metropolis. In addition, we purchased parking pricing data from Parking In Motion, Inc. (PIM), which included rate information for almost 10,000 off-street parking lots across the United States. PIM collects these data through a telephone survey of parking facility operators and follow-up field work. The format of the parking lot prices as provided was extremely messy, and we needed to determine one overall parking rate by parsing the information from various time categories such as daily, hourly, early bird, every 30 minutes, first 30 minutes + additional hourly, and daily max, to name a few. Once we assigned a general 8-hour parking rate to each lot, we used the geographic coordinates that were identified for each lot to geographically relate them and their parking rate information to the closest transit station in our database (and the respective investment, where relevant). This allowed us to determine average parking prices within a given catchment area around each station. We attempted to collect traffic condition data at the corridor level, but the inconsistency of traffic database systems across U.S. cities led us to rely exclusively on two nationwide sources—the Federal Highway Administration’s (FHWA) Annual Highway Statistics and the Texas

B-7 Transportation Institute’s (TTI) Annual Urban Mobility Report. For both we reassigned congestion data from UZA to metropolitan area based upon population. Annual Highway Statistics (Table HM- 72) provided us with average daily vehicle miles-traveled (VMT) per freeway lane mile in 513 UZAs from 1995 through 2008, estimated as thousands of daily vehicle-miles traveled on freeways divided by total estimated freeway lane miles. The Annual Urban Mobility Report contains a travel time index, which measures road congestion by comparing travel conditions in the peak period to those in free-flow, in 85 selected UZAs from 1993 to 2009. The traffic conditions measured by FHWA are highly correlated with the travel time index calculated for TTI’s 85 selected UZAs. B.2.6 Other System-Level Predictors As a measure of station-area walkability, we assigned a Walk Score to each station in our database. Walk Score is a number between 0 and 100 that measures the walkability of any address, from “car-dependent” to “walker’s paradise.” More walkable neighborhoods are characterized by more amenities (e.g., parks and grocery stores) within walking distance, higher intersection densities and shorter average block lengths. Across the United States our stations cover the full spectrum of possible Walk Scores, from 0 to 100, with a mean Walk Score of 73. For our system- level analysis we calculated the mean Walk Score within each region. Regional Walk Scores range from 29 (Poughkeepsie-Newburgh-Middletown, NY) to 85 (Hagerstown-Martinsburg, MD-WV), and the average regional Walk Score is 66. To account for physical as well as built environment conditions, we collected from NCDC data on average temperature, sunlight, and precipitation within each region. We measured regulatory restrictiveness using a 2003 survey of jurisdictions conducted in 2003 by Rolf Pendall (see Pendall 2006). Pendall surveyed more than 1,800 localities, asking about various land use ordinances in place in each jurisdiction, with questions about planning, zoning, expansion potential, housing construction, public facilities, and affordable housing. We selected 12 survey questions and indicated with a dummy variable whether the given regulation was in place for each jurisdiction. To aggregate to the regional level, we calculated the percentage of the surveyed population within each metropolitan area to which the given regulation applied. B.3 Predictors of Transit Project Success: Project-Level The project-level analysis utilized many of the same predictors, including some metropolitan area-level variables such as regional population, household, economic, and work force characteristics. We applied the station catchment area spatial analysis described above, but we aggregated the catchment area demographic data by project instead of by metropolitan area. Within the catchment areas we also investigated the effect of average private off-street parking lot prices. We calculated marginal changes in transit network connectivity and complexity as well as gravity- based job accessibility measures for each station and aggregated the results to the project level. We recorded the opening year, the number of stations, and the total route-miles of every project, which we compiled from transit agency websites, urbanrail.net, FTA reports, descriptions and maps on agency websites, and maps provided by the National Transportation Atlas Database (NTAD). We approximated a number of project-level service characteristics, including speed and frequency, using individual transit agency websites, maps, and schedules. Finally, we used federal databases, Google Earth, and transit agency websites to augment data originally collected by Guerra and Cervero (2011) on a project’s service features, such as track grade, station park-and-ride spots, and bus connections from stations along line. Track grade for projects not included in their study was

B-8 found using the FTA Capital Cost Database (Booz Allen Hamilton, Inc. 2005) when possible. In other cases, it was estimated using the ruler tool in Google Earth. Grade transitions were generally counted as not-at-grade. Supplemental data on station park-and-ride spots and bus connections for projects not examined by Guerra and Cervero were estimated using transit agency information on station amenities and bus routes. B.4 Additional Variables Considered In addition to the variables listed above, we considered information on numerous other potential measures and predictors of transit project success. A complete summary of our data (collected, tested, and modeled) can be found in Appendix E, including what data we compiled, the geographic level and date range of the data, the sources from which the data came, what data entered into our analysis and its observed effects on ridership and PMT.

C-1 APPENDIX C: All Fixed-Guideway Transit Projects in the United States Note: This appendix is included in order to inform the reader about the fixed-guideway transit projects included in our modeling process. Projects were excluded when key data were unavailable, such as the LEHD data used to es mate employment near sta ons; data about parking cost in the CBD; and ridership, which we were some mes unable to procure by sta on from the relevant transit agency. We also excluded some projects in early stages of our data collec on process because capital cost data were not available. The first 55 projects in the table were those used in our ridership model. State City Project Name Mode Type Opening Year Route- miles Ridership Model Reason Excluded AZ Phoenix Metro Light Rail LRT Ini al 2008 20 YES CA Los Angeles Long Beach Blue Line LRT Ini al 1990 45 YES CA Los Angeles Green Line LRT Expansion 1995 20 YES CA Los Angeles Pasadena Gold Line LRT Expansion 2003 14 YES CA Los Angeles Red Line (Segment 1) HRT Expansion 1993 3 YES CA Los Angeles Red Line (Segment 2) HRT Expansion 2000 7 YES CA Los Angeles Red Line (Segment 3) HRT Expansion 1996/1999 7 YES CA Los Angeles Orange Line BRT Expansion 2005 14 YES CA Sacramento Sacramento Stage I LRT Ini al 1987 18 YES CA Sacramento Mather Field Road Extension LRT Extension 1998 6 YES CA Sacramento South Phase 1 LRT Expansion 2003 6 YES CA Sacramento Sacramento Folsom Corridor LRT Extension 2005 11 YES CA San Diego Blue Line LRT Ini al 1981 25 YES CA San Diego Orange Line LRT Expansion 1986 22 YES CA San Diego Mission Valley East LRT Extension 2005 6 YES CA San Francisco Ini al BART HRT Ini al 1972 72 YES CA San Francisco BART SFO Extension HRT Extension 2003 9 YES CA San Jose San Jose North Corridor LRT Ini al 1987 17 YES CA San Jose Tasman West LRT Expansion 1999 8 YES CA San Jose Tasman East LRT Expansion 2001 5 YES CA San Jose VTA Capitol Segment LRT Extension 2004 3 YES CA San Jose VTA Vasona Segment LRT Expansion 2005 5 YES CO Denver Central Corridor LRT Ini al 1994 5 YES CO Denver Denver Southwest Corridor LRT Extension 2000 9 YES CO Denver Denver Southeast (T- LRT Expansion 2006 19 YES

C-2 State City Project Name Mode Type Opening Year Route- miles Ridership Model Reason Excluded REX) FL Miami Metrorail HR Inial 1984 21 YES FL Miami South Florida Tri-Rail Upgrades CR Enhancement 2007 72 YES GA Atlanta North / South Line HRT Expansion 1981 22 YES GA Atlanta North Line Dunwoody Extension HRT Extension 1996 2 YES IL Chicago O'Hare Extension (Blue Line) HRT Extension 1984 8 YES IL Chicago Orange Line HRT Expansion 1993 9 YES IL Chicago Douglas Branch HRT Extension 2005 7 YES IL Chicago Metra North Central CR Expansion 1996 55 YES IL Chicago Metra Southwest Corridor CR Extension 2006 11 YES MD Balmore Central Line LRT Expansion 1992 23 YES MD Balmore Three extensions LRT Extension 1997 7 YES MD Balmore Balmore Metro HRT Inial 1983 12 YES MN Minneapolis Hiawatha Corridor LRT Inial 2004 12 YES NJ Jersey City Hudson-Bergen MOS 1 and 2 LRT Expansion 2003 15 YES NJ Newark Newark Elizabeth MOS-1 LRT Expansion 2006 1 YES NJ Trenton Southern NJ Light Rail Transit System LRT Expansion 2004 28 YES NY Buffalo Buffalo Metro Rail LRT Inial 1985 6 YES OH Cleveland Cleveland Healthline BRT Expansion 2008 7 YES OR Eugene Eugene EmX BRT Inial 2007 4 YES OR Portland Portland MAX Segment I LRT Inial 1986 15 YES OR Portland Portland Westside/Hillsboro MAX LRT Extension 1998 18 YES OR Portland Portland Airport MAX LRT Expansion 2001 6 YES OR Portland Portland Interstate MAX LRT LRT Expansion 2004 6 YES PA Philadelphia SEPTA Frankford Rehabilitaon HRT Enhancement 2003 5 YES TX Dallas S&W Oak Cliff and Park Lane LRT Extension 1996 20 YES TX Dallas North Central LRT Extension 2002 13 YES UT Salt Lake City North-South Corridor LRT Inial 1999 15 YES UT Salt Lake City Medical Center Ext. LRT Extension 2003 2 YES UT Salt Lake City University Ext. LRT Extension 2003 3 YES WA Sea¡le Sea¡le Central Link LRT Inial 2009 14 YES

C-3 State City Project Name Mode Type Opening Year Route- miles Ridership Model Reason Excluded Light Rail Project CA Los Angeles MetroLink CR Ini al 1992 NO CA Los Angeles MetroLink Riverside Orange County Lines CR Expansion 1994 NO Ridership CA Los Angeles MetroLink Inland Empire Orange County Line CR Expansion 1995 NO Ridership CA Los Angeles MetroLink 91 Line CR Expansion 2002 NO Ridership CA San Diego Sprinter LRT Expansion 2008 NO Ridership CA San Diego Coaster CR Expansion 1995 NO Ridership CA San Francisco Muni J-Church Extension LRT Extension 1991 NO CA San Francisco Muni T-Third Extension LRT Expansion 2007 NO CA San Francisco BART Colma Extension HRT Extension 1996 NO CA San Francisco BART PiŒsburgh Bay Point Extension HRT Extension 1996 NO CA San Francisco BART Dublin Pleasanton Extension HRT Expansion 1997 NO CA San Jose Altamont Commuter Express CR Expansion 1998 NO Ridership CO Denver Central PlaŒe Valley LRT Expansion 2002 NO CT New Haven Shoreline East CR Expansion 1990 NO Ridership DC Washington DC Addison (G) Blue Line HRT Expansion 1977 4 NO LEHD DC Washington DC Glenmont (B) red HRT Extension 1978 12 NO LEHD DC Washington DC New Carrollton (D) Orange HRT Expansion 1978 12 NO LEHD DC Washington DC Yellow Line HRT Expansion 1983 14 NO LEHD DC Washington DC Shady Grove (A) red HRT Extension 1984 18 NO LEHD DC Washington DC Vienna (K) Orange HRT Extension 1986 12 NO LEHD DC Washington DC Franconia/Springfield (J/H) Blue Line HRT Extension 1997 4 NO LEHD DC Washington DC Anacos a Outer (F) HRT Extension 2001 7 NO LEHD DC Washington DC U street (E) green HRT Expansion 2001 2 NO LEHD DC Washington DC Largo Metrorail Extension HRT Extension 2004 3 NO LEHD DC Washington DC Virginia Railway Express CR Expansion 1992 NO LEHD FL Miami Tri-Rail CR Expansion 1989 NO LEHD GA Atlanta East-West Line HRT Ini al 1979 NO LEHD

C-4 State City Project Name Mode Type Opening Year Route- miles Ridership Model Reason Excluded GA Atlanta Proctor Creek Branch HRT Expansion 1992 NO LEHD IL Chicago Metra UP West Corridor CR Rehab 2009 9 NO Ridership IL Chicago Green Line Rehabilita€on HRT Rehab 1996 NO MA Boston Southwest Corridor HRT Expansion 1987 5 NO LEHD MA Boston MBTA Worcester Line CR Extension 1994 NO LEHD MA Boston MBTA Old Colony Lines CR Expansion 1997 NO LEHD MA Boston MBTA Greenbush Line CR Expansion 2007 NO LEHD MA Boston South Boston Piers - Phase 1 BRT Expansion 2004 NO LEHD MD Bal€more Owings Mills Extension HRT Extension 1987 NO MD Bal€more Johns Hopkins Hospital Extension HRT Extension 1995 NO MN Minneapolis Northstar Line CR Expansion 2009 NO Timeframe MO St. Louis MetroLink LRT Ini€al 1993 NO MO St. Louis St. Louis St. Clair County Extension LRT Extension 2001 NO Parking rate MO St. Louis Cross County Extension LRT Extension 2006 NO NC Charlo”e Charlo”e South Corridor LRT Ini€al 2007 10 NO Ridership NJ Newark Hudson-Bergen MOS 1 and 2 LRT Expansion 2000 NO Parking NJ Newark Midtown Direct CR Upgrade 1996 NO NJ Newark Montclair Connec€on CR Upgrade 2002 NO NM Albuquerque New Mexico Rail Runner Express CR Ini€al 2006 NO NY New York Archer Avenue Line HRT Extension 1988 NO OR Portland Portland Streetcar LRT Expansion 2001 NO Ridership OR Portland Green Line LRT Expansion 2009 NO Timeframe OR Portland WES CR Expansion 2009 NO Timeframe PA Philadelphia Center City Commuter Connec€on CR Upgrade 1984 NO PA Philadelphia SEPTA Airport Line CR Expansion 1985 NO PA Pi”sburgh Light Rail Stage I LRT Expansion 1984 16 NO Ridership PA Pi”sburgh Light Rail Stage II LRT Expansion 2004 5 NO Ridership PA Pi”sburgh South Busway BRT Ini€al 1977 NO PA Pi”sburgh East Busway BRT Expansion 1983 NO PA Pi”sburgh West Busway BRT Expansion 2000 NO TN Memphis Memphis Medical Center LRT Expansion 2004 NO

C-5 State City Project Name Mode Type Opening Year Route- miles Ridership Model Reason Excluded TN Nashville Music City Star CR Ini al 2006 NO TX Aus n Capital MetroRail CR Ini al 2010 NO Timeframe TX Dallas Northeast Extension LRT Extension 2002 NO TX Dallas Green Line LRT Expansion 2009 NO Timeframe TX Dallas Trinity Railway Express CR Expansion 1996 NO TX Dallas A-Train CR Expansion 2011 NO Timeframe TX Houston Houston METRO LRT Ini al 2004 NO UT Salt Lake City Intermodal Hub Extension LRT Extension 2008 NO UT Salt Lake City FrontRunner CR Expansion 2008 NO WA SeaŒle South Lake Union Streetcar LRT Expansion 2007 NO WA SeaŒle Sounder Commuter Rail CR Ini al 2000 NO WA Tacoma Tacoma Link LRT Ini al 2003 NO

D-1 APPENDIX D: Network Measures We calculated and tested measures of network connectivity by metropolitan area for both railway and highway networks, as described below. As noted in the report, we found that these measures tended not to be statistically significant with the inclusion of simpler indicators. • Number of Nodes (v) and Links (e), and Total Length of a Graph (L(G)). Links are segments of track or roadway. Nodes are locations where segments meet (e.g., intersections). The term “graph” can be understood to mean “network.” Total length is simply the summer linear length of links. • Diameter (DM or D(d)) is the length of a straight path between the two nodes of a network that are farthest away from each other. (Similarly, the theoretical diameter length can be computed from the actual area of the region [= Pi *(DM/2)2].) • Number of Cycles (u), or the maximum number of “independent cycles” in a graph, is estimated by the number of nodes, links (and sub-graphs, not explained here, which are usually equal to 1) in each metropolitan area. The more complex a network is, the higher the value of u, so the measure can be used as an indicator of the level of development and complexity in a transport system. u = e – v + p Based on the elements above, we computed the following network connectivity indices: • Alpha Index measures connectivity by comparing the number of cycles in a graph with the maximum possible number of cycles. The higher the alpha value, the more connected the network. Trees and simple networks have an alpha value of 0, whereas completely connected networks have an alpha value of 1. The alpha index measures the level of connectivity independent of the number of nodes in the network. µ 2v - 5 • Beta Index measures connectivity by evaluating the relationship between the number of links and the number of nodes. Trees and simple networks have a beta value of less than 1, connected networks with one cycle have a beta value of 1, and more complex networks have a beta value greater than 1. Complex networks have a high value of beta, as more links equates to more possible paths in the network (assuming fixed number of nodes). =

D-2 = e v • Gamma Index measures connectivity by evaluating the relationship between the number of observed links and the number of possible links. Values of gamma fall between 0 and 1, with a gamma value of 1 indicating a completely connected network. In reality, a gamma value equal to 1 is extremely unlikely. The gamma index is used to efficiently measure the progression of a network over time. = e 3(v – 2) • Eta Index measures the average length per link. Adding a new node to a network while maintaining the overall length of a graph will cause a decrease in the eta value. = µ 2v - 5 • Diameter Ratio measures the relationship between the diameter of a graph L(G) and the theoretical diameter of a metropolitan area (Area). A high Diameter Index value reflects a relatively large-size network to the area (closer to or even more than 1); on the other hand, a low value (closer to 0) represent a relatively small-size network to the area. DI = D(d) 0.5×( Area )0.5 • Pi Index measures the ratio of the total length of a graph L(G) and the distance along its diameter. The index is labeled Pi because it closely resembles the actual Pi value (3.1415), which expresses the ratio between the circumference and the diameter of a circle. A high pi value reflects a well-developed network, whereas a low pi value (closer to 1) represents a linear corridor. = L(G) D(d) We also generated another measure expressing the situation of a node (station) in each regional space (MSA): • The Average Nearest Neighbors Distance (NND) Index measures the ratio of the Observed Mean Distance to the Expected Mean Distance. The expected distance is the

D-3 average distance between neighbors in a hypothetical random distribution (our calculations are based on Euclidean distance). If the index is less than 1, the pattern exhibits clustering; if the index is greater than 1, the pattern is dispersed. This ratio can be automatically generated using ArcGIS. Source: ArcGIS 10

E-1 APPENDIX E: Variables List Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Transit Service Characteriscs Transit Service Characteriscs Direconal length of new service Project various sources Direconal length of new service X Percent of track elevated Project 2010 FTA Capital Cost Database; Google Earth Percent of track elevated X Percent of track at grade Project 2010 FTA Capital Cost Database; Google Earth Percent of track at grade X -0.08 Percent of track below ground Project 2010 FTA Capital Cost Database; Google Earth Percent of track below ground X Percent of track in highway median Project 2010 FTA Capital Cost Database; Google Earth Percent of track in highway median X Presence of parking at sta€ons (dummy) Project 2011-2012 Transit agency websites Presence of parking at sta€ons (dummy) X Number of park-and- ride spaces Project 2011-2012 Transit agency websites Number of park-and- ride spaces X 0.37 Frequency of service in morning peak hour Project 2011-2012 Transit agency websites Frequency of service in morning peak hour X Average speed of service Project 2011-2012 Transit agency websites Average speed of service X

E-2 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Number of bus lines that intersect project line Project 2011-2012 various sources Number of bus lines that intersect project line X New line (dummy) Project various sources New line (dummy) X Expansion of exisng line (dummy) Project various sources Expansion of exisng line (dummy) X Enhancement of exisng line (dummy) Project various sources Enhancement of exisng line (dummy) X Extension to exisng line (dummy) Project various sources Extension to exisng line (dummy) X Capital cost (millions, 2009 dollars) Project Guerra and Cervero, 2011 Capital cost (millions, 2009 dollars) X Received FTA New Starts funding (dummy) Project 2010 FTA Received FTA New Starts funding (dummy) X Dollars spent per mile (millions, 2009 dollars) Project Guerra and Cervero, 2011 Dollars spent per mile (millions, 2009 dollars) X Opening year Project various sources Opening year X Age Project various sources Age X 0.14 0.21 Number of staons Project 2011-2012 various sources Number of staons X Number of terminals Project 2011-2012 various sources Number of terminals X Number of airports served by project line Project 2011-2012 various sources Number of airports served by project line X Mode heavy rail transit (dummy) Project various sources Mode heavy rail transit (dummy) X Mode bus rapid transit (dummy) Project various sources Mode bus rapid transit (dummy) X Mode light rail (dummy) Project various sources Mode light rail (dummy) X Transit Network Characteriscs Transit Network Characteriscs

E-3 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Number of jobs within 0.5 mi of all fixed- guideway staons in metropolitan area Region Catchments LEHD Number of jobs within 0.5 mi of all fixed- guideway staons in metropolitan area X X Populaon within 0.5 mi of all fixed-guideway staons in metropolitan area Region Catchments Census 2000 Populaon within 0.5 mi of all fixed-guideway staons in metropolitan area X X Direconal route-miles of the transit network Region 2002-2008 NTD Direconal route-miles of the transit network X X Direconal route-miles of heavy rail transit Region 2002-2008 NTD Direconal route-miles of heavy rail transit X X Direconal route-miles of light rail Region 2002-2008 NTD Direconal route-miles of light rail X X Total passenger miles traveled (thousands) Region 2002-2008 NTD Total passenger miles traveled (thousands) X X Network connecvity index a Region Links 2008 GIS Calculaons Network connecvity index a X Network connecvity index b Region Links 2008 GIS Calculaons Network connecvity index b X Network connecvity index c Region Links 2008 GIS Calculaons Network connecvity index c X Network connecvity index g Region Links 2008 GIS Calculaons Network connecvity index g X Network connecvity index e Region Links 2008 GIS Calculaons Network connecvity index e X Network connecvity index s Region Links 2008 GIS Calculaons Network connecvity index s X Network connecvity index p Region Links 2008 GIS Calculaons Network connecvity index p X Length of all rail links (meters) Region Links 2008 GIS Calculaons Length of all rail links (meters) X Number of links in the network Region Links 2008 GIS Calculaons Number of links in the network X Number of nodes Region Links 2008 GIS Calculaons Number of nodes X Link density (km per square km) Region Links 2008 GIS Calculaons Link density (km per square km) X

E-4 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Node density (number per square km) Region Links 2008 GIS Calculaons Node density (number per square km) X Diameter of rail network Region Links 2008 GIS Calculaons Diameter of rail network X Nearest neighborhood distance Region Links 2008 GIS Calculaons Nearest neighborhood distance X Percent of PMT traveled on bus Region Links 2002-2008 NTD Percent of PMT traveled on bus X Staon-Area Characteriscs Staon-Area Characteriscs Land area within 0.5 mi of project sta ons Project Catchments 2008 Census Land area within 0.5 mi of project sta ons X Land area within 0.5 mi of fixed-guideway sta ons in metropolitan area Region Catchments 2008 Census Land area within 0.5 mi of fixed-guideway sta ons in metropolitan area X X Popula on within 0.5 mi of project sta ons Project Catchments 2000 Census Popula on within 0.5 mi of project sta ons X 0.05 0.04 Popula on within 0.5 mi of fixed-guideway sta ons in metropolitan area Region Catchments 2000 Census Popula on within 0.5 mi of fixed-guideway sta ons in metropolitan area X -0.87 X White popula on within 0.5 mi of project sta ons Project Catchments 2000 Census White popula on within 0.5 mi of project sta ons X White popula on within 0.5 mi of fixed- guideway sta ons in metropolitan area Region Catchments 2000 Census White popula on within 0.5 mi of fixed- guideway sta ons in metropolitan area X X La no popula on within 0.5 mi of project sta ons Project Catchments 2000 Census La no popula on within 0.5 mi of project sta ons X La no popula on within 0.5 mi of fixed- guideway sta ons in metropolitan area Region Catchments 2000 Census La no popula on within 0.5 mi of fixed- guideway sta ons in metropolitan area X X Median age within 0.5 mi of project sta ons Project Catchments 2000 Census Median age within 0.5 mi of project sta ons X

E-5 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Median age within 0.5 mi of fixed-guideway staons in metropolitan area Region Catchments 2000 Census Median age within 0.5 mi of fixed-guideway staons in metropolitan area X Number of residents under 18 within 0.5 mi of project staons Project Catchments 2000 Census Number of residents under 18 within 0.5 mi of project staons X Number of residents under 18 within 0.5 mi of fixed-guideway staons in metropolitan area Region Catchments 2000 Census Number of residents under 18 within 0.5 mi of fixed-guideway staons in metropolitan area X X Number of residents over 65 within 0.5 mi of project staons Project Catchments 2000 Census Number of residents over 65 within 0.5 mi of project staons X Number of residents over 65 within 0.5 mi of fixed-guideway staons in metropolitan area Region Catchments 2000 Census Number of residents over 65 within 0.5 mi of fixed-guideway staons in metropolitan area X X Number of housing units within 0.5 mi of project staons Project Catchments 2000 Census Number of housing units within 0.5 mi of project staons X Number of housing units within 0.5 mi of fixed-guideway staons in metropolitan area Region Catchments 2000 Census Number of housing units within 0.5 mi of fixed-guideway staons in metropolitan area X X Number of occupied housing units within 0.5 mi of project staons Project Catchments 2000 Census Number of occupied housing units within 0.5 mi of project staons X Number of occupied housing units within 0.5 mi of fixed-guideway staons in metropolitan area Region Catchments 2000 Census Number of occupied housing units within 0.5 mi of fixed-guideway staons in metropolitan area X X Number of vacant housing units within 0.5 mi of project staons Project Catchments 2000 Census Number of vacant housing units within 0.5 mi of project staons X

E-6 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Number of vacant housing units within 0.5 mi of fixed-guideway staons in metropolitan area Region Catchments 2000 Census Number of vacant housing units within 0.5 mi of fixed-guideway staons in metropolitan area X X Number of owner- occupied housing units within 0.5 mi of project staons Project Catchments 2000 Census Number of owner- occupied housing units within 0.5 mi of project staons X Number of owner- occupied housing units within 0.5 mi of fixed- guideway staons in metropolitan area Region Catchments 2000 Census Number of owner- occupied housing units within 0.5 mi of fixed- guideway staons in metropolitan area X X Number of renter- occupied housing units within 0.5 mi of project staons Project Catchments 2000 Census Number of renter- occupied housing units within 0.5 mi of project staons X Number of renter- occupied housing units within 0.5 mi of fixed- guideway staons in metropolitan area Region Catchments 2000 Census Number of renter- occupied housing units within 0.5 mi of fixed- guideway staons in metropolitan area X X Number of one-person households within 0.5 mi of project staons Project Catchments 2000 Census Number of one-person households within 0.5 mi of project staons X Number of four-person households within 0.5 mi of project staons Project Catchments 2000 Census Number of four-person households within 0.5 mi of project staons X Number of seven- person households within 0.5 mi of project staons Project Catchments 2000 Census Number of seven- person households within 0.5 mi of project staons X Number of households with no vehicles within 0.25 mi of project staons Project Catchments 2000 Census Number of households with no vehicles within 0.25 mi of project staons X

E-7 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Number of households with one vehicle within 0.25 mi of project staons Project Catchments 2000 Census Number of households with one vehicle within 0.25 mi of project staons X Number of households with two vehicles within 0.25 mi of project staons Project Catchments 2000 Census Number of households with two vehicles within 0.25 mi of project staons X Number of households with three vehicles within 0.25 mi of project staons Project Catchments 2000 Census Number of households with three vehicles within 0.25 mi of project staons X Number of households with four vehicles within 0.25 mi of project staons Project Catchments 2000 Census Number of households with four vehicles within 0.25 mi of project staons X Number of households with five+ vehicles within 0.25 mi of project staons Project Catchments 2000 Census Number of households with five+ vehicles within 0.25 mi of project staons X Number of jobs within 0.5 mi of project staons Project Catchments 2002-2008 LEHD Number of jobs within 0.5 mi of project staons X 0.18 0.21 Number of jobs within 0.5 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Number of jobs within 0.5 mi of fixed- guideway staons in metropolitan area X -0.67 X Percent of metropolitan-area jobs that fall within 0.5 mi of fixed-guideway staons Region Catchments 2002-2008 LEHD Percent of metropolitan-area jobs that fall within 0.5 mi of fixed-guideway staons X Number of workers under 30 within 0.5 mi of project staons Project Catchments 2008 LEHD Number of workers under 30 within 0.5 mi of project staons X

E-8 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Number of workers under 30 within 0.5 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Number of workers under 30 within 0.5 mi of fixed-guideway staons in metropolitan area X X Number of workers aged 30-54 within 0.5 mi of project staons Project Catchments 2008 LEHD Number of workers aged 30-54 within 0.5 mi of project staons X Number of workers over 54 within 0.5 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Number of workers over 54 within 0.5 mi of fixed-guideway sta ons in metropolitan area X Number of jobs paying less than $1250 per month within 0.25 mi of project sta ons Project Catchments 2008 LEHD Number of jobs paying less than $1250 per month within 0.25 mi of project sta ons X Number of jobs paying less than $1250 per month within 0.25 mi of fixed-guideway sta ons in metropolitan area Region Catchments 2002-2008 LEHD Number of jobs paying less than $1250 per month within 0.25 mi of fixed-guideway sta ons in metropolitan area X X Number of jobs paying more than $1250 and less than $3333 per month within 0.25 mi of project sta ons Project Catchments 2008 LEHD Number of jobs paying more than $1250 and less than $3333 per month within 0.25 mi of project sta ons X Number of jobs earning more than $3333 per month (higher wage jobs) within 0.5 mi of fixed-guideway sta ons in metropolitan area Region Catchments 2002-2008 LEHD Number of jobs earning more than $3333 per month (higher wage jobs) within 0.5 mi of fixed-guideway sta ons in metropolitan area X 0.44 Agriculture, Forestry, Fishing and Hun ng jobs within 0.25 mi of project sta ons Project Catchments 2008 LEHD Agriculture, Forestry, Fishing and Hun ng jobs within 0.25 mi of project sta ons X

E-9 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Agriculture, Forestry, Fishing and Hunng jobs within 0.25 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Agriculture, Forestry, Fishing and Hunng jobs within 0.25 mi of fixed-guideway staons in metropolitan area X X Mining, Quarrying and Oil and Gas Extracon jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Mining, Quarrying and Oil and Gas Extracon jobs within 0.25 mi of project staons X Mining, Quarrying and Oil and Gas Extracon jobs within 0.25 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Mining, Quarrying and Oil and Gas Extracon jobs within 0.25 mi of fixed-guideway staons in metropolitan area X X Ulies jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Ulies jobs within 0.25 mi of project staons X Ulies jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Ulies jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Construcon jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Construcon jobs within 0.25 mi of project staons X Construcon jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Construcon jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Manufacturing jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Manufacturing jobs within 0.25 mi of project staons X Manufacturing jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Manufacturing jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Wholesale Trade jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Wholesale Trade jobs within 0.25 mi of project staons X

E-10 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Wholesale Trade jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Wholesale Trade jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Retail Trade jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Retail Trade jobs within 0.25 mi of project staons X Retail Trade jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Retail Trade jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Transportaon and Warehousing jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Transportaon and Warehousing jobs within 0.25 mi of project staons X Transportaon and Warehousing jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Transportaon and Warehousing jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Informaon jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Informaon jobs within 0.25 mi of project staons X Informaon jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Informaon jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Finance and Insurance jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Finance and Insurance jobs within 0.25 mi of project staons X Finance and Insurance jobs within 0.25 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Finance and Insurance jobs within 0.25 mi of fixed-guideway staons in metropolitan area X X Real Estate and Rental and Leasing jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Real Estate and Rental and Leasing jobs within 0.25 mi of project staons X

E-11 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Real Estate and Rental and Leasing jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Real Estate and Rental and Leasing jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Professional, Scienfic and Technical Services jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Professional, Scienfic and Technical Services jobs within 0.25 mi of project staons X Professional, Scienfic and Technical Services jobs within 0.25 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Professional, Scienfic and Technical Services jobs within 0.25 mi of fixed-guideway staons in metropolitan area X X Management of Companies and Enterprises jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Management of Companies and Enterprises jobs within 0.25 mi of project staons X Management of Companies and Enterprises jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Management of Companies and Enterprises jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Administrave and Support and Waste Management and Remediaon Services jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Administrave and Support and Waste Management and Remediaon Services jobs within 0.25 mi of project staons X Administrave and Support and Waste Management and Remediaon Services jobs within 0.25 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Administrave and Support and Waste Management and Remediaon Services jobs within 0.25 mi of fixed-guideway staons in metropolitan area X X

E-12 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Educaonal Services jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Educaonal Services jobs within 0.25 mi of project staons X Educaonal Services jobs within 0.25 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Educaonal Services jobs within 0.25 mi of fixed-guideway staons in metropolitan area X X Health Care and Social Assistance jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Health Care and Social Assistance jobs within 0.25 mi of project staons X Health Care and Social Assistance jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Health Care and Social Assistance jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Arts, Entertainment and Recreaon jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Arts, Entertainment and Recreaon jobs within 0.25 mi of project staons X Arts, Entertainment and Recreaon jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Arts, Entertainment and Recreaon jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Accommodaon and Food Services jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Accommodaon and Food Services jobs within 0.25 mi of project staons X Accommodaon and Food Services jobs within 0.25 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Accommodaon and Food Services jobs within 0.25 mi of fixed- guideway staons in metropolitan area X X Other Services (except Public Administraon) jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Other Services (except Public Administraon) jobs within 0.25 mi of project staons X

E-13 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Other Services (except Public Administraon) jobs within 0.25 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Other Services (except Public Administraon) jobs within 0.25 mi of fixed-guideway staons in metropolitan area X X Public Administraon jobs within 0.25 mi of project staons Project Catchments 2008 LEHD Public Administraon jobs within 0.25 mi of project staons X Public Administraon jobs within 0.25 mi of fixed-guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Public Administraon jobs within 0.25 mi of fixed-guideway staons in metropolitan area X X Retail, entertainment and food jobs (a„racon-based) within 0.5 mi of fixed- guideway staons in metropolitan area Region Catchments 2002-2008 LEHD Retail, entertainment and food jobs (a„racon-based) within 0.5 mi of fixed- guideway staons in metropolitan area X 0.68 Gravity-based job accessibility measure: linear, 0.25 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: linear, 0.25 mi catchment X Gravity-based job accessibility measure: linear, 0.5 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: linear, 0.5 mi catchment X Gravity-based job accessibility measure: linear, 1 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: linear, 1 mi catchment X Gravity-based job accessibility measure: exponenal, 0.25 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: exponenal, 0.25 mi catchment X

E-14 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Gravity-based job accessibility measure: exponenal, 0.5 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: exponenal, 0.5 mi catchment X Gravity-based job accessibility measure: exponenal, 1 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: exponenal, 1 mi catchment X Gravity-based job accessibility measure: 0.25 power law, 0.25 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 0.25 power law, 0.25 mi catchment X Gravity-based job accessibility measure: 0.25 power law, 0.5 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 0.25 power law, 0.5 mi catchment X Gravity-based job accessibility measure: 0.25 power law, 1 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 0.25 power law, 1 mi catchment X Gravity-based job accessibility measure: 0.5 power law, 0.25 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 0.5 power law, 0.25 mi catchment X Gravity-based job accessibility measure: 0.5 power law, 0.5 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 0.5 power law, 0.5 mi catchment X Gravity-based job accessibility measure: 0.5 power law, 1 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 0.5 power law, 1 mi catchment X Gravity-based job accessibility measure: 0.75 power law, 0.25 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 0.75 power law, 0.25 mi catchment X Gravity-based job accessibility measure: 0.75 power law, 0.5 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 0.75 power law, 0.5 mi catchment X

E-15 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Gravity-based job accessibility measure: 0.75 power law, 1 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 0.75 power law, 1 mi catchment X Gravity-based job accessibility measure: 2 power law, 0.25 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 2 power law, 0.25 mi catchment X Gravity-based job accessibility measure: 2 power law, 0.5 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 2 power law, 0.5 mi catchment X Gravity-based job accessibility measure: 2 power law, 1 mi catchment Region Catchments 2008 GIS and LEHD Gravity-based job accessibility measure: 2 power law, 1 mi catchment X Interacƒon of jobs and populaƒon within 0.5 mi of all project staƒons with daily parking rates in the CBD Project Catchments Census, LEHD and Parking In Moƒon Interacƒon of jobs and populaƒon within 0.5 mi of all project staƒons with daily parking rates in the CBD X 0.39 0.659 Interacƒon of jobs and populaƒon within 0.5 mi of all fixed-guideway staƒons in metropolitan area with average daily traffic per fwy lane mile Region Catchments Census, LEHD and FHWA Interacƒon of jobs and populaƒon within 0.5 mi of all fixed-guideway staƒons in metropolitan area with average daily traffic per fwy lane mile X 0.57 Metropolitan Area Characteriscs Metropolitan Area Characteriscs Populaƒon of metropolitan area Region 2002-2008 BEA Populaƒon of metropolitan area X 0.12 X Total income of metropolitan area Region 2002-2008 BEA Total income of metropolitan area X

E-16 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Per capita income of metropolitan area Region 2002-2008 BEA Per capita income of metropolitan area X Populaon growth rate of metropolitan area Region 2002-2008 BEA Populaon growth rate of metropolitan area X Real GDP of metropolitan area Region 2002-2008 BEA Real GDP of metropolitan area X Per capita real GDP of metropolitan area Region 2002-2008 BEA Per capita real GDP of metropolitan area X Percent white residents in metropolitan area Region 2000 Census Percent white residents in metropolitan area X Percent Hispanic residents in metropolitan area Region 2000 Census Percent Hispanic residents in metropolitan area X Percent residents under 18 in metropolitan area Region 2000 Census Percent residents under 18 in metropolitan area X Percent residents over 65 in metropolitan area Region 2000 Census Percent residents over 65 in metropolitan area X Percent residents enrolled at undergraduate university in metropolitan area Region 2000 Census Percent residents enrolled at undergraduate university in metropolitan area X Percent residents that immigrated since 2000 in metropolitan area Region 2000 Census Percent residents that immigrated since 2000 in metropolitan area X Number of commuters who commute by motorcycle in metropolitan area Region 2000 Census Number of commuters who commute by motorcycle in metropolitan area X Number of jobs in metropolitan area Region 2002-2008 BEA Number of jobs in metropolitan area X X Populaon-weighted average congeson of metropolitan area Region 2002-2008 TTI Populaon-weighted average congeson of metropolitan area X X Average daily VMT per freeway lane mile in metropolitan area Region 2002-2008 FHWA Average daily VMT per freeway lane mile in metropolitan area X -0.03

E-17 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Share of jurisdicons covered with a comprehensive plan in metropolitan area Region 2003 Pendall et al. Share of jurisdicons covered with a comprehensive plan in metropolitan area X X Share of jurisdicons subject to zoning ordinances in metropolitan area Region 2003 Pendall et al. Share of jurisdicons subject to zoning ordinances in metropolitan area X X Share of jurisdicons subject to low-density- only zoning in metropolitan area Region 2003 Pendall et al. Share of jurisdicons subject to low-density- only zoning in metropolitan area X X Share of jurisdicons subject to high-density allowed zoning in metropolitan area Region 2003 Pendall et al. Share of jurisdicons subject to high-density allowed zoning in metropolitan area X X Share of jurisdicons that employ growth management tools in metropolitan area Region 2003 Pendall et al. Share of jurisdicons that employ growth management tools in metropolitan area X X Share of jurisdicons with building moratoria in metropolitan area Region 2003 Pendall et al. Share of jurisdicons with building moratoria in metropolitan area X X Share of jurisdicons with an Adequate Public Facilies ordinance in metropolitan area Region 2003 Pendall et al. Share of jurisdicons with an Adequate Public Facilies ordinance in metropolitan area X X Share of jurisdicons with an Affordable Housing density bonus in metropolitan area Region 2003 Pendall et al. Share of jurisdicons with an Affordable Housing density bonus in metropolitan area X X Share of jurisdicons with affordable housing inclusionary zoning in metropolitan area Region 2003 Pendall et al. Share of jurisdicons with affordable housing inclusionary zoning in metropolitan area X X

E-18 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Share of jurisdicons where the allowable density was reduced since 1994 in metropolitan area Region 2003 Pendall et al. Share of jurisdicons where the allowable density was reduced since 1994 in metropolitan area X X Share of jurisdicons where the allowable density was increased since 1994 in metropolitan area Region 2003 Pendall et al. Share of jurisdicons where the allowable density was increased since 1994 in metropolitan area X X Share of jurisdicons with building or populaon growth restricons in metropolitan area Region 2003 Pendall et al. Share of jurisdicons with building or populaon growth restricons in metropolitan area X X Average annual precipitaon (inches) of metropolitan area Region 2012 NCDC Average annual precipitaon (inches) of metropolitan area X X Percent of possible sunlight in metropolitan area Region 2012 NCDC Percent of possible sunlight in metropolitan area X X Average temperature of metropolitan area Region 2012 NCDC Average temperature of metropolitan area X X Average number of days per year with highs over 90°F in metropolitan area Region 2012 NCDC Average number of days per year with highs over 90°F in metropolitan area X X Average number of days per year with lows below 32°F in metropolitan area Region 2012 NCDC Average number of days per year with lows below 32°F in metropolitan area X X Average snowfall per year (inches) in metropolitan area Region 2012 NCDC Average snowfall per year (inches) in metropolitan area X X Daily parking rate within 0.5 mi of project sta‚ons Project Catchments 2011 Parking In Mo‚on Daily parking rate within 0.5 mi of project sta‚ons X

E-19 Metropolitan-Level Models Project-Level Models Indicator Geographic Level Date Range Source Indicator Considered Observed Effect Considered Observed Effect (incl. endogenous vars) Observed Effect (excl. endogenous vars) Daily parking rate in the CBD Region 2011 Parking In Moon Daily parking rate in the CBD X -0.07 -0.08 Maximum hourly parking rate Region 2009 Colliers Internaonal Maximum hourly parking rate X Maximum parking rate for 12 hrs Region 2009 Colliers Internaonal Maximum parking rate for 12 hrs X Maximum parking rate for 24 hrs Region 2009 Colliers Internaonal Maximum parking rate for 24 hrs X Maximum monthly parking rate (r) Region 2009 Colliers Internaonal Maximum monthly parking rate (r) X Maximum monthly parking rate (u) Region 2009 Colliers Internaonal Maximum monthly parking rate (u) X Average gas price within counes within the metropolitan area Region 2009 NHTS Average gas price within counes within the metropolitan area X Populaon-weighted average gas price within counes within the metropolitan area Region 2009 NHTS Populaon-weighted average gas price within counes within the metropolitan area X Average distance to the CBD of project staons Project 2008 GIS Average distance to the CBD of project staons X Average Walk Score at all staon locaons within project Project 2012 Walk Score Average Walk Score at all staon locaons within project X Average Walk Score at all staon locaons in metropolitan area Region 2012 Walk Score Average Walk Score at all staon locaons in metropolitan area X X

F-1 APPENDIX F: Fixed-Guideway Projects Included in Analysis State City Project Name Mode Type Opening Year Route- miles Avg Daily Wkdy Ridership Capital Cost (M$2009) Sta‚on-Area Employment Sta‚on- Area Popula‚on Daily CBD Parking AZ Phoenix Metro Light Rail LRT Inial 2008 20 40,772 1,231 187,816 74,135 5 CA Los Angeles Long Beach Blue Line LRT Inial 1990 45 79,349 1,658 185,178 180,511 15 CA Los Angeles Green Line LRT Expansion 1995 20 30,935 1,225 66,818 74,088 15 CA Los Angeles Pasadena Gold Line LRT Expansion 2003 14 23,681 1,022 102,982 105,065 15 CA Los Angeles Red Line (Segment 1) HRT Expansion 1993 3 26,073 2,566 136,311 48,170 15 CA Los Angeles Red Line (Segment 2) HRT Expansion 1996 7 45,410 2,891 70,634 174,905 15 CA Los Angeles Red Line (Segment 3) HRT Expansion 1999 7 30,138 1,733 25,292 28,817 15 CA Los Angeles Orange Line BRT Expansion 2005 14 21,940 371 46,107 83,112 15 CA Sacramento Sacramento Stage I LRT Inial 1985 18 31,071 360 63,851 42,573 12 CA Sacramento Mather Field Road Extension LRT Extension 1998 6 6,711 44 7,599 18,996 12 CA Sacramento South Phase 1 LRT Expansion 2003 6 9,877 225 9,559 27,610 12 CA Sacramento Sacramento Folsom Corridor LRT Extension 2004 11 6,587 274 40,202 15,579 12 CA San Diego Blue Line LRT Inial 1981 25 41,361 986 187,905 93,665 16 CA San Diego Orange Line LRT Expansion 1986 22 23,113 633 38,798 81,575 16 CA San Diego Mission Valley East LRT Extension 2005 6 4,203 521 10,650 18,710 16 CA San Francisco Inial BART HRT Inial 1974 72 284,162 6,960 311,300 269,182 30 CA San Francisco BART SFO Extension HRT Extension 2003 9 19,501 1,598 27,877 14,065 30 CA San Jose San Jose North Corridor LRT Inial 1987 17 11,272 757 100,999 56,579 14

F-2 State City Project Name Mode Type Opening Year Route- miles Avg Daily Wkdy Ridership Capital Cost (M$2009) Sta‚on-Area Employment Sta‚on- Area Popula‚on Daily CBD Parking CA San Jose Tasman West LRT Expansion 1999 8 1,977 416 38,728 15,101 14 CA San Jose Tasman East LRT Expansion 2001 5 3,340 335 17,452 20,494 14 CA San Jose VTA Capitol Segment LRT Extension 2004 3 2,385 205 4,819 29,645 14 CA San Jose VTA Vasona Segment LRT Expansion 2005 5 3,848 374 29,902 38,766 14 CO Denver Central Corridor LRT Ini‚al 1994 5 36,403 161 96,104 25,269 13 CO Denver Denver Southwest Corridor LRT Extension 1999 9 8,728 228 16,780 9,893 13 CO Denver Denver Southeast (T-REX) LRT Expansion 2006 19 16,298 876 86,349 26,811 13 FL Miami Metrorail HRT Ini‚al 1984 21 58,121 2,366 146,439 109,235 9 FL Miami South Florida Tri-Rail Upgrades CR Enhancement 2007 72 36,510 394 76,384 52,405 9 GA Atlanta North / South Line HRT Expansion 1985 22 113,948 3,194 176,597 47,472 7 GA Atlanta North Line Dunwoody Extension HRT Extension 1996 2 9,381 611 16,327 4,253 7 IL Chicago O'Hare Extension (Blue Line) HRT Extension 1984 8 21,350 469 30,026 10,811 29 IL Chicago Orange Line HRT Expansion 1993 9 32,334 778 20,176 65,718 29 IL Chicago Douglas Branch HRT Extension 2005 7 16,035 503 28,652 115,554 29 IL Chicago Metra North Central CR Expansion 1996 55 2,201 247 23,971 34,463 29 IL Chicago Metra Southwest Corridor CR Extension 2006 11 4,125 211 14,978 35,312 29 MD Bal‚more Central Line LRT Expansion 1992 23 24,541 531 106,966 62,984 14 MD Bal‚more Three extensions LRT Extension 1997 7 4,448 140 35,891 15,304 14 MD Bal‚more Bal‚more Metro HRT Ini‚al 1985 12 39,023 2,040 72,145 59,848 14 MN Minneapolis Hiawatha Corridor LRT Ini‚al 2004 12 30,518 454 167,692 42,224 11 NJ Jersey City Hudson-Bergen MOS 1 and 2 LRT Expansion 2003 15 40,100 1,809 88,742 211,414 38 NJ Newark Newark Elizabeth MOS-1 LRT Expansion 2006 1 1,065 214 16,108 19,599 38 NJ Trenton Southern NJ Light Rail Transit System LRT Expansion 2002 28 8,150 1,166 24,910 64,862 24 NY Buffalo Buffalo Metro Rail LRT Ini‚al 1984 6 24,076 951 65,298 45,417 7 OH Cleveland Cleveland Healthline BRT Expansion 2008 7 12,850 197 114,837 32,797 12 OR Eugene Eugene EmX BRT Ini‚al 2007 4 6,600 26 27,994 17,128 4

F-3 State City Project Name Mode Type Opening Year Route- miles Avg Daily Wkdy Ridership Capital Cost (M$2009) Sta‚on-Area Employment Sta‚on- Area Popula‚on Daily CBD Parking OR Portland Portland MAX Segment I LRT Ini al 1986 15 60,229 508 116,225 63,679 9 OR Portland Portland Westside/Hillsboro MAX LRT Extension 1996 18 34,223 1,320 64,900 54,053 9 OR Portland Portland Airport MAX LRT Expansion 2001 6 3,005 156 5,319 3,108 9 OR Portland Portland Interstate MAX LRT LRT Expansion 2004 6 7,992 333 16,343 18,279 9 PA Philadelphia SEPTA Frankford Rehabilita on HRT Enhancement 2005 5 45,103 1,186 24,336 110,510 24 TX Dallas S&W Oak Cliff and Park Lane LRT Extension 1997 20 46,713 1,137 145,557 68,864 6 TX Dallas North Central LRT Extension 2002 13 12,304 450 57,228 20,750 6 UT Salt Lake City North-South Corridor LRT Ini al 1998 15 31,405 412 74,476 27,619 12 UT Salt Lake City Medical Center Ext. LRT Extension 2003 2 3,358 87 22,057 1,709 12 UT Salt Lake City University Ext. LRT Extension 2001 3 7,285 111 17,532 15,945 12 WA Sea—le Sea—le Central Link Light Rail Project LRT Ini al 2006 14 19,719 2,583 161,394 61,817 22

G-1 APPENDIX G: Model Technical Information The project-level ridership models were executed in Stata using the regress command. The coefficients were estimated using ordinary least squares and robust errors to account for clustering across metropolitan areas. The system-level PMT models were executed in Stata using the xtreg command with metropolitan-area ID as the cluster-level variable and maximum likelihood parameter estimation. A comprehensive guide to implementing panel regressions in Stata is Multilevel and Longitudinal Modeling Using Stata by Sophia Rabe-Hesketh and Anders Skrondal (2nd Edition, Stata Press, College Station TX, 2005). To compare the goodness-of-fit between multiple models, we employed the Bayesian Information Criterion (BIC) post-estimation statistic. BIC is a penalized goodness-of-fit measure that approximates the probability that a model is most likely given the data (Washington, Karlaftis & Mannering 2011, p. 400). Generally BIC values closer to zero are associated with better models, and we used this value to iteratively compare pairs of models with different forms.

H-1 APPENDIX H: Focus Groups, Phase 2, Topic Responses The sections below provide more detail on responses to the following questions: • Would a spreadsheet tool as proposed be useful to you? In what circumstances might you use it? • Who would be the audience for the model outputs? • What sorts of outputs would be most helpful for you? • How hard would it be for you to generate the input data needed to apply the tool? • What improvements can be made to the tool’s input and output interfaces? • How can we make the handbook most useful to practitioners? Question: Would a spreadsheet tool as proposed be useful to you? In what circumstances might you use it? Who would be the audience for the model outputs? Participants in the focus group at the APTA Rail Conference agreed that the tool would be useful for a quick evaluation of corridors and/or new transit lines that are often suggested by transit agency board members and the public. It would demonstrate to board members and citizens where their ideas fall, in terms of ridership and cost per rider, compared to national examples. The tool would show some communities whether or not the rail transit they want has potential merit. It could also be used for scenario testing, such as testing changes in land use and parking policy, changes in land use, and changes in alignment and station locations. Regional prioritization of potential corridors was also mentioned as a potential use. Participants in the Houston–Galveston Area Council (HGAC) focus group and the telephone interviews had a similar reaction with regard to the spreadsheet’s utility for quickly comparing scenarios. Those attending the focus group held with staff at the San Francisco Bay Area Metropolitan Transportation Commission (MTC) thought that the spreadsheet tool might be of use by some of the transit agencies and local jurisdictions in the Bay Area. They noted that project sponsors generally do these sorts of analyses before presenting projects to MTC. As far as the MTC itself is concerned, however, the sense was that they would continue to use the regional model for planning analyses. Their impression was that the spreadsheet tool would not save much time, and they would have less confidence in the reliability of the results. It was agreed that the tool is not a substitute for a good regional model. One of the MTC participants asked why a regression framework had been used, which he said does not capture how people behave when they travel and does not offer insights into the potential competitiveness of other modes. He wondered what the outputs of the model tell the user, i.e., what the user learns that can be acted upon. The telephone interviewees found utility in the spreadsheet tool, however. One interviewee, from a smaller transit agency, thought that the tool could be useful for prioritizing corridors and for

H-2 conversations with local governments about station-area development plans/policies. A second small transit agency interviewee called the tool “just what we need.” Another interviewee, from a large MPO, saw utility for scenario planning and noted that the regional model could not be used for everything. These responses suggest that some agencies may find the outputs helpful while others may not. Question: What sorts of outputs would be most helpful for you? Participants in the APTA Rail focus group suggested several additional histograms or bar charts allowing users to compare proposed projects with others in the database. In addition to cost per rider, it was suggested that users be able to compare projects in terms of the number of riders per mile. They would like to be able to turn off and on bars in the chart so comparisons can be made with similar projects (e.g., compare with other starter lines, same mode, peer cities, etc.). Participants would like to be able to add all their existing lines and highlight them in the histogram, so that new projects could be compared to existing lines within the same urban area. Both focus group participants and telephone interviewees said that users would also like to be able to save scenario results so that they can easily compare one scenario with another. Question: How hard would it be for you to generate the input data needed to apply the tool? Participants in the focus groups and telephone interviews noted that assembling the station-area population and employment data using GIS was somewhat demanding, but expressed no significant concerns other than the time required. Most planning agencies are thought to have the necessary capability. One MPO interviewee stated that agencies should be “doing this anyway” to evaluate alternative projects and corridors. At the MTC focus group, one participant suggested that the tool would be easier to use if population and employment data were embedded into the database. Users would still have to create their own buffers around stations, but the base data would be there. The focus group at the APTA Rail Conference pointed out that some entries on the input screen use terms that have different meanings in the transit industry. The input screen should ask for end- to-end travel time, rather than speed, because speed is often interpreted as average top speed between stations. The meaning of “bus connections” needs to be spelled out—is it the number of bus routes or number of buses? The meaning of “route-miles” needs to be made explicit. A participant in the MTC focus group cautioned that some smaller MPOs do not have the capability to use macros in Excel. Question: What improvements can be made to the tool’s input and output interfaces? One of the telephone interviewees from a small transit agency suggested that, in addition to reporting out ridership and cost per rider, the tool might identify those inputs which were most favorable to ridership and those where the project could be strengthened. The user would then know where to focus efforts to enhance a project’s likelihood of success. It was also suggested by the HGAC focus group that a help function be added to the tool. For example, users might like to be able to click on the inputs on the input screen to get a pop-up box with definitions or instructions.

H-3 Question: How can we make the handbook most useful to practitioners? The HGAC focus group said that their confidence in the tool’s results would be greater if the handbook explained the inner workings of the spreadsheet tool and how it was validated. This might include a comparison between what the tool might have predicted for completed projects, compared with the actual ridership. Noting that there were likely to be outliers in the data, one might question the tool’s reliability for particular cases. They suggested that the handbook express appropriate caveats on the use of results. Similarly, an interviewee urged that the handbook explain the inner workings of the spreadsheet sufficiently that users could explain it to elected officials and other interested parties. There was also discussion, in the HGAC and MTC focus groups, of putting the handbook and spreadsheet tool online for ease of access. A participant in the MTC group urged us to make the tool fun and easy to use, like a game. He pointed the research team to a Transit Competitiveness Index tool prepared for MTC’s Sustainability Study, and specifically to that tool’s graphical interface, that made the application fun to use. Similarly, a telephone interviewee referred the research team to a user-friendly tool on Portland Metro’s website that allows users to test a variety of transit scenarios and make trade-offs. Some participants believed that the method would be useful for an initial evaluation of potential transit projects, helping prioritize alternatives, providing a means for scenario testing, and demonstrating to the community the implications of different options. Comments were offered on additional capabilities that would make the tool more useful, including improved visualization of the tool’s outputs in the form of charts and tables. Overall, participants expressed no significant concerns about the difficulty of generating the input data for the tool, aside from time required. Many requested that the handbook provide clear and detailed information on the underlying mechanisms of the spreadsheet tool in order to feel more confident about the validity of its results and to better explain the tool’s outcomes to interested parties.

I-1 I.1 South Line (Charlotte, NC) The South Line, now called the LYNX Blue Line, is a 9.6-mile, 15-station light rail project extending south from Uptown Charlotte (the city’s central business district) to Interstate 485 in southern Mecklenburg County near the South Carolina State border. The facility, completed in 2007, generally parallels North-South Interstate 77 and serves considerable commuter traffic accessing the 80,000 jobs located in Charlotte’s CBD. This case study suggests that transit planning can be driven by land use aspirations as well as transit criteria. Additionally, the case demonstrates the complex balancing act that occurs as planners seek to achieve multiple goals, as well as the detrimental sacrifices that can occur in an attempt to maximize a single measure of success. This case study highlights the art of transit planning as opposed to the engineering and quantitative rigor often associated with major transportation projects. I.1.1 Establishing Charlotte Rail Transit Rail transit planning was initiated in Charlotte in the 1980s and culminated in the Transit Corridor System Planning Study of 1989 (FTA and CATS 2002). By 1994, the Charlotte Transitional Analysis had identified rail transit corridors that would support the region’s overall Centers and Corridors Concept Plan, which identified five radial corridors of dense urban Figure I-1: Route Diagram for LYNX Blue Line, Charlotte, North Carolina APPENDIX I: Detailed Case Study Write-Ups and Regional Profiles

I-2 development with wedges of low-density single-family housing in between. In 1998, Mecklenburg County residents voted for a half-cent sales tax measure that was dedicated to implementing the region’s 2025: Integrated Transportation and Land Use Plan (FTA and CATS 2003). In 1997, before citizens approved the transit tax, the city purchased a 3.3-mile segment of abandoned Norfolk Southern Railroad right-of-way south of Uptown Charlotte and began operating a historic trolley over just a few miles of the tracks between downtown and several emerging pockets of redeveloped warehouses.1 After passage of the 1998 tax measure, planning began for light rail transit along a much longer segment of that same Norfolk Southern Railroad right-of-way and culminated with the South Corridor Major Investment Study in 2000 (FTA and CATS 2002). The FTA approved the South Corridor LRT project for preliminary engineering in August 2000, and a Record of Decision on the project’s environmental documentation was issued in May 2003 (FTA NSFA 2005). On May 6, 2005, the FTA entered into a Full-Funding Grant Agreement (FFGA) providing a federal commitment of $192.94 million in New Starts funds. The total project cost under the Full- Funding Grant Agreement (FFGA) was $426.85 million, with the majority of funds coming from state and local sources (FTA SCLRT 2005). In addition to and separate from the project budget, the City of Charlotte provided $72 million in complementary infrastructure improvements as part of the South Corridor Infrastructure Program (SCIP) (CATS 2012). The South Corridor project, called the LYNX Blue Line, opened for revenue service in November 2007. The South Corridor’s northern terminus is at the intermodal Charlotte Transportation Center (CTC) in the heart of Uptown Charlotte, which houses approximately 80,000 jobs and 15,000 residents.2 Major central city destinations include Charlotte’s convention center and two major league sports arenas. The route heads south from there through the mixed-use South End neighborhood and parallels South Boulevard (NC 521) and I-77 until it reaches its terminus at a park-and-ride station near Interstate 485, roughly three miles from the South Carolina border. The South Corridor was the first rail corridor to be built among several rail corridors envisioned for the Charlotte region. The decision to construct this line before others provided us with useful insight into indicators that have been used to predict transit project success. Based on a high-level assessment of right-of-way availability, constraints on parallel regional travel network segments, and opportunities to promote regional agglomeration, the South Corridor was prioritized over others. Notably, there was very little debate about the selection of this alignment and few alternatives were seriously considered. In one circumstance, the City of Pineville was originally imagined as the rail line’s terminus south of Interstate 485, but the rail line proposal was truncated when the city’s council declined rail transit. While their decision was politically divisive and represented one of the few times that the Charlotte region’s political bodies were not united behind the centers and corridors vision, the shorter line was considered by many to be a benefit for transit operations and capital costs that made the project more attractive for federal funding.3 1 Interviewee AR, telephone conversation, 5/18/12. 2 Charlotte Center City Partners; http://www.charlottecentercity.org; Accessed 10/22/12. 3 Interviewee AS, in-person conversation, 8/30/12.

I-3 Though the South Corridor remains the only operating rail transit segment in Charlotte, a $1.2 billion extension of the South Corridor recently received federal funding and will double the extent of the rail system upon opening in 2017 (Spanberg 2012). The rail system is complemented by Charlotte Area Transit System’s (CATS’) extensive bus operations. At the time the South Corridor was planned, CATS operations were divided between express, local, and cross-town bus services (CATS 2012). Local and express bus services had typically terminated downtown at the Charlotte Transportation Center (CTC). When the South Corridor opened, many routes of the regional bus system were reoriented to serve as rail feeder services. Additionally, the sales tax funding used for the South Corridor helped expanded bus services throughout the region. As of 2012, 28% of rail riders also boarded buses during their trips. CATS rail services are branded distinctly from bus services. LYNX rail service along the South Corridor is called the Blue Line and operates seven days a week (weekdays 5:20 am to 2 am, Saturday 5:45 am to 1:56 am, Sunday 6:25 am to 12:26 am).4 During peak hours in the peak direction, headways are as low as 10 minutes. Generally, service is offered at 15-minute headways, with 20-minute headways later in the evening. In its first few months of operation, the Blue Line averaged between 12,000 and 17,000 weekday riders (CATS 2012). Ridership grew steadily, peaking at 21,700 in the fourth quarter of 2008.5 However, in the years since then ridership dropped to a consistent average weekday estimate of approximately 15,000 boardings (CATS 2012). Fluctuations in ridership were attributed to the 2008 peak in gas prices and bus service cuts due to declining sales tax receipts during the recession. Nearly 80% of ridership emanates from origin locations within the defined South Corridor study area, and over half of all trips are to and from downtown Charlotte. Actual ridership in 2009 exceeded the projected 2010 ridership estimates by approximately 2,800 trips. This is attributed to the fact that transportation models underestimated the number of riders that would travel long distances to ride the line for relatively short rail trips (CATS 2012). Interestingly, transit agency staff had anticipated this travel pattern based on parking costs to access government buildings and special event venues in the Uptown area.6 I.1.3 Planning a Successful Transit Project Close collaboration of local agencies led to a transit project that epitomizes transportation and land use coordination.7 The South Corridor project was planned by a consortium of Charlotte- Mecklenburg County agencies. CATS is the builder and operator of the line as well as an agency of the consolidated city-county local government. As such, it is part of the same organization as the Charlotte Department of Transportation (CDOT) and the Charlotte-Mecklenburg Planning Department (Planning), two other significant players in the planning and implementation of the South Corridor transit project.8 In addition, planning of the line was strongly influenced by the Federal Transit Administration, which provided nearly half of the funding for the project. As discussed below, several interviewees suggested that federal measures of transit project success led 4 Lynx Service; http://charmeck.org/city/charlotte/cats/Pages/default.aspx; Accessed 10/22/12. 5 APTA; http://www.apta.com/resources/statistics/Documents/Ridership/2008_q4_ridership_APTA.pdf; Accessed 7/20/12. 6 Interviewee AS, in-person conversation, 8/30/12. 7 Interviewee AT, telephone conversation, 8/27/12. 8 Interviewee AR, telephone conversation, 5/18/12. I.1.2 South Line Operations

I-4 to decisions that undermined the quality of the South Corridor and contributed to its costly retrofit as the Blue Line is extended to northeast Charlotte. The selection of the South Corridor as Charlotte’s first rail investment over the four other corridor options identified in the Centers and Corridors Concept Plan could be interpreted as a choice that optimally met the diverse interests of three stakeholder local departments—CDOT, CATS, and Planning—by addressing traffic concerns, selecting a viable route from a cost perspective, and aiding land use change. As several interviewees shared with me on separate occasions, the Charlotte City Council selected the South Corridor in the 1990s for three primary reasons: the corridor was parallel to heavily congested Interstate 77 in the fastest growing corridor in the region; the corridor coincided with available Norfolk Southern right-of-way; and there was tremendous potential for redevelopment along the corridor that was fitting with the region’s long-range land use vision.9,10,11,12,13 During the 1990s, the second-largest U.S. banking center established itself in Charlotte’s downtown, one of the fastest growing cities in the nation in one of the nation’s fastest growing regions at the time (FTA and CATS 2003). Much of the region’s housing and commercial development in the 1980s and 1990s occurred between the Interstate 77 corridor and the Independence Boulevard corridor, which proceeds radially southeast out of downtown Charlotte.14 High-end shopping malls were expanded in the quadrant, high-end housing was built within the belt loop, and sprawl continued to spread further south in South Carolina. Freeways north of Charlotte had not yet been expanded and growth had not yet occurred there. As one transit professional put it, “There was momentum to the south.” With growth came congestion. Though touted as providing a transport alternative for transit- dependent populations, in many ways the South Corridor project was framed in community meetings as a release valve for growing traffic as more and more households moved to northern South Carolina and workers commuted to downtown Charlotte.15 The major transportation infrastructure within the area was at its maximum physical capacity.16 In South Carolina and just inside the North Carolina border, Interstate 77 was an eight-lane roadway. Within the Interstate 495 beltway, urban development and bridge bottlenecking constrained the expansion of the Interstate to six lanes. There were few alternate radial routes into the city, and gridlock was common on the Interstate. One of the possible alternate routes was the South Boulevard radial arterial that paralleled Interstate 77. South Boulevard extended from central Charlotte to the Interstate 485 belt loop and was four lanes with vehicle volumes exceeding 65,000 along much of its length.17 Paralleling the corridor was a Norfolk Southern Railroad right-of-way, part of which was disused and part of which was a limited-use spur to access a handful of industrial sites. The remainder of the railroad right-of- way, paralleling the southern reaches of South Boulevard before crossing south of Interstate 495, was a fully operational Norfolk Southern mainline. 9 Interviewee AD, telephone conversation, 8/24/12. 10 Interviewee AN, in-person conversation, 8/30/12. 11 Interviewee AS, in-person conversation, 8/30/12. 12 Interviewee AT, telephone conversation, 8/27/12. 13 Interviewee AR, telephone conversation, 5/18/12. 14 Interviewee AD, telephone conversation, 8/24/12. 15 Interviewee AU, in-person conversation, 8/30/12. 16 Interviewee AD, telephone conversation, 8/24/12. 17 Interviewee AT, telephone conversation, 8/27/12.

I-5 In 1997, the City of Charlotte purchased 3.3 miles of disused rail property to preserve for future transit use.18 In 1998, the city council allocated $16.7 million to build along it a two-mile trolley line from Uptown Charlotte to Charlotte’s Historic South End.19 It was intended to accommodate vintage trolley services and serve as a capital “down payment” on eventual light rail transit services. The Charlotte Trolley operation was embraced by the community and delineated a clear path forward for expanded rail services. One city planner suggested that, “It was a Disney-like ride but it got the imagination going, and property owners were very enthusiastic about expanding the transit operations.”20 When it came time to grow Charlotte’s transit system, the planner thought “we had a clear and obvious location for the route along an abandoned freight rail right-of-way. It was a fairly simple decision-making process.” The congested Interstate 77 corridor and available right-of-way coincided with Charlotte’s overall regional development vision. As one transit planner noted, “The driver that started the Charlotte Transit program was really the community’s vision for how they wanted to develop as a city.”21 As a land use planner explained, “We had planted the seed of rapid transit in the early to mid-1980s”22 The regional vision was enumerated in the ‘2025: Integrated Transportation and Land Use Plan’. Generally, the idea of the plan was to concentrate the majority of future regional growth into five radial corridors. Per a transit professional, “The idea was to [use the] transit system to help create an environment for higher density, pedestrian-friendly, mixed-use, so-called transit-oriented or transit- supportive land use.”23 This smart growth strategy would allow for growth while minimizing traffic impacts and maintaining the suburban single-family neighborhoods that defined much of Charlotte. Regional discussions were focused on providing options to a diverse population with diverse interests, so that the region could remain competitive in the long term.24 According to public sector staff, the community was very much in favor of the holistic transportation and land use vision for the South Corridor. When discussing the importance of affirming the regional vision before advocating for a particular transportation project, one planner stated that “we didn’t sell the South Corridor, we sold a system.”25 The South Corridor was considered an obvious place to begin implementation. As evidence of the consensus on this matter, one City of Charlotte staffer noted that after numerous public meetings they received approximately five comments on the South Corridor’s draft environmental statement and only one comment on the final.26 Additionally, a 2007 measure to repeal the county’s transit sales tax was soundly defeated by voters just weeks before the South Corridor opened for revenue service.27 Consensus was also sought within government. In discussing land use planning that occurred alongside the project, a planner reminisced about their refrain during the South Corridor planning process: “Transit isn’t the end, it’s the means for us to accomplish this new community that we’re trying to build.”28 As another transit planner noted, “We had these joint collaborative meetings and 18 Interviewee AS, in-person conversation, 8/30/12. 19 Charlotte Trolley; http://www.charlottetrolley.org; Accessed 11/2/12. 20 Interviewee AR, Telephone conversation, 5/18/12. 21 Interviewee AT, telephone conversation, 8/27/12. 22 Interviewee AN, in-person conversation, 8/30/12. 23 Interviewee AD, telephone conversation, 8/24/12. 24 Interviewee AN, in-person conversation, 8/30/12. 25 Interviewee AN, in-person conversation, 8/30/12. 26 Interviewee AT, telephone conversation, 8/27/12. 27 Interviewee AV, in-person conversation, 8/30/12. 28 Interviewee AN, in-person conversation, 8/30/12.

I-6 we did things which I think are pretty unique for transit agencies. We met [with the other departments] and talked about where we wanted the parking lots. Not just, ‘Where’s the demand [for parking] going to be?’ 29 Transit professionals were also at the forefront of the land use conversation. As one transit veteran explained, “One of the things that always concerned me about our plan here in Charlotte was its fit with the size of the downtown population.” 30 Reflecting the research of Pushkarev and Zupan, the veteran went on to say, “We were talking about 55,000-60,000 jobs in the downtown area on a daily basis and that worried me. I always had that in mind that we needed to have at least 100,000 jobs to support five corridors of rapid transit. I had a lot of discussion with the downtown business interest about attracting additional major employers for the downtown area.” Also a reflection of the Pushkarev and Zupan research, transit planners referenced land use plans when refining station locations. As one transit planner recalled telling a land use planner, “Land has to be [planned to] a certain intensity for us to consider a station there.”31 The selection of the South Corridor over one of the other four corridors was partly predicated on localized economic development potential. As one planner described the selection process, “There was so much more momentum, in terms of fulfilling and achieving truly transit-oriented development. I think that corridor had a lot more momentum behind it from the development community, from property owners, from [our interpretation of] where we could be successful. And I don’t think there were many of the other corridors that were as [well] positioned as the South Corridor was.” 32 Also, several property-owning constituents who had been proponents of the trolley alignment had promised to invest in the South Corridor.33 As the planner described it, “We also had some grassroots folks who were in what’s now called South End who were strong advocates for a pedestrian-friendly type of development, either with rapid transit or not. […] So I think we had to head south.” The design of the South Corridor project was also based on local real estate development potential. A transit planner explained, “The emphasis for locating the stations was almost more development first, access second. Where did it make sense from a […] development perspective and the land use perspective to have stations, and then how can we provide access to those locations?”34 Based on this set of criteria, the original proposal for the project included 19 stations. As a land use planner explained, “The stations were considered beads on a chain, and they moved up and down as we considered the walkability of areas and other factors.”35 The interviewee went on to say, “There was political pressure to have more stations closer together. Every property owner wants to be right at the station and not farther away.” Transit planners worked with land use planners to eliminate some proposed stations to improve transit efficiency. As a transit planner explained, “I didn’t want to have a street car—a light rail running on a street car line, if you will—with stations every couple of blocks.” Transit planners added grade separations, shifted station locations, and balanced the operational requirements of light rail with the land use intentions of the region and local land use planners. 29 Interviewee AW, in-person conversation, 8/30/12. 30 Interviewee AD, telephone conversation, 8/24/12. 31 Interviewee AR, telephone conversation, 5/18/12. 32 Interviewee AN, in-person conversation, 8/30/12. 33 Interviewee AN, in-person conversation, 8/30/12. 34 Interviewee AT, telephone conversation, 8/27/12. 35 Interviewee AR, telephone conversation, 5/18/12. ”

I-7 The entire alignment was even shifted off the existing right-of-way in one circumstance to accommodate real estate development near a proposed station. Scallybark station and several hundred yards of track were located in the median of South Boulevard to provide access to land that was considered ripe for development by urban planners. The original Norfolk Southern Railroad tracks had been slightly elevated on a berm at the edge of South Boulevard, blocking access to several undeveloped acres. To promote transit-oriented development, “We took out the berm and made the land accessible to the major thoroughfare and a rail station.”36 As an alignment option that addressed regional traffic concerns, had an existing right-of-way adequate for light rail transit services, and aided land use change, the South Corridor was the obvious choice for a rail project in Charlotte. In fact, there were no serious alternatives officially considered. One alternative in the Draft Environmental Impact Statement (DEIS) considered an extension of the light rail alignment to Pineville south of Interstate 485 but the northern portions of the proposal were nearly identical to what was ultimately built (FTA and CATS 2002). While some of the earliest studies included 19 stations, most station locations were initially aligned with major East-West arterials and were never significantly altered (CATS 2012). Other alternatives in the final environmental documents included a no-build alternative and a “Transportation Systems Management” option that included enhanced bus services on South Boulevard (FTA and CATS 2003). Interviewees repeatedly suggested that there was consensus about the South Corridor alignment’s design and that the locally preferred transit alignment was an obvious solution. As one planner put it, “I really think it was by accident that all the planets aligned for us to move forward with the South Corridor.”37 In spite of the consensus around the South Corridor’s selection and light rail proposal, some problems did arise because the success metrics used to define, select, and design the South Corridor were not synonymous with the federal funding criteria. Interviewees in Charlotte suggested that prioritizing one measure of success over others, as was the case with the federal New Starts funding criteria during the George W. Bush administration, was not a valid means of evaluating success. As one transit professional framed it, “There's an art associated with making these projects happen.”38 In fact, they considered the FTA’s focus on a single quantitative measure detrimental to the project’s success. Economic development, walkability, and sustainability were strong considerations in Charlotte but were not highly regarded by the Federal Transit Administration at that time. “That stuff was important to us locally but wasn't important to those people who were evaluating whether or not they wanted to invest a couple of hundred million dollars in the construction of this project.”39 Federal funding evaluations included a ridership-based cost-effectiveness hurdle regardless of the other benefits of the project. Charlotte, with its focus on changing the way the region developed, did not meet the cost- effectiveness hurdle in early evaluations. Modifications for the sake of cost savings were required. Planners could lower costs by increasing proposed train frequencies and thereby reducing the number of train cars that needed to be purchased, running two-car operations rather than three-car train sets.40 This allowed them to correspondingly reduce platform lengths, power system units, and 36 Interviewee AD, telephone conversation, 8/24/12. 37 Interviewee AN, in-person conversation, 8/30/12. 38 Interviewee AD, telephone conversation, 8/24/12. 39 Interviewee AD, telephone conversation, 8/24/12. 40 Interviewee AT, telephone conversation, 8/27/12.

I-8 track ballast quality. As one planner put it, “We ended up coming up with a [plan] that justified the project according to their cost-effectiveness criteria and travel time savings.”41 One land use planner suggested that Charlotte was doing whatever it took to get their first line and get the system plan off the drawing boards.42 Today, demand exceeds the two-car train sets that are operated on the line, and the extension of the Blue Line to northeast Charlotte will further increase that demand. With considerable inconvenience and at great expense, CATS is currently retrofitting the existing South Line to accommodate three-car trains. The FTA’s singular focus on cost effectiveness led to short-sighted decision-making that complicated matters in the future. In addition to considering the federal cost-effectiveness criteria and the three measures of success enumerated above—alleviation of traffic, viable completion, and beneficial land use impacts — Charlotte interviewees noted two other measures of success they considered during the planning of the South Corridor. First, they sought to make the corridor safer. Prior to the implementation of the line, pedestrians were subjected to hostile environments along South Boulevard and its cross streets. The project and attendant land use changes were planned to address some of these concerns.43 Second, there had been a number of fatalities resulting from collisions between freight rail traffic and automobiles crossing the railroad tracks.44 Roadway intersection conditions that had caused traffic to sometimes back up across the tracks were addressed with South Boulevard intersection improvements. Also, the light rail line was elevated over several arterials to reduce dangerous interactions with high-volume roadways. From the standpoint of project planners, the light rail project had to improve safety conditions on the corridor to be successful. Additionally, transit project stakeholders sought to design a project that could successfully launch the region’s overall transit system implementation.45 The South Corridor line was selected because it had the greatest chance of success and would provide local political momentum for further investment. Inadvertently, the project also paved the way for the region’s second project, the northeast extension of the South Line, by setting a precedent with the North Carolina Department of Transportation (NCDOT). South Boulevard was also the North Carolina state highway NC 521, and the shift of the Scallybark station to the median of South Boulevard was the first implementation of light rail on a NCDOT facility.46 Numerous issues had to be addressed on that short segment of shared road space, including the city taking responsibility for maintenance along that section of roadway. As one planner explained, “With the state being able to see that it does work, our next project is going to be in a U.S. route for the majority of it. If we hadn’t done that, we might have a lot stiffer opposition to move ahead in the next project.”47 In addition to the numerous measures of success considered by South Corridor stakeholders, Charlotte planners also mentioned several qualitative “rules of thumb” that they considered when predicting the potential success of the LYNX Blue Line. The most often mentioned of these were threshold indicators of political support that suggested to planners that the project would actually be implemented. One planner recalled how he evaluated a job offer to join the transit planning team during the early stages of the project and made his decisions based on three project champions: the 41 Interviewee AD, telephone conversation, 8/24/12. 42 Interviewee AN, in-person conversation, 8/30/12. 43 Interviewee AV, in-person conversation, 8/30/12. 44 Interviewee AT, telephone conversation, 8/27/12 45 Interviewee AS, in-person conversation, 8/30/12 46 Interviewee AU, in-person conversation, 8/30/12 47 Interviewee AU, in-person conversation, 8/30/12

I-9 mayor, a county commissioner, and the new chief executive of the transit agency.48 The mayor and county commissioner were from different political parties and took turns touting the project depending on the audience at their joint appearances.49 The mayor’s focus was making sure the project was “driven by data, technical information, [and] the ability to really build our community. [He asked], “where’s the need, where’s the economic development potential, where’s the best ridership?””50 To implement the transit system expansion, the city hired the then Chair of the American Public Transportation Association. This was a clear sign that the project leadership was skilled and politically savvy.51 Additionally, the team leading the rail system planning had considered how attractive each of Charlotte’s proposed rail corridors would be to the North Carolina congressional delegation that would be relied upon to advocate for the project.52 Coming out of the ‘2025 Plan’ process, they decided to pursue the South Corridor because it would have the support of several members of the South Carolina congressional delegation as well. Planners also thought that building the region’s first line to the South Carolina border was an indicator of ridership potential. Fast-growing South Carolina suburbs represented a major origin of transit trips and served as one end of a “barbell,” with the South Corridor’s northern terminus in downtown as the other end.53 Initially, the southern end of the transit line’s “barbell” was intended to be the local mall in fast-growing Pineville.54 However, it was determined that a parking garage near Interstate 485 would serve as a much larger ridership generator because it could attract patrons from Pineville as well as other southern suburbs. Another indicator of ridership noted by project planners was the expansion of bus services and the realignment of bus routes that coincided with the start of Blue Line revenue service.55 For instance, several long haul bus routes were converted into multiple rail feeder bus routes. Planners considered this an access extension for bus service that allowed buses to reach further into neighborhoods that had only been served at the periphery prior to rail. Rising downtown parking costs also served as an indicator of potential ridership.56 Park-and-ride license plate surveys have borne this out. A measureable portion of riders drove from east and west Charlotte to ride the train north into downtown. Parking lot owners in downtown have also confirmed a decline in demand at their facilities. The South Corridor case study suggests that transit project planners consider a wide array of success indicators to predict performance across a number of measures. Those measures may be more related to indirect transit outcomes like land use impacts than to direct measures of success like ridership. This case study suggests that transit planning is a complex art that uses both qualitative indicators and quantitative forecasts to balance a number of expectations for a single fixed-guideway transit project. 48 Interviewee AT, telephone conversation, 8/27/12 49 Interviewee AS, in-person conversation, 8/30/12 50 Interviewee AN, in-person conversation, 8/30/12 51 Interviewee AT, telephone conversation, 8/27/12 52 Interviewee AS, in-person conversation, 8/30/12 53 Interviewee AS, in-person conversation, 8/30/12 54 Interviewee AR, in-person conversation, 8/30/12 55 Interviewee AS, in-person conversation, 8/30/12 56 Interviewee AV, in-person conversation, 8/30/12

I-10 I.2 North Central Corridor (Dallas, TX) The North Central Corridor, completed in 2002, is a rail extension of the original DART light rail starter system that opened in 1996. The 13.8-mile Red Line extension was located in a former Southern Pacific rail corridor paralleling North-South United States Highway 75, known locally as the North Central Expressway. The light rail corridor passes through the cities of Dallas and Richardson, terminating in the City of Plano. The project consists of nine new stations and the reconstruction of Park Lane Station, the former terminus of the Red Line. This North Central Corridor case study suggests that rail transit may be implemented as a release valve in highly congested, auto-oriented locations. For the North Central Corridor, the transit planning process focused on automobile congestion on parallel routes, park-and-ride stalls, and other auto-related criteria. Predictors of choice riders were a keen focus for planners, because only through their mode choice shift would the project achieve its traffic mitigation intent. While the project ultimately failed to meet its ridership targets, the project was still considered a success according to a number of alternative measures. Most fundamental, several of the alternative measures relate to the fact that a regional rail project was actually completed in this auto-oriented metropolis and that it remains in service today. I.2.1 Expanding Dallas Rail Transit In the early 1980s, the North Central Texas Council of Governments (NCTCOG) worked with the Urban Mass Transportation Administration (UMTA) to develop a rail plan for greater Dallas. (UMTA 1982) In 1983, the Dallas Area Rapid Transit Authority (DART) was formed upon the passage of a referendum in 14 cities and Dallas County.57 By 1984, DART was operating commuter-oriented, non-stop express bus service from Plano and Richardson in the northern Dallas suburbs to downtown Dallas. Also in 1984, the DART Board adopted light rail as the preferred mode for a planned 147-mile network of regional rail. This original plan included the North Central Corridor.58 Given revenue constraints, the system plan was reduced to 93 miles and later to 65 miles, but plans never excluded the North Central Corridor. As one DART planner described it, the North Central Corridor was expected to be such an outstanding corridor in terms of cost and ridership that it “might have been able to stand alone.”59 The cost of the corridor was anticipated to be relatively low for several reasons. In April 1988, DART purchased 34.5 miles of railroad right-of-way from the Southern Pacific Transportation Company, including rails paralleling the North Central Expressway.60 Additionally, the systems’ planners phased transit implementation. They determined that the line could be built in two waves, first to Arapaho Road in Richardson and later to Parker Road in Plano. (DART 1991) They also planned to implement a single track north of the Arapaho station until ridership demand dictated a second track. Finally, they constructed DART’s express bus facilities at proposed rail station sites so that some North Central Corridor infrastructure would be built in advance. 57 DART; “DART History”; http://www.dart.org/about/history.asp; Accessed 10/26/12 58 DART; “DART History”; http://www.dart.org/about/history.asp; Accessed 10/26/12 59 Interviewee AX, Telephone conversation, 7/24/12 60 DART; “DART History”; http://www.dart.org/about/history.asp; Accessed 10/26/12

I-11 In 1991, the Richardson Transit Center opened for bus park-and-ride and bus transfer operations.61 It would later become the Arapaho Center Station on the North Central Red Line. In 1992, the East Plano Transit Center opened just north of downtown Plano. It would later become Parker Road Station, the terminus station of the North Central Red Line. Reflecting the congestion- oriented nature of DART’s mission, DART’s bus facilities were also aligned with DART-funded high-occupancy vehicle (HOV) lanes on US-75 and other major Dallas freeways. By 1992, DART had broken ground on its light rail “starter system” and began the official federal planning process for the North Central Corridor.62 The starter system extended north from a downtown Dallas Transit Mall (a vehicle-free city street) to Park Lane (a station on the North Central Corridor and the initial station of the North Central Corridor extension). It continued south of downtown Dallas to Leddbetter Road on a South Oak Cliff alignment and to Westmoreland Road on a West Oak Cliff alignment. While the North Central Corridor extension was projected to be the best extension based on ridership, cost, and a comparison of benefit-cost ratios, it was decided that other lines would be built first.63 The decision to prioritize the southern routes was motivated by the transit-dependent population in the south Dallas area. As DART began the official federal planning process for the North Central Corridor, there was little debate over the preferred alignment or station locations.64 At various points, alternatives under consideration included a no-build alternative, an HOV expansion on the parallel freeway, an HOV facility within the rail right-of-way, and a shared rail and HOV facility in the rail right-of-way. However, with a former freight rail right-of-way already purchased by the agency, most alternatives were put forward as straw men, and the light rail facility was the clear intent of DART’s board and staff.65 In 1996, the first 11 miles of the 20-mile starter system opened for revenue operations, and the following year service was initiated at the Park Lane Station—the starting point of the North Central Corridor extension.66 At the time, the DART Board voted to accelerate light rail construction to the member cities of Garland, Richardson and Plano, including the installation of double-track north of Arapaho Station on the North Central Corridor.67 This decision eliminated two cost-saving measures—staged implementation of the full line and single track on the northern segment. Construction on the North Central Corridor north of Park Lane began in February 1999, just before DART signed a Full-Funding Grant Agreement with the Federal Transit Administration in October 1999. (DART 2006)68 The North Central Corridor extension opened in two phases: the first nine miles to Galatyn Park in July 2002, and the remaining three miles to Parker Road in December of that year.69 61 DART; “DART History”; http://www.dart.org/about/history.asp; Accessed 10/26/12 62 DART; “DART History”; http://www.dart.org/about/history.asp; Accessed 10/26/12 63 Interviewee AX, Telephone conversation, 7/24/12 64 Interviewee AJ, in-person conversation, 6/5/12 65 Interviewee AY; Telephone conversation, 9/7/12 66 DART; “DART History”; http://www.dart.org/about/history.asp; Accessed 10/26/12 67 DART; “DART History”; http://www.dart.org/about/history.asp; Accessed 10/26/12 68 Federal Transit Administration; http://www.fta.dot.gov/printer_friendly/12304_3053.html; Accessed 10/26/12 69 DART; “DART History”; http://www.dart.org/about/history.asp; Accessed 10/26/12

I-12 Figure I-2: Route Diagram for DART North Central Corridor, Dallas, Texas I.2.2 North Central Operations The North Central Corridor project is an extension to the DART Starter System. It connects to the system at Park Lane Station, two stations north of Mockingbird Station where the Blue Line splits with the Red Line to head northeast to Garland, TX. South of Mockingbird, rails continue into the downtown Dallas Transit Mall, where they also merge with Green Line operations. All DART LRT lines serve the Dallas Transit Mall. After passing through the Transit Mall, Orange Line and Green Line trains turn northwest, while the Red Line trains (with Blue Line service) continue southeast before turning south. A few miles south of downtown Dallas, the Blue Line service continues southward while Red Line rails turn southwest to Westmoreland Station near Interstate 20 in southwest Dallas. Upon opening in 2002, Red Line service on the North Central Corridor operated from Arapaho Center Station in the northern Dallas suburbs to Westmoreland in the southeastern Dallas suburbs.70 In late 2002, trains began to serve the far north-Dallas suburbs from the Parker Road Station. In 2009, with the opening of the first Green Line stations to the northwest of downtown Dallas, Orange Line service began to share Red Line tracks along the North Central Corridor but split off from the Red Line after the Dallas Transit Mall using Green Line tracks heading northwest. As of October 2012, Red Line trains operate at 15-minute headways during morning and evening peaks, at 20-minutes headways throughout midday, and at 20-minute to 30-minute headways in the early morning and at night.71 Orange Line trains operate all the way to Parker Road Station at approximately 15-minute headways from around 5:30AM to 8:00AM and from 3:30PM to 7:00PM. In the midday and at night, Orange Line trains operate as far north as LBJ/Central Station at 70 DART; “DART History”; http://www.dart.org/about/history.asp; Accessed 10/26/12 71 DART; http://www.dart.org/schedules/schedules.asp; Accessed 10/26/12

I-13 approximately 20-minute headways. During the peak, the North Central Corridor service between downtown Dallas and Plano operates at an effective headway of 7.5 minutes due to the overlapping Orange Line and Red Line services. As of 1996, DART estimated that over 11,000 daily riders would use the North Central Corridor extension in the year 2010. (FTA 1996) With double tracking and updated modeling, estimates were expanded to 17,000 riders in the year 2010. (FTA 1998) As of 2010, approximately 11,000 weekday riders boarded at the nine new stations of the North Central Corridor extension. As of 2010, approximately 18% of total LRT system ridership boarded along the North Central Line (which represented 23% of the system’s stations). (DART 2012) The difference in riders has been primarily attributed to DART service levels though a number of confounding issues also affected patronage. (DART 2006) As the Federal Transit Administration summarized: “The total efficiency and effectiveness of the LRT system is impacted by numerous factors outside the control of DART, such as economic conditions and developments and construction near station sites. However, factors controlled by DART, such as parking availability, station locations and bus system interactions, also affect ridership noticeably.” (DART 2006) Due to the agency’s budget constraints, DART reduced rail service frequencies from those present on opening day, which had already been reduced from planned frequencies. (DART 2006) DART also reduced bus services in the region through route consolidation and changes to service frequency. In addition, reduced employment in the Telecom Corridor impacted commute travel in the area. Only three stations have exceeded ridership forecasts. The three northernmost stations along the extension were anticipated to have approximately 4,000 riders, but their ridership has approached 5,000. (DART 2012, DART 2006) Two of the stations that exceeded ridership expectations (Parker Road and Bush Turnpike) serve as large park-and-ride facilities adjacent to major Interstate interchanges. Parking demand at these facilities led to the construction of additional parking stalls. The third station that has exceeded ridership forecasts is the downtown Plano station that experienced significant real estate investment in the walk-shed over the last decade. It was built in anticipation of 450 daily users while approximately 600 boarded at the station on weekdays in 2010. I.2.3 Planning a Successful Transit Project The North Central Corridor was part of a much larger transit system envisioned by NCTCOG, UMTA, and DART. UMTA’s funding criteria put significant emphasis on executing low-cost, high- ridership projects, which led to the prioritization of the North Central Corridor as an early expansion of the core system that passed through downtown Dallas. (DART, 1991) Expectations of a low-cost alignment were driven by the fact that right-of-way had been procured from Southern Pacific in the 1980s.72 Dallas planners had sought out continuous corridors where rail transit infrastructure could be accommodated. That included highway corridors, railroad corridors, and even electricity transmission line corridors. The availability of the freight rail corridor parallel to the North Central Expressway was considered unusual and fortunate. 72 Interviewee AY, Telephone conversation, 9/7/12

I-14 The freight rail corridor paralleled U.S. Highway 75, which was high volume and over capacity. The route also ran north though the City of Dallas to Richardson and Plano, two of the fastest growing suburbs in the 1980s.73 The roadway itself was slated for reconstruction and expansion due to the high demand. At one point, planners imagined the rail could be built quickly enough to serve frustrated drivers during the reconstruction slated for much of the 1990s. Because automobile congestion was a major impetus for the transit project, planners focused on providing a competitive alternative in terms of travel time.74 While early planning studies determined most riders would be transit-dependent, planners focused on attracting choice ridership. Therefore, travel time savings became a primary consideration and roadway levels of service were a primary indicator of where transit could have a competitive advantage. Average operating speeds had been an issue on earlier system investments that ran in arterial roadway medians and interacted heavily with traffic at intersections. This motivated planners to consider a multi-mile tunnel from downtown Dallas to the area near present day Mockingbird Station and to elevate tracks over roadways as often as possible throughout the Dallas and Richardson portions of the alignment. While the right-of-way had been procured, tunnels and elevated tracks added to the cost of the line and had to be justified to the Federal Transit Administration—a major project funder—on the basis that these features provided travel time savings that would attract significantly more riders to the line.75 Planners “knew it would be successful from a ridership standpoint” because it would connect fast-growing suburbs with downtown Dallas along an existing congested traffic corridor, and regional travel models were used to validate this assertion.76 Employment growth along the North Central Corridor was another justification for the route.77 There was a dense employment node at the Presbyterian Hospital campus near the proposed Walnut Hill Station and additional technology job centers near several stations north of Walnut Hill. While the route was typically considered a downtown commuter line, it was politically important for the DART Board to provide service to transit-dependent populations and minority populations. While there were some Asian American enclaves along the North Central Corridor in Richardson, it was mainly argued that the Starter System would provide access from predominantly African American communities near south Dallas station areas to jobs in downtown Dallas, and the North Central Corridor would later provide access from south Dallas to jobs in the emerging Telecom Corridor north of Dallas. Connections to job opportunities along the extension were touted frequently. For instance, a planned station at Campbell Road was deferred until anticipated demand finally warranted a station nearby. Interest from the City of Richardson to build a station near some proposed office developments motivated DART to eliminate the Campbell Road Station from plans and construct Galatyn Park Station just to the north of the planned Campbell Road site.78 This was considered proof of the economic development potential of the proposed line and became part of the rhetoric that the extension would not merely serve Plano and Richardson commuters bound for downtown Dallas. Yet, Texas Instruments, one of the largest employers in North Texas, declined to put a station near its headquarters and several of their major manufacturing facilities that were located adjacent 73 Interviewee AX, Telephone conversation, 7/24/12 74 Interviewee AY, Telephone conversation, 9/7/12 75 Interviewee AJ, in-person conversation, 6/5/12 76 Interviewee AJ, in-person conversation, 6/5/12 77 Interviewee AX, Telephone conversation, 7/24/12 78 Interviewee AJ, in-person conversation, 6/5/12

I-15 to the rail corridor.79 Transit planners did not consider wading into the politics of advocating for station locations near Texas Instruments largely because their campuses were auto-oriented and were considered unlikely to generate ridership. Though DART planners did not explicitly measure the density of the projects at the time, they tacitly considered campus-style technology offices a low-density use that—per their understanding of research on the topic of density and rail transit— would not support light rail service.80 In spite of arguments to the contrary, the line was considered a park-and-ride-accessed facility for downtown office workers and only modestly as a service upgrade for transit-dependent bus riders.81 Station locations were set in the very earliest plans based largely on where the line intersected with major East-West arterial roadways and where physical geometries allowed for long linear station platforms.82 Such locations were typically strip commercial or light industrial uses that had co- existed with freight rail operations. While this meant that there was some employment within walking distance of stations, evaluations found that there was very little housing near the line and it was expected that many nearby residents would opt to drive a short distance to the station.83 In fact, after initial station sites were selected, DART conducted accessibility studies that considered park- and-ride and bus feeder access. Walk-up potential and transit-oriented development potential were only marginally considered and were, for the most part, not influential in the design of facilities. In one exceptional instance, the City of Plano worked with DART to build a station near downtown Plano to help spur economic development.84 The city helped develop a large apartment complex adjacent to the line, improved streetscapes and parks near the station, and promoted the rejuvenation of their downtown commercial storefronts. However, in most instances, plans developed in final engineering focused on auto-centric priorities and were only modified in instances where parking capacity was added. For instance, the project’s surplus capital funds were used to add parking at Walnut Hill Station.85 Also, DART later supplemented parking at the Bush Turnpike and Parker Road stations to accommodate demand. Aside from the minor changes enumerated above, there were few modifications to the initial project proposal. In addition, there were no legitimate alternatives ever considered for implementation.86 As one planner stated, “The line was fixed in space [based on the fright rail right- of-way] so it was a matter of making tweaks to maximize ridership.”87 In spite of ridership that has underperformed relative to projections, the line is today considered a great success.88 For one, planners, politicians, and others believe it has mitigated some level of highway congestion and, thus, helped the region avoid detrimental impacts to its economic development. In general, rail transit is considered a significant regional economic advantage. To regional planners, DART makes the region more marketable as it competes for jobs and growth in a global 79 Interviewee AJ, in-person conversation, 6/5/12 80 Interviewee BA, in-person conversation, 8/14/12 81 Interviewee AX, Telephone conversation, 7/24/12 82 Interviewee AF, in-person conversation, 8/13/12 83 Interviewee AF, in-person conversation, 8/13/12 84 Interviewee AX, Telephone conversation, 7/24/12 85 Interviewee AX, Telephone conversation, 7/24/12 86 Interviewee AZ, in-person conversation, 8/14/12 87 Interviewee AJ, in-person conversation, 6/5/12 88 Interviewee AZ, in-person conversation, 8/14/12

I-16 marketplace.89 Rail transit is also considered a prerequisite for being classified as a global city and the Dallas region has focused on its ability to build DART’s light rail infrastructure when other Texas cities have failed to do so.90 Irrespective of ridership, Texas political leaders proudly focus on the fact that DART operates the longest light rail system in the country.91 Also irrespective of performance, DART and its regional partners are proud to be in the process of connecting their system to both of the region’s major passenger airports.92 It would seem that in the eyes of many Dallas stakeholders the most important measure of success for the North Central Corridor extension—and any other DART projects—is that rail transit was ever built in unabashedly automobile-centric Dallas, Texas. I.2.4 Commuter Rail Insights – Trinity Railway Express The Dallas-Fort Worth region is also home to the successful Trinity Railway Express (TRE) commuter rail service between downtown Dallas and downtown Fort Worth. We asked interviewees about their planning of commuter rail service, the differences they see between commuter rail and other fixed-guideway services, and the applicability of our indicator-based method to such transit proposals. The 35-mile, 10-station TRE project opened in three phases between 1996 and 2001.93 Right-of- way for the line was procured from the Chicago Rock Island and Pacific Railroad in 1983 by the cities of Dallas and Fort Worth at the same time as railroad right-of-way was procured for the Dallas North Central light rail alignment. The TRE rail is now jointly owned by DART and the Fort Worth transit agency, The T. The project was planned by North Central Texas Council of Governments, built by DART and The T, and is currently operated by a private vendor. Project funding came from the region and the Federal Transit Administration. TRE service connects downtown Dallas and downtown Fort Worth with stations in between. Most stations have free park-and-ride facilities with bus transfer centers.94 The route generally parallels Texas Highway 183, the DFW Airport Freeway. Fares are zone-based. Trains operate from 5:00AM to 11:00PM on weekdays and 9:00AM to 11:00PM on Saturdays with additional trains for special events. Headways are as frequent as 20 minutes at the peak of the AM peak and as infrequent as two hours in late evenings and on weekends. Freight railroads continue to use the tracks during off-peak times.95 North Central Texas Council of Governments planners characterized the project as an “opportunistic” rail transit investment.96 The right-of-way was purchased well in advance of the service being planned and at significantly less than it would cost to procure right-of-way today. The service was started because upgrade costs were minimal and operations costs were also marginal. The initial project was not expected to generate significant ridership.97 The more costly extension from the outskirts of Dallas to downtown Fort Worth was justified by the incremental ridership that 89 Interviewee AO, in-person conversation, 8/14/12 90 Interviewee AY, Telephone conversation, 9/7/12 91 Interviewee AJ, in-person conversation, 6/5/12 92 Interviewee AF, in-person conversation, 8/13/12 93 DART; www.dart.org; “TRE Facts”; Accessed 10/19/12 94 Trinity Railway Express; http://www.trinityrailwayexpress.org; Accessed 10/19/12 95 DART; www.dart.org; “TRE Facts”; Accessed 10/19/12 96 Interviewee AZ, in-person conversation, 8/14/12 97 Interviewee AO, in-person conversation, 8/14/12

I-17 the longer, two-ended route could generate. The project benefited from having major downtown business districts at both ends of the line because bi-directional traffic could be generated. The line also serves a hospital complex outside of downtown Dallas that is a major regional traffic generator. While ridership has been substantial and has justified service expansion, regional planners believe service frequency hampers attracting more ridership and attracting TOD investment.98 They hope to transition the service from locomotive push-pull train sets to lighter weight, faster, and more efficient DMU trains like those that now operate on the region’s new northern commuter rail service. They believe this technological change will allow for service changes that can make the service significantly more attractive to riders and to real estate investors. Planners felt that our indicator-based method could be adapted for use on commuter rail projects if it was sensitive to service frequency and the varied peak and off-peak schedules common to commuter services.99 In their own practice, NCTCOG planners have never formally added commuter rail service to their regional model. Instead, their models consider TRE to be an oversized commuter bus service. While they believed the indicator-based method would be useful in other circumstances, they were not sure that it would have been employed to evaluate TRE. They felt this particular commuter rail project was so obvious based on much simpler indicators— providing connections between two downtowns on an inexpensive freight rail right-of-way—that modeling was not required to justify the initial right-of-way purchase or the station locations. More sophisticated and FTA-approved regional travel models were used to apply for federal funding and there were few instances during that process when an intermediate ridership prediction tool would have been valuable. 98 Interviewee AO, in-person conversation, 8/14/12 99 Interviewee BB, in-person conversation, 8/14/12

I-18 I.3 EmX Phase I Bus Rapid Transit (Lane County, Oregon) The EmX Phase I project is a four-mile, 10-station bus rapid transit (BRT) facility connecting downtown Eugene and downtown Springfield in Lane County, OR. As the transit connection between the region’s two major transit hubs, the $25 million investment is considered the backbone of a proposed 61-mile regional BRT system. In operation since January 2007, the line has exceeded ridership forecasts from opening day. This case study suggests that many of the same rail transit planning “rules of thumb” identified in other TCRP H-42 case studies are also relevant for fixed- guideway bus projects. I.3.1 Establishing Lane County Bus Rapid Transit The EmX concept arose out of comprehensive transportation and land use planning conducted in Lane County, OR.100 In the 1970s, the Lane County metropolitan area was required by Oregon state law to develop comprehensive regional land use plans and urban growth boundaries. Throughout the 1970s and 1980s, the regional government discussed focusing urban growth in walkable, mixed- use development nodes. Because the designated nodes were typically existing commercial crossroads, these regionally significant locations were already served by Lane Transit District (LTD) bus routes. The EmX enhanced bus service arose from environmental interests.101 In response to the national 1992 Clean Air Act Amendment, the State of Oregon enacted Transportation Planning Rule Goal 12, which required Oregon cities to gradually reduce vehicle-miles traveled (VMT). In 1996, as part of a Regional Transportation Plan update, dedicated guideway bus service was promoted as an option that could make bus service more attractive than automobile travel and therefore achieve VMT reduction goals. Although many citizens encouraged regional planners to implement light rail, an urban rail feasibility study in the late 1990s deemed the mode to be out of scale with the land use density of the region and with LTD’s expansive service area, in addition to being too expensive to construct.102 By 2001, an extensive network of bus-only transitways connecting many of the 100 Interviewee BC, in-person conversation, 8/08/12 101 Interviewee BC, in-person conversation, 8/08/12 102 Interviewee BC, in-person conversation, 8/08/12 Figure I-3: Route Diagram for EmX in Eugene and Springfield, Oregon

I-19 region’s designated growth nodes was adopted in the long-range transport plan by Eugene, Springfield, Lane County, and LTD. The 2011 regional transport plan includes 61 miles of BRT in its fiscally constrained long-range project list (Central Lane MPO 2011). Opened in 2007, the Phase I EmX line from downtown Eugene to downtown Springfield was the first BRT route implemented in the region. EmX service was subsequently extended with a route that departs from the easternmost end of the Phase I project and proceeds in a loop through north Springfield. Recently, plans were approved for an additional extension that will extend EmX service from the westernmost end of the Phase I alignment at the Eugene Station through west Eugene. Though alternate versions of the Phase I alignment were considered, interviewees consistently agreed that the Phase I route was the consensus favorite among starter line options and provided the best opportunity to showcase the new technology and advance the vision of the 2001 regional transportation policy. Construction on the Phase I EmX route began in 2004 and was budgeted to be approximately $25 million.103 Funding came from Federal Transit Administration Section 5307 and 5309 funding sources ($19.2 million) and LTD transit funds. The project was completed under budget and on time with service initiated in January 2007. I.3.2 EmX Phase I Operations The Phase I alignment runs from the Eugene Station bus terminal to the Springfield Station bus terminal. Most bus service in the eastern part of the urbanized area passes through the Springfield Station’s eight bus bays, while approximately 30 bus routes pass through the Eugene Station’s 20 bus bay facilities. Sixty percent of the Phase I corridor consists of exclusive bus lanes. 104 Also, queue jumping lanes exist at the McVay Station, and signal priority exists at other locations. A 1.5-mile portion of the alignment passing along Franklin Boulevard through the Glenwood area was not constructed with exclusive right-of-way due to extremely low intersection density and because that portion of the corridor is slated for major roadway improvements in coming years, which could include EmX upgrades if ridership demand necessitates it.105 The EmX operation utilizes distinct buses from the rest of the LTD fleet.106 The hybrid-electric buses are 60-foot articulated New Flyer buses with three doors on the right side and two doors on the left side of the bus. They have low floors that allow level boarding from roadside platforms. The buses accommodate multiple wheelchairs and multiple bicycles. EmX runs approximately every 10 minutes on weekdays, every 15 minutes during weekday evenings and on Saturdays, and every 30 minutes during late evenings and on Sundays.107 Ridership on Route 11, EmX’s predecessor, averaged just under 2,700 weekday boardings.108 The LTD predicted EmX ridership would average 4,200 weekday riders (FTA 2009). Initial 103 Lane Transit District; “EmX History”; http://www.ltd.org/; Accessed 7/20/12 104 Lane Transit District; “EmX History”; http://www.ltd.org/; Accessed 7/20/12 105 Interviewee BD, in-person conversation, 8/08/12 106 Lane Transit District; “EmX”; http://www.ltd.org/; Accessed 7/20/12 107 Lane Transit District; “EmX”; http://www.ltd.org/; Accessed 7/20/12 108 Interviewee BD, in-person conversation, 8/08/12

I-20 observations greatly surpassed this prediction, with ridership hitting 6,600 per weekday by October of 2008. That year EmX reached a single-day ridership record of over 8,000 riders.109 Ridership declined by approximately 10% after proof of payment was required on the EmX route in September 2009, but it has since recovered.110 Service was provided free of charge until an off- board payment technology could be implemented (a larger order of fare collection machines was planned as part of the expansion of EmX services into north Springfield with the Gateway extension project).111 Prior to the implementation of fare collection, onboard surveys found that fewer than one-quarter of all riders on the Phase I alignment would require cash payments. The vast majority of riders possessed Lane Transit District group passes, typically because riders were University of Oregon students or staff, middle and high school students in the public school district, or employees of the regional hospital—all transit pass participants. I.3.3 Planning a Successful Transit Project The planning of the Phase I EmX was a collaborative effort spearheaded by an LTD board member. Key participating agencies included Lane Council of Governments, Oregon DOT, Lane Transit District, the cities of Eugene and Springfield, Lane County, the Federal Transit Administration, and the Federal Highway Administration. In the mid-1990s Rob Bennett, a former city councilman, an LTD board member, and an MPO policy committee member, responded to the statewide VMT reduction mandate with a visionary challenge.112 He applied his business operations experience, particularly marketing, to the discussion of transportation mode choice. He asked transportation staff at the MPO and LTD to produce a “quantum leap forward” in transit service. With his challenge in mind, the region set out to reduce VMT by making transit more attractive than driving. The LTD board identified several key components that would make transit competitive with automobile travel.113 Paramount was dedicated guideway to remove transit vehicles from congested roadways. Secondary were distinct vehicles that provided an enhanced in-vehicle experience and off-board fare payment that took the hassle out of boarding transit. Lastly, the service would need to be provided along major corridors to coordinate the transport investment with the region’s nodal land use aspirations. According to the personal experiences of most Oregonians, these features were readily available with light rail technology like that found in Portland, OR. Many public commenters argued on the basis of regional pride and the availability of funding at the state and federal level that Eugene “deserved” light rail transit.114 Cost was at the center of a debate between rail fans and those who believed enhanced bus service could achieve the same benefits.115 Bus detractors focused on bus service’s perceived unreliability, environmental impacts, and impermanent infrastructure. Bus advocates set out to alleviate concerns about impermanence with unique station designs, addressed environmental impacts by specifying hybrid bus technology, and sought to eliminate reliability issues with dedicated guideways, signal preemption technology, and digital wait-time indicators. In addition, advocates promoted the idea 109 Lane Transit District; “History”; http://www.ltd.org/; Accessed 7/20/12 110 Interviewee BE, in-person conversation, 8/06/12 111 Interviewee BD, in-person conversation, 8/08/12 112 Interviewee BE, in-person conversation, 8/06/12 113 Interviewee BE, in-person conversation, 8/06/12 114 Interviewee BC, in-person conversation, 8/08/12 115 Interviewee BC, in-person conversation, 8/08/12

I-21 that any enhanced bus corridor would be ready for conversion to light rail if demand supported the conversion. Advocates of bus focused their arguments against light rail on total cost and investment efficiency. While it was clear that rail would be considerably more expensive than enhanced bus services, it was arguments about the densities required to support rail service that became the lynchpin of their pro-bus advocacy. Consults were retained to conduct a feasibility study that rested on the notion that efficient transit services of various technologies required commensurate population densities to support them. Ultimately, the decision to pursue BRT was a contingent one.116 To address the uncertainty of an untested, unfamiliar technology, a Eugene City Councilman established policy laying out the conditions with which LTD would need to comply to gain the city’s support for the regional BRT plan: demonstrate that local governments unanimously supported the final design, that funding was available, and that outputs from the regional model indicated the proposed project would increase the transit mode share along a corridor. With these assurances, the regional BRT plan was adopted and an initial project was identified. Very little analysis was conducted to select the first project.117 LTD decided to replace Route 11 because it was the agency’s highest ridership route.118 The original concept for the Route 11 BRT upgrade was an 11-mile corridor from east Springfield to west-central Eugene.119 In addition to high ridership, the portion of the route in the City of Eugene had a grass median along an arterial roadway—Franklin Boulevard—that could accommodate a two-lane busway. It served two existing hubs and was located next to the University of Oregon campus, which had plans to expand without adding parking supply and instead raising parking prices.120 The route was also considered politically feasible because of its regional scope; it would serve the two primary cities in Lane County.121 Additionally, many staff and board members advocated for the Route 11 alternative because “it made sense as a pilot case to form a basis for future expansion.”122 Upon receiving pushback from several city council members who believed LTD should instead prioritize the improvement of low ridership routes, LTD staff carried out a back-of-the-envelope evaluation of proposed BRT routes.123 The evaluation considered bus and car travel times between route ends, existing bus ridership on the proposed routes, and the ease of implementing dedicated bus lanes on the corridors. Without producing a prediction of ridership based on the region’s travel model or another method, these success criteria validated the prioritization of the Route 11 upgrade. Despite the cost-based arguments for BRT over light rail, the 11-mile EmX project proved difficult to fund.124 The unproven service and estimated cost levels were not compatible with many transit funding programs and guidelines. To comply with funding source requirements, LTD and regional planners eventually scaled the project down to keep its costs in line with a particular Federal Transit Administration funding category.125 This, along with staff and local funding 116 Interviewee BC, in-person conversation, 8/08/12 117 Interviewee BE, in-person conversation, 8/06/12 118 Lane Transit District; “EmX History”; http://www.ltd.org/; Accessed 7/20/12 119 Interviewee BE, in-person conversation, 8/06/12 120 Interviewee BF, in-person conversation, 8/08/12 121 Interviewee BE, in-person conversation, 8/06/12 122 Interviewee BE, in-person conversation, 8/06/12 123 Interviewee BE, in-person conversation, 8/06/12 124 Interviewee BE, in-person conversation, 8/06/12 125 Interviewee BD, in-person conversation, 8/08/12

I-22 capacity constraints, ultimately led to the diminution of the project from the proposed 11 miles to the “backbone” four-mile segment between the two existing transit hubs. Once the four-mile alignment was selected, few alternatives were compared because it was an established route with existing bus stops located at critical intersections.126 The route was modified in only one instance, where Walnut Station was adjusted one block from the bus stop’s original location to be adjacent to a vacant auto dealership and a former Oregon Department of Transportation yard that were both slated for redevelopment. While farther from existing land uses, the revised station location also worked better from an engineering perspective. The EmX Phase I BRT project is considered a success because it has attracted new ridership with only modest investment and few changes to the Route 11 services. As one planner put it, “The [regional transportation] models couldn’t even handle the concept because the old service and the new service had 10-minute headways, but we knew that the old bus route was invisible to [citizens] and we were making [major service enhancements].”127 The notion was to simplify and re-brand bus transit to attract new riders. “[EmX] was point A to B, it looked different, no timetables necessary, no system maps needed.” Several interviewees attribute the doubling of ridership along the route to a profound change in transit service perceptions rather than operational improvements attributable to dedicated right-of-way. For politicians who approved the first EmX project, its success hinged on growing choice ridership and reducing VMT.128 It was generally believed by interviewees that EmX has successfully attracted those riders. As one regional planner argued during our interview, “LTD was running the Number 11 [bus] every ten minutes. Students had a class pass just as they do now. So, two of the factors that are key to attracting ridership were the same with the regular bus service that preceded BRT and the BRT that was implemented.”129 Therefore, planners believe that at least some of the new riders on EmX have shifted from another travel mode to the BRT service. In addition to benefitting from an increase in choice riders, LTD has seen bus operations improve. We argued [for BRT] on the basis of the quantum leap needed to attract choice riders and didn’t really put together a business case for why these operations would actually benefit our bottom line. In fact, the BRT service reduced [bus] travel times by 35%, increased corridor boardings by 270%, and reduced cost per boarding by 30% relative to the Number 11 service that operated on the corridor. BRT helped LTD overcome the biggest conundrum in bus operations. If you want to serve corridors where people want to access popular establishments or concentrations of dwellings, then one must operate a bus on a congested corridor. BRT provides, in particular, the transit infrastructure and service elements that overcome [that congestion].130 In retrospect, local planners believe operational benefits were a primary benefit, second to successfully attracting choice riders with a relatively inexpensive infrastructure investment. As one 126 Interviewee BE, in-person conversation, 8/06/12 127 Interviewee BE, in-person conversation, 8/06/12 128 Interviewee BG, in-person conversation, 8/08/12 129 Interviewee BH, in-person conversation, 8/08/12 130 Interviewee BC, in-person conversation, 8/08/12

I-23 planner put it, “Operations cost are more important locally than capital costs. Operating burden is local and capital cost burden is partially taken on by [the state and federal governments].”131 Another fundamental success of the project was its focus on accessibility.132 LTD was the first agency to provide universal wheelchair accessibility on all of it routes.133 EmX stations were designed with audible signals at busy intersections, platforms that provide level boarding, and buses equipped with rear facing wheel chair bays that allow for unassisted ingress and egress (LTD brochure Date Unknown). Ultimately, the Phase 1 EmX project is considered successful because of the political support it garnered for future investments. Because LTD services are funded by a payroll tax, “some of [our local business owners] pay close attention.”134 Many business owners have stepped forward to defend EmX expansion when other local business owners have complained about new services impacting their street frontages. It would seem that the stature of LTD has been elevated. Just prior to our case study visit, representatives from New Zealand had visited Eugene to tour the EmX facilities. A map on the wall indicated over 100 visits by delegations from all over the globe. EmX is a point of political pride in Lane County, Oregon. “We have the mayors of Eugene and Springfield talking about EmX every opportunity they get.”135 In spite of significant tea party opposition to the latest 8.8-mile extension of EmX into west Eugene, a third phase of EmX development, the city council affirmed implementation in a 7-1 vote on September 26, 2012, and the LTD board of directors voted 5-1 on October 8, 2012, in favor of initiating design of the new corridor.136 131 Interviewee BE, in-person conversation, 8/06/12 132 Interviewee BC, in-person conversation, 8/08/12 133 Interviewee BD, in-person conversation, 8/08/12 134 Interviewee BC, in-person conversation, 8/08/12 135 Interviewee BI, in-person conversation, 8/08/12 136 Lane Transit District; “West Eugene EmX Extension;” www.ltd.org; Accessed 10/29/12

I-24 I.4 Interstate MAX (Portland, OR) The Portland, Oregon, region, the 23rd most populous metropolitan area in the United States, operates a 50-mile MAX light rail system that was envied by several of our interviewees from much larger regions. The Interstate MAX project, a 5.8-mile extension of Portland’s system, was completed in 2004. The line connects downtown Portland to its northern suburbs in the state of Oregon and was designed with an intent to eventually extend the line further north, over the Columbia River, to Vancouver, WA. This Interstate MAX case study suggests that definitions of “success” are fungible and that qualitative factors may outweigh quantitative evaluation metrics. Figure I-4: Route Map for Interstate MAX, Portland, Oregon I.4.1 Expanding Portland Rail Transit The Interstate MAX project was part of a longstanding plan to expand Portland’s MAX light rail system. The long-range plans for the Portland region’s light rail system are based largely on Metro’s 1982 Light Rail System Plan, which identified bus routes with high enough ridership to justify conversion to higher-capacity transit.137 Construction of Eastside MAX (part of today’s Blue Line) commenced in 1982, and the line from downtown Portland to Greshman was opened in September 1986.138 Building on that line’s success, voters approved the Westside MAX (also part of today’s Blue Line service) to Beaverton and Hillsboro in 1990. After the completion of the EIS for TriMet’s Westside MAX line in 1994, regional transit planning focus shifted to the region’s next priority, the South/North Transit Corridor, which stretched from the southern suburb of Milwaukie in Clackamas County through Portland and across the Columbia River into Vancouver, WA. The Interstate MAX alignment that was ultimately constructed had been identified in the plan as one of two “Northern Alternatives” connecting downtown Portland with Vancouver. 137 Interviewee AA, in-person conversation, 8/7/12 138 TriMet; www.TriMet.org; Accessed 10/2/12

I-25 The FTA had already approved Metro’s request to undertake alternatives analysis on the South/North Corridor in 1993, and light rail was selected as the locally preferred alternative in December of 1994 (Metro 1998). As one planner suggested, “The mode was pre-selected for us based on the mode chosen for the previous [MAX light rail] projects.”139 In contrast to the official framing of the federally-overseen alternatives analysis process, early system planning focused on selecting routes where light rail would succeed. Planners did not identify problematic travel corridors and then select the optimal transportation improvement. In fact, local jurisdictions were fighting to be the next city in the Portland region to get a light rail project and took whatever actions they had available to them to help justify light rail.140 In 1994, Portland area voters approved a bond measure to finance their portion of the South/North Light Rail Project. However, voters in Washington State voted against a bond measure that would have financed their portion of the South/North Transit Corridor. TriMet continued planning the Portland portion of the line, but the project failed to win support at the ballot in 1998. Portland region voters ultimately rejected a $475 million General Obligation bond measure that would have funded the project’s construction later that year. Though the regional bond failed, results showed that 54% of city of Portland voters and 55.1% of Portland residents within one-half-mile of the alignment north of downtown supported the bond measure. In March 1999, a group of local business and community leaders asked TriMet to investigate a scaled back alignment on the northern portion of the corridor, from the Rose Quarter to Expo Center (city of Portland 2001). TriMet, Metro, and the Portland City Council were able to complete and adopt a Final EIS and Conceptual Design Report for the Interstate MAX project later that year. The FTA and TriMet signed a full-funding agreement (FFGA) in September 2000.141 TriMet reports the total project cost as $350 million, of which nearly 74% ($257.5 million) was federally funded. The remainder of the project was paid for by the city of Portland, Metro, and TriMet. Construction started in November 2000, and lasted almost four years. Major features included the 4,000-foot-long Vanport Bridge, significant streetscape enhancements, including a tripling in the number of street trees along the corridor, and the relocation of a 37-foot-tall Paul Bunyan statue in the Kenton neighborhood.142 Interstate MAX opened on May 1, 2004, four months ahead of schedule. I.4.2 Interstate MAX Operations Today, Interstate MAX (Yellow Line) is a 5.8-mile, 10-station line from downtown Portland, through North Portland neighborhoods to the Expo Center, near the border with Washington State. The northern terminus was selected to enable future expansion across the Columbia River to Vancouver, WA. The southern end of the line initially tied into the original East-West downtown transit alignment on SW Morrison and Yamhill Streets, shared with the Red and Blue Lines. Currently the Yellow Line utilizes the revitalized North-South Portland Transit Mall to travel through downtown Portland to its current terminus at Portland State University. The Yellow Line runs seven days a week, from roughly 5AM to 1AM, with 15-minute headways. During early mornings, midday, and in the evening, service is slightly less frequent. The vast 139 Interviewee AA, telephone conversation, 7/11/12 140 Interviewee BJ, in-person conversation, 8/6/12 141 Interviewee AA, telephone conversation, 7/11/12 142 TriMet; “Interstate MAX: Yellow Line”; www.TriMet.org; Accessed 10/2/12

I-26 majority of trains operate the full length of the current line, from Portland State University on the south side of downtown to the Expo Center terminus in North Portland. TriMet’s planning model (run in 2000) forecast 13,900 Interstate MAX riders in 2005, and between 18,100 and 18,860 by 2020 (FTA 2007). Actual 2005 ridership was slightly lower than projected, at 11,830, but it has been growing steadily at a rate of about a thousand additional weekday riders per year. Given the ridership growth trends, the FTA expected the project to “easily achieve better than 80 percent of its predicted ridership by the forecast year(s), indicating a relatively reliable ridership forecast” (FTA 2007). Presently, the Portland-Milwaukie light rail line is being constructed from the current terminus of the Yellow Line south to inner Southeast Portland, Milwaukie, and Oak Grove in north Clackamas County.143 The route follows a southern portion of the original South/North Transit Corridor Project. I.4.3 Planning a Successful Transit Project The primary agencies involved in the planning of the Interstate MAX were Metro, Portland’s unique elected regional government; TriMet, the regional transit agency covering Multnomah, Clackamas and Washington counties; and the city of Portland. Metro is responsible for the planning of the region’s transportation system and publishes the Regional Transportation Plan (RTP), which includes a plan for capital investments in high-capacity transit corridors. For the Interstate MAX project, Metro and TriMet worked together, with Metro as lead agency, to prepare environmental documents and secure funding from the FTA.144 TriMet managed the project’s construction. The city of Portland’s Office of Transportation conducted an expansive community outreach effort, building local support for the line, soliciting feedback on design details, and ensuring minimal negative impacts on local businesses during construction.145 This came on the heels of another community planning effort called the Albina Community Plan that was considered by transit planners to encompass the land use and economic development goals of the neighborhoods around much of the Interstate MAX corridor. To understand the planning of Interstate MAX, one must understand the planning of its predecessor, the South/North Transit Corridor. This corridor had been identified in the region’s 1982 rail plan. The success of the overall transit plan was considered to hinge on connecting the Portland region’s major poles, particularly transit centers (transfer hubs) and concentrations of employment.146 As one transit planner explained, in most instances the transit alignments defined on early plans were based on professional intuition using aerial photographs and accreted knowledge of regional travel patterns.147 Reflecting this planning technique, north of downtown Portland, the proposed South/North light rail alignment exited downtown’s Transit Mall to pass through the Rose Quarter event district, served several hospital campuses, skirted one of the region’s remaining port and industrial districts, served the city of Portland’s Exposition Center near the Columbia River, and passed through downtown Vancouver, WA—a fast-growing, northern suburb of Portland. Early 143 TriMet; “Portland-Milwaukee: a vital transportation link”; www.TriMet.org; Accessed 10/2/12 144 Interviewee BK, in-person conversation, 8/7/12 145 Interviewee BJ, in-person conversation, 8/6/12 146 Interviewee AK, in-person conversation, 8/6/12 147 Interviewee BK, in-person conversation, 8/7/12

I-27 considerations for station locations focused on serving these centers and also aligning transfer points for bus patrons on cross-town routes along major East-West thoroughfares.148 Between downtown Portland and the Columbia River, both the South/North corridor and Interstate MAX projects were planned in two segments. One segment consisted of track from the Banfield project’s existing downtown rail right-of-way to the Kaiser Hospital campus just northwest of the Interstate 5/Interstate 405 interchange. A second segment consisted of the straight route north from Kaiser to the Expo Center on Portland’s northernmost border. Figure I-5: Segment 7: Steel Bridge to Kaiser (Metro 1995) Throughout the planning of the South/North corridor, two possible alignments were considered for the segment of the light rail from the Kaiser facility, north of downtown Portland, to the border with Vancouver, WA. These alignments were called “Interstate Avenue” and “I-5” because one alignment would run down the center line of Interstate Avenue for much of the way (a four-lane state route that served as the primary North-South traffic artery prior to the opening of Interstate 5) and the other route would parallel Interstate 5 in right-of-way along the west side of the Interstate. An equal number of stations were to be located along the two routes and the stations were proposed at the same cross streets along the route. 148 Interviewee AA, in-person conversation, 8/7/12

I-28 Figure I-6: Segment 8: Kaiser to Expo Center149 As of 1994 evaluations, comparative characteristics of the two alignments suggest that the I-5 alternative was to be cheaper, faster, have higher ridership, and have fewer nuisance impacts on the neighborhood than the Interstate Avenue alignment. Table I-1: Summary Characteristics of Proposed Alignments (PMG 1998) Characteristic Interstate Avenue I-5 Year of Expenditure Cost (millions) $1,199 $1,085 LRT Weekday Ridership from Oregon City to 179th 64,000 65,400 Total Weekday Corridor Transit Ridership 131,350 132,800 Effective LRT Operating Cost (millions) from Oregon City to 179th $18.14 $18.02 Cost-Effectiveness Ratio (lower is better) 8.36 7.94 Residential and Business Displacements (Interstate Avenue variations reflect different roadway designs to accommodate varied levels of automobile capacity) 40/65/120 70 According to planning documents, there were significantly more advantages related to the I-5 proposal when compared to the Interstate Avenue alternative (PMG 1998). Modeling of the I-5 proposal suggested that the project would yield higher transit system ridership as well as higher ridership on the route. Much of that ridership differential from the Interstate Avenue alignment was 149 Portland Metro; “Segments and Design Options”; Portland Metro; April 13, 1995; p. 53

I-29 related to the shorter travel time along the I-5 route (two minutes shorter) that would make the service more attractive to Clark County, WA, residents as a commute alternative to downtown Portland. The I-5 alignment was also expected to have lower capital and operating costs than the Interstate Avenue alignment. Thus, the I-5 alignment was preferable when benefit-cost was measured as capital cost per rider and operating cost per rider. Planners also thought the I-5 alignment would provide better access to the Portland Community College (PCC) campus on N.E. Killingsworth and neighborhoods east of Interstate 5 while providing excellent accessibility to the high-density development between Interstate Avenue and Interstate 5 that was identified during the city’s Albina Community Plan process. The I-5 alignment also would have had significantly fewer impacts on businesses and residents during construction. Operating noise impact would have also been minimal along the I-5 alignment because noise walls would have been installed along the route. The walls would have also provided sound protection from Interstate 5 traffic noise. According to the planning documents, the Interstate Avenue alignment had fewer advantages relative to the I-5 option. Interstate Avenue operations would have provided more rail visibility and more direct access to existing retail, commercial, and residential properties along Interstate Avenue and within the Kenton area. The alignment would have provided equal accessibility benefits for new dense developments considered within the Albina Community Plan while providing greater accessibility to residential areas west of Interstate Avenue. One Portland planner suggested that the neighborhood was primarily interested in achieving dense development along Interstate Avenue—neighborhood-serving retail per the Albina Community Plan—and thought the development potential would be maximized if the rail ran along the Interstate Avenue corridor.150 Transport planners were interested in the operational benefits of the I-5 line, which minimized grade crossings and maximized travel speeds. The Project Management Group suggested that a modified alternative be studied, one that merged the two concepts by utilizing the Interstate 5 right-of-way between stations and then diverting the line several blocks to accommodate station platforms on Interstate Avenue (PMG 1994). Planners considered several hybrid variations of the alignment.151 However, the operational benefits were considered much too small relative to the number of property impacts that would have occurred. The final recommendations for Draft Environmental Impact Statement (DEIS) alternatives focused on the Interstate Avenue and I-5 options and suggested that a tradeoff would exist between cost and enhancing certain “land use opportunities.” (PMG 1996) While not explicitly mentioned in any transit planning documents we reviewed, interviewees pointed to safety concerns as an issue that tipped the scales in favor of the Interstate Avenue alignment. In the case of the I-5 alternative, stations had to be set back from major cross streets to accommodate conflicts with Interstate 5 on-ramps and off-ramps. Planners believed that real and perceived lack of safety for patrons accessing stations and waiting on platforms near Interstate 5 could negatively impact ridership on that alignment and harm the MAX brand. As one transit planner described the process, “Part of the argument [for the Interstate Avenue alternative] was that the stations would be safer in the middle of Interstate Avenue where there are eyes on the streets, people passing by, grocery stores, restaurants, and bicyclists, as opposed to near the freeway where—because these [rail] cars are 190 feet long—part of the station is going to be isolated. How are you going to protect it?”152 As part of their practice, the planner asks, “How do you make 150 Interviewee BJ, in-person conversation, 8/6/12 151 Interviewee AA, in-person conversation, 8/7/12 152 Interviewee AK, in-person conversation, 8/6/12

I-30 it an attractive and a comfortable space for your daughter who is 16 years old?” While the region’s ridership models did not account for perceived safety, planners had learned from experience on the Banfield line, Portland’s first light rail line, that platform safety was a critical issue for light rail operations. By 1995, the Interstate Avenue and I-5 options were viewed largely through the lens of the revitalizing potential and perceived safety of the Interstate Avenue alignment, making it a clear preference among the community and regional planners.153 Attention began to focus on the second segment of the project, the approach into downtown Portland. Project alternatives showed the South/North project entering downtown from the north across Portland’s Steel Bridge, sharing tracks with the existing Gresham LRT line (PMG 1995). This accommodated a station in the existing transit center located between the Oregon Convention Center and the Rose Garden Arena on the east side of the Willamette River. Per Portland’s planning imperative to connect employment centers, four alternatives showed the line continuing from the transit center and passing along either the east or west side of Interstate 5 with stations at the Emanuel Hospital facility. The advantages of these alignments were the access provided to the Emanuel Hospital employment center and the Eliot neighborhood. In fact, an evaluation of advantages and disadvantages consistently described the number of employees and residential units accessible within a five and ten minute walk of the proposed stations (PMG 1995). Concerns regarding the alignments focused on operating issues as well as costly design components (PMG 1995). In at least one of the four options, the Emanuel Hospital station would have been a costly underground station. Other concerns about the alignments included operations impacts when passing through the Rose Quarter during events, the cost of passing either over or under Interstate 5 (in some instances, multiple times), and potential operating conflicts when running on non-exclusive right-of-way (i.e., in neighborhood streets). As of early 1996, two route options for the segment of the project from downtown Portland to the Kaiser facility had been identified for inclusion in a DEIS. Both options passed through the Rose Quarter transit center, had stations when crossing Broadway, and had stations adjacent to the Emanuel Hospital facilities (PMG 1996). Notably, no alternative alignment west of Interstate 5 was recommended for further analysis because such a route would not provide access to the Eliot neighborhood or Emanuel Hospital, which were considered priority service areas (access from a station west of Interstate 5 to the neighborhood would have necessitated crossing the Interstate and negotiating an 80-foot grade change). This alternative was not included for further analysis also because the station on the west side of Interstate 5 would have been located in a zone designated for continued urban industrial uses, and it was feared that a station would have produced “non- industrial redevelopment pressures which contradict city objectives for this area” (PMG 1996). Planning proceeded after 1996 toward another major funding milestone. Within the Portland metropolitan region, Oregon voters had approved funding for the South/North line in 1994 while Washington State voters had not. In 1998, Portland voters were asked to approve a local bond to finance the South/North project using the pre-approved revenues. Portland voters rejected the measure, though a majority of North Portland voters did vote for the bond. In response to the failed vote, regional elected officials held a series of “listening posts” to determine next steps. The community suggested moving forward with a shorter, less expensive project in North Portland where voters were supportive of the funding measure. Planning for the shorter route proceeded quickly based on prior planning conducted for the South/North project. 153 Interviewee BL, in-person conversation, 8/7/12

I-31 Business leaders formally requested a segment be built between downtown Portland and the Expo Center in March 1999 (City of Portland 1991). By April 1999, staff had prepared a Supplemental Draft Environmental Impact Statement (SDEIS). By June 1999, the Interstate Avenue alignment was identified as the preferred route in the North Corridor by the Portland City Council, TriMet, and Metro. Finally, in October 1999, Portland City Council adopted the Final Environmental Impact Statement (FEIS). The project that was ultimately approved in 1999 consisted of the Interstate Avenue alignment in the northern segment and, for the segment leaving downtown Portland, a route along the Willamette River, significantly west of the Interstate 5 corridor—a route not studied in the mid-1990s planning process but one subsequently considered because of the project’s limited budget. After leaving downtown Portland via the Steel Bridge, the line diverged from shared tracks at an intersection prior to the pre-existing transit center in the Rose Quarter. While all prior proposals had passed through that transit center, transferring patrons would walk as much as several hundred yards to reach certain bus bays. This change accommodated a sharp left turn to the northwest so that the alignment could follow an existing multilane arterial through an industrial zone—a much lower cost route than previously conceived. Unlike prior plans, no station was provided at Broadway and no access was provided to the Eliot neighborhood or Emanuel Hospital complex. Planners determined that the cost of providing those connections far outweighed the benefits of actually getting a project built within the limited budget.154 Part of the motivation for the original South/North alignment was to serve as a salve for community interests upset over prior government interventions.155 The Eliot neighborhood and Rose Quarter had been significantly impacted by urban renewal projects. The area’s neighborhoods, predominantly minority and lower income than much of Portland, had also been impacted by the construction of the Interstate 5 corridor. Relatively recent displacements for the construction of the convention and arena complexes were also fresh on the mind of community and local government officials. Even so, access to many of these communities was sacrificed for a lower cost route on the segment leaving downtown Portland. On the other hand, there was not an option for the segment from Kaiser to Expo Center that was magnitudes cheaper than other alternatives but there was one option that could serve as the salve for prior government interventions. As one planner stated, “The residents saw the value of transit and [attendant] reinvestment [to] recreate a neighborhood that was lost because of the freeway [construction].”156 Planners were persuaded to pursue the Interstate Avenue alignment over the I-5 alignment even though the I-5 alternative was superior by most quantitative metrics. Further cost-saving measures were also identified. One such tradeoff reduced auto-mobility and impacted transit operations to reduce costs while simultaneously meeting neighborhood preferences. Whereas the original Interstate Avenue route plans had assumed that as many as 125 businesses and residences would have to be displaced to accommodate road widening, transportation planners determined that the Interstate Avenue tracks could be built without significant changes to the existing road right-of-way. Taking lanes without replacing them reduced automobile throughput capacity but provided adequate capacity for near-term automobile demand. To address longer-term auto demand, some automobile turn movements were accommodated in lanes shared with light rail tracks. However, the interactions between automobiles and trains negatively impacted the proposed 154 Interviewee AK, in-person conversation, 8/6/12 155 Interviewee BJ, in-person conversation, 8/6/12 156 Interviewee BJ, in-person conversation, 8/6/12

I-32 transit operations along the Interstate Avenue alignment. That said, the changes allowed the transit project to be built where the community wanted it, without displacements, and within the available budget. In a retrospective evaluation of the project’s performance, it was found that planners overestimated the travel time impacts of operating light rail in city streets and underestimated the attractiveness of the service to non-commuters (FTA 2008). Additionally, FTA found that planners calibrated their ridership models with land use changes that did not materialize and used walk-up and park-and-ride ridership assumptions that were overly optimistic. Ultimately, the project was built on budget and attracted approximately the number of riders predicted during the planning phases of the project. All of our interviewees believed that the Interstate MAX project was a success. When asked what they might do differently, no one suggested that the less costly I-5 alignment would have been preferable. Some interviewees believed that more could have been done to capitalize on the project through proactive land use planning.157 Likewise, some suggested that even stronger community engagement would have been beneficial had more funding been available.158 Another thought it had been successful at attracting riders but not necessarily the choice riders that are highly prized by the regional agency.159 Despite the line’s minor shortcomings, it is widely believed that Interstate MAX has provided several years of travel benefits for citizens, generated significant community development benefits for the neighborhoods it currently serves, and preserved opportunities to expand the project as envisioned by the original South/North Corridor project. In fact, planning for an extension of the Interstate MAX line to Vancouver, WA, is ongoing, and an extension of the line to the south is under construction. I.4.4 Commuter Rail Insights – Westside Express The Portland region is also home to Westside Express Service (WES), a commuter rail project extending from Wilsonville in the southwest of the region to Beaverton in the central-west of the region. Similar to Dallas, we asked interviewees about their planning of commuter rail service, the differences they see between commuter rail and other fixed-guideway services, and the applicability of our indicator-based method to such transit proposals. The 14.7-mile, five-station WES project cost $161MM to build and opened in 2009.160 The project was implemented by TriMet on an operating freight railroad right-of-way in partnership with Washington County, Oregon Department of Transportation, Metro, and the cities of Wilsonville, Tualatin, Tigard, and Beaverton. TriMet and Washington County shared costs above base elements funded by FTA. The suburb to suburb line connects four communities in the southwest of the region to the MAX light rail system via an intermodal station in Beaverton on the Westside LRT line. The service generally parallels a North-South highway corridor consisting of Interstate 5 in the south and state highway 217 in the north. Service operates on 30-minute headways, Monday through Friday, during the morning and afternoon peak.161 157 Interviewee BL, in-person conversation, 8/7/12 158 Interviewee BJ, in-person conversation, 8/6/12 159 Interviewee AI, in-person conversation, 8/7/12 160 TriMet; http://trimet.org/about/history/wes.htm; Accessed 10/19/12 161 TriMet; http://trimet.org; Accessed 10/19/12

I-33 The project was envisioned and advocated by stakeholders in the western part of the Portland region.162 Due to low anticipated ridership relative to cost and alternative regional projects, the project was not initially supported by either the regional MPO or the rail transit agency.163 Before models were even run, Oregon Metro argued against the line because of the extremely low housing densities near the right-of-way. Regional funding equity drove the decision to move forward with planning and a downsizing of the project made it a justifiable investment. Ridership on the line has met projections made early in the planning process but is far short of the revised numbers that were eventually used to justify federal and regional funding.164 In spite of the heavy rail technology required by the Federal Railroad Administration on the alignment and limited operating schedule, neither Metro nor TriMet consider this a commuter rail project.165 According to Metro planners, the project is essentially a cost-effective LRT extension in a technologically constrained corridor.166 Regional planners suggested that the service has already spurred several transit-oriented real estate investments, akin those along MAX light rail lines, in spite of the current operating limitations. They hope to one day expand to all-day service and gradually invest in the corridor (e.g., double tracking) until it can be cost effectively transitioned to MAX LRT technology and provide a one-seat ride to downtown Portland on the existing Westside LRT corridor. Interviewees suggested that the project’s success has not been hampered by technology or the setting of the project but by the frequency of service and the limited hours.167,168 They believe that improved service could even justify a costly shift in alignment from the existing freight rails over limited distances to provide greater accessibility to certain land uses, particularly a mall that is a major regional trip generator. Interviewees indicated that they used many of the same rules of thumb for designing WES commuter rail and MAX light rail projects. In general, they believe our proposed indicator-based method could have been applied to WES if it took into account the reduced operating schedules that are typical of commuter rail. 162 Interviewee AA, in-person conversation, 8/7/12 163 Interviewee AI, in-person conversation, 8/7/12 164 Interviewee AM, in-person conversation, 8/7/12 165 Interviewee AA, in-person conversation, 8/7/12 166 Interviewee BL, in-person conversation, 8/7/12 167 Interviewee AI, in-person conversation, 8/7/12 168 Interviewee AB, in-person conversation, 8/7/12

I-34 I.5 University & Medical Center Extensions (Salt Lake City, UT) The 3.8-mile, seven-station extension of the Salt Lake City, Utah, TRAX light rail system connects the original starter line in downtown Salt Lake City to the University of Utah campus to the east. Though conceived of and planned as part of one larger project, the University and Medical Center extensions represent two subdivisions of that original line. The phased implementation ensured the first phase from downtown to the University of Utah campus, the university extension, was in service when the 2002 Winter Olympic opening ceremonies were held in the University of Utah football stadium. This case study offers insights into the myriad measures of success and indicators of success that can influence transit project proposals, particularly rules of thumb used to define course-grained transit system plans. Figure I-7: Route Diagram for Utah Transit Authority (UTA) University and Medical Center Extensions, Salt Lake City, Utah I.5.1 Expanding Salt Lake City Rail Transit Initial planning for a light rail system in the Salt Lake City area began in 1983 and, motivated as a mitigation measure for Interstate 15 expansion, the North-South Corridor was identified as the region’s initial rail project in 1988. Utah Transit Authority (UTA) utilized a federal grant to acquire and preserve right-of-way in the same time period (FTA 2007). The concept of the 15-mile North- South light rail project was included in a tax funding plan that failed when put before voters in 1992.169 In spite of the tax measure’s failure, planning for an East-West rail project between the Salt Lake City Airport and the University of Utah began in 1993 in anticipation of several major events (FTA 2007). After Salt Lake City won its 1995 bid to host the 2002 Winter Olympics, light rail planning was fast-tracked. A Major Investment Study conducted in 1996 identified a 10.11-mile light rail line from the Salt Lake City International Airport to aid with transportation during the 2002 Olympics. 169 Interviewee BM, in-person conversation, 8/20/12

I-35 Opened in December 1999, the 15-mile starter line from downtown Salt Lake City south to Sandy Civic Center was paid for largely from Interstate 15 reconstruction funds reallocated to the project soon after the Winter Olympic announcement.170 Ridership on the initial line quickly surpassed projections, and voters passed a quarter-cent sales tax to fund future transit expansion in November 2000, particularly the “West-East Line.” Due to federal funding limitations and time constraints related to the 2002 Winter Olympics, the West-East Line was divided into four separate segments in 1999: the Airport Extension, the Downtown Loop, the University Line, and the Medical Center Extension of the University Line (FTA 2007). In spite of dividing the line for funding purposes, by all other measures the lines were considered one project and there was little doubt that all four segments would eventually be constructed.171 In early 2000, the FTA approved the final design on the scaled back portion of the West-East alignment that extended from the existing North-South line in downtown to the University of Utah. In mid-2001, UTA received federal approval to begin final designs of the medical center extension as well. Construction commenced on the University Line in the spring of 2000 and a Full-Funding Grant Agreement was signed in August 2000 (FTA 2007). The relatively fast pace of approvals and construction were based on a desire to open the University Line in time for the 2002 Winter Olympics. Perhaps due to the expedited timeframe, preferred plans for the University and Medical Center projects changed little during the course of planning (Parsons 1999, Parsons 1997). The first phase of the extension, to the University of Utah football stadium (Stadium Station) was opened on December 15, 2001 (FTA 2007). Construction of the medical center extension followed immediately after construction was completed on the University Line, significantly before a Full- Funding Grant Agreement was signed in May 2002. The second phase, bringing the line to its current terminus at the University Medical Center, was opened on September 29, 2003, a year after the Olympic Games and a full 15 months ahead of schedule (Salt Lake Tribune 9/28/03). It is thought that the cost and schedule efficiencies were the product of the seamless construction process between the first and second lines (FTA 2007). UTA reports the cost of the extensions at $148.5 million and $89.4 million, respectively, and roughly 65% ($96.5 million) and 60% ($53.6 million) of the extensions were federally funded (FTA 2000, FTA 2002). The remainder of the projects were funded by local sales tax revenues. I.5.2 University/Medical Center Operations UTA’s 3.8-mile, two-part extension proceeds east from downtown Salt Lake City in the center lanes of E 400 S (also known as University Boulevard), briefly swinging south onto E 500 S before continuing east and entering the University of Utah campus adjacent to the university’s football stadium. The university/medical center extension proceeds east from the stadium and then north to the medical center campus. The first phase of the extension, to the university, included three new stations along 400 S and a station at the University of Utah football stadium. The second project extended the line a further three stations, all of which serve the University of Utah campus, to its current terminus at the University Medical Center. Under current service patterns, the University/Medical Center extension is part of the Red Line, which shares tracks with the Blue Line (North-South Line) south of downtown Salt Lake City until 170Utah Transit Authority; http://www.rideuta.com/uploads/FactSheet_History_2012new.pdf; accessed 10/22/12 171 Interviewee BN, in-person conversation, 8/20/12

I-36 branching to the west at Fashion Place West along the more recently opened Mid-Jordan extension to Daybreak Parkway. The Red Line runs seven days a week, from roughly 5AM to 12AM Monday through Saturday and from 9:30 a.m. to 11:00 p.m. on Sundays. Service is provided at 15-minute headways throughout the day on weekdays, and every 20 minutes on weekends. The vast majority of trains operate the full length of the line, with the exception of several of the earliest and latest trains. Prior to opening, the University Line corridor was served by multiple UTA bus routes (Chatman 2012). Bus service on the eastern portion of the University Line, along 400 South, was mostly replaced by light rail. Also, the bus routes in the downtown area were modified to facilitate better connectivity between light rail and bus. Significant bus service along 200 South, considered a less- congested and more bus-friendly route than 400 South, continues to operate between downtown and the campus.172 According to the TCRP H-42 transit project database, ridership at stations along the two project segments were approximately 7,300 and 3,400 per weekday. This ridership nearly meets forecasts estimated for 2020. In 2020, ridership was expected to be 7,600 weekday boardings on the downtown to university segment of the line, with 3,100 of those expected to be new transit riders (FTA 2000). Also, ridership on the medical center extension was predicted to be 4,100 on an average weekday, with 3,400 new riders (FTA 2002). I.5.3 Planning a Successful Transit Project As noted, planning for the University/Medical Center extension was initially included in a larger West-East Line from the Salt Lake City International Airport, due east to downtown, and then further east to the University of Utah campus. The Wasatch Front Regional Council (WFRC), working with UTA, completed environmental studies in 1997 and 1999 on the 10.9-mile West-East Corridor. These documents were required because the project sought funding from FTA, which also influenced the design of the line. A great deal of information about measures of success and indicators of success can be gleaned from planning documents for the West-East Line, the debates that occurred over the alignment, and the after-the-fact assessments of the line that interviewees shared during our conversations about the University and Medical Center extensions. During early transit system studies, regional planners hired consultants to identify routes that could be viable rail transit projects.173 The regional evaluation considered three factors sequentially: major regional destinations, origin and destination pairing between major destinations, and existing traffic congestion on corridors linking paired regional destinations. As a regional planner expressed during our interview, “[Automobile congestion] is the reason people will ride transit.” The West- East Corridor concept arose out of this form of high-level system planning analysis. Whereas the region’s original North-South alignment had been motivated largely by bus rationalization and availability of right-of-way, the West-East Corridor was identified because it met the primary criteria of the rail transit study because it served the region’s primary airport, downtown Salt Lake City, and the University of Utah and congestion between them would only grow worse.174 In fact, the university was the second biggest generator of traffic in the state, second 172 Interviewee AE, in-person conversation, 8/20/12 173 Interviewee BN, in-person conversation, 8/20/12 174 Interviewee BM, in-person conversation, 8/20/12

I-37 only to downtown Salt Lake City.175 Of particular interest to Utah planners in the early 1990s, the proposed line passed several Olympic venues located in downtown Salt Lake City and on the University of Utah campus, and 47% of Olympic lodging was located within the West-East Corridor. (Parsons 1999) In addition to those primary destinations, documents claimed that high levels of transit and travel demand existed because of special trip generators along the corridor including: the LDS Church’s downtown campus, Utah State Fairpark, Delta Center basketball arena, Salt Lake Arts Center, Abravanel Hall, Salt Palace Convention Center, Capitol Theater, John W. Gallivan Utah Center, Hansen Planetarium, Fine Arts Museum, Museum of Natural History, Pioneer Memorial Theater, Kingsbury Hall, Rice-Eccles Football Stadium, John M. Huntsman Center (Parsons 1999). Planning documents suggested that many of the trips generated by these uses occurred within the corridor as people moved from the airport, campus, and venues to hotels and restaurants also located along the route. At the behest of the FTA, several transit modes were considered for the service along this destination-rich corridor. Among them were standard bus service, bus lanes, and LRT (Parsons 1999). Bus lanes were motivated by FTA’s interest in pursuing bus rapid transit in the mid-1990s.176 While a feasible and cost-effective option, the region argued that BRT would eventually need to be upgraded to light rail in the corridor at much greater expense and with greater service impacts. Light rail was preferred because of its compatibility with the existing system and the area’s aesthetic, the perceived reliability improvements relative to bus service, and the role rail had played in defining the region’s long-term land use vision (Parsons 1997). Light rail was also considered superior at the time because of the region’s focus on air quality, an argument based on the electric motive power and rail’s ability to attract choice riders and reduce regional VMT (Parsons 1997). There was also a desire to provide a world-class urban transit connection to the University of Utah campus where many Olympic venues were located.177 The University of Utah was amenable to the light rail transit but argued that it would be best for it to remain on major roadways so that the center of campus could remain a pedestrian-oriented environment. Some planners argued that light rail could be integrated into the heart of the dense campus environment but it was resolved that shuttle services would help move people from rail to the various quadrants of the campus. Past the main campus area, on the easternmost end of the West-East Corridor, planners considered serving either the medical center to the northeast or a research park to the southeast. They determined that existing and future land uses favored the medical center alignment. As one planner explained, research parks have “long distances between the streets and buildings in a park- like setting: not real transit conditions.”178 Without conducting extensive analysis, the additional trip distance between transit stations and employment destinations in the research park were determined to be indicative of low ridership. Ultimately, it was determined that there was “a much bigger concentration of trips to the medical center.”179 Additionally, transit planners learned that master plans for the University of Utah campus, including the medical center, called for additional facilities within the existing footprint.180 175 Interviewee BN, in-person conversation, 8/20/12 176 Interviewee BN, in-person conversation, 8/20/12 177 Interviewee BN, in-person conversation, 8/20/12 178 Interviewee AE, in-person conversation, 8/20/12 179 Interviewee BN, in-person conversation, 8/20/12 180 Interviewee AE, in-person conversation, 8/20/12

I-38 Future land use changes were also a consideration when planners evaluated which alignment they would recommend for the connection between downtown and the University of Utah. Planners generally evaluated potential land use changes based on local land use policies, but did not attempt to quantify the scale of real estate development that might have occurred because of the transit project (PBQ&D 1994). Local government staff was adamant that the alignment be located on 400 South because they saw more development potential along that corridor than 200 South, an alternative route along a residential corridor that provided access to the University of Utah’s ceremonial campus entrance and was unlikely to experience land use changes.181 In fact, the city had previously promised the 200 South neighborhood that no density increases would be allowed. The impetus for the 200 South proposal had been automobile traffic priorities along 400 South. Early transit feasibility studies had identified the 400 South Corridor due to physical constraints on parallel routes and transit planners advocated for the route as planning progressed.182 The local government agreed to take over maintenance of the roadway, a state route, from Utah DOT to facilitate the implementation of light rail. However, pressure to relocate the project to 200 South or another 400 South alternative came from Utah DOT when their planning process for the Interstate 15 corridor identified 400 South as an interchange. Legislators and others forced a re-evaluation of the light rail route. Ultimately, a compromise solution was developed that retained capacity on the roadway by sharing left turn lanes with light rail tracks. While an operational setback, planners believe the corridor is preferable for a number of reasons. While planners evaluated a number of measures of success during the planning process, they have found the constructed alignment has been successful for a number of unanticipated reasons. For instance, the transit project was anticipated to influence real estate development on the corridor and a recent UTA study has identified $1 billion in private real estate investment along the corridor.183 As the UTA’s community outreach staff person has stated, “Rail is a big motivator for developers.”184 However, it was not anticipated that the rail would be so successful at allowing for significant public sector real estate investment on the University of Utah campus. Prior to the light rail line, the university had 10,000 occupied parking stalls. In recent years, even with the addition of more occupiable space on campus and more student enrollment, the university experiences demand for approximately 7,000 parking stalls.185 Development has occurred on several surface parking lots. With the shift in travel patterns the university is able to better utilize its limited land area while avoiding pushing parking demand into neighborhoods. Many donations received by the university will fund new structures but will not pay for parking.186 Overflow parking into nearby neighborhoods had historically been a major concern and it was difficult to build buildings using donations without identifying additional funding sources to build attendant parking facilities. However, the utilization of transit services by students and staff has allowed the university to grow without adding parking. In another instance, transit planners anticipated the traffic mitigation benefits of a light rail project but turned their attention to the traffic safety benefits of the transit project only during the later project design phases. West-East Corridor plans focused on the need for a transit alternative in the growing region as vehicle-miles traveled were anticipated to rise faster than population or 181 Interviewee BN, in-person conversation, 8/20/12 182 Interviewee BN, in-person conversation, 8/20/12 183 Interviewee AE, in-person conversation, 8/20/12 184 Interviewee BM, in-person conversation, 8/20/12 185 Interviewee BO, in-person conversation, 8/20/12 186 Interviewee AE, in-person conversation, 8/20/12

I-39 employment and roadway capacity would not keep pace (Parsons 1999). Yet, highly localized traffic safety benefits were produced as the light rail plan helped to address problem intersections and pedestrian safety issues. The implementation of a roundabout and significant pedestrian infrastructure provided a safer environment, particularly near the University of Utah.187 While it was anticipated by planners that connecting regional trip generators would be beneficial for transit, they did not anticipate the operational efficiencies that were gained by serving the University of Utah campus in particular. As one planner explained, students and staff generate significant midday ridership due both to the staggered class schedules of students and the opportunity for students and staff to reach lunch destinations and convenience retail just off campus.188 While adding significantly to the ridership on the line, this off-peak demand does not require UTA to add additional train cars or increase service frequency. By almost every measure, the University and Medical Center extensions of the TRAX light rail system were considered a success by interviewees. Success has been defined a number of ways, many of which were not stakeholder priorities during the planning process. This suggests that some of the criteria that informed the conception of these projects—perhaps the density of destinations and the significant barbell trip generators/attractors—may effectively address multiple measures of success simultaneously. The successes have furthered the region’s resolve to implement an extensive light rail system. The two extensions were originally envisioned to be part of a West-East Line from the airport to the university and the airport extension of that project is expected to open in 2013. Adding to the University and Medical Center projects, the airport connection will fulfill the complete vision of the late 1990s major investment studies (Salt Lake Tribune 5/1/12). I.5.4 Commuter Rail Insights – FrontRunner North The Salt Lake City region is also home to FrontRunner commuter rail service. Again, we asked interviewees about their planning of commuter rail service, the differences they see between commuter rail and other fixed-guideway services, and the applicability of our indicator-based method to such transit proposals. The 44-mile, nine-station FrontRunner North project extends from downtown Salt Lake City to Ogden in the north, cost $551 million to build, and was completed in 2008 (UTA 2005).189 An expansion of the service is now operational from downtown Salt Lake City to Provo in the south of the region. The project was planned and built cooperatively by regional bodies and UTA. Funding came from the region and FTA. The service connects northern cities and suburban communities to Salt Lake City along a route that parallels Interstate 15. Service operates from 5:00AM to 11:30AM Monday through Friday and 7:00AM to 11:30AM on Saturday.190 Trains operate every 30 minutes in the peak and every hour midday and evenings during the week, and every hour and a half on Saturdays except for special services provided for events. Fares vary by distance. UTA buses serve all stations and extensive park-and-ride facilities ranging from 235 stalls to 874 stalls are located at all stations outside of downtown Salt Lake City. 187 Interviewee BN, in-person conversation, 8/20/12 188 Interviewee BP, in-person conversation, 8/20/12 189 Utah Transit Authority; http://www.rideuta.com; Accessed 10/19/12 190 Utah Transit Authority; http://www.rideuta.com; Accessed 10/19/12

I-40 Interviewees believed the success of the FrontRunner service is its centrality in the valley, competitive travel times, and high frequency.191 The route is aligned with the long, linearly- constrained valley geography that has defined the region’s urban growth. The rails parallel an Interstate corridor. Stations have both excellent park-and-ride access and bus feeder service along the major arterials that run perpendicular to the North-South Corridor. Ridership was lower than initially anticipated. Planners attributed this to the Interstate 15 widening that occurred just before opening as well as the economic downturn that has reduced commute travel and roadway congestion in the region.192 Interestingly, the line was proposed as a mitigation measure for congestion on the adjacent Interstate that was simultaneously widened. The service has been close to ridership forecasts more recently, which planners attribute to the high frequency, economic recovery, and special event services provided at various times during the year. Interviewees indicated that they used many of the same rules of thumb for selecting the FrontRunner commuter rail alignment as TRAX light rail routes during their regional system planning process.193 However, they suggested that our proposed indicator-based method would be more reliable if it included multiple commuter rail projects because they felt service characteristics, particularly the limitations of railroad operations related to speed and service frequency, made commuter rail services significantly different from other rail transit modes. 191 Interviewee BN, in-person conversation, 8/20/12 192 Interviewee AE, in-person conversation, 8/20/12 193 Interviewee BN, in-person conversation, 8/20/12

I-41 I.6 Branch Avenue Extension (Washington, DC, Prince George’s County, MD) The Washington, D.C., region, including the District of Columbia and parts of the states of Maryland, Virginia, and West Virginia, is served by multiple modes of fixed-guideway transit. The primary urban rail transit system serving the District is the Washington Metropolitan Area Transit Authority’s (WMATA) 106-mile Metrorail subway system. The Branch Avenue extension—also called the Outer F extension in planning documents—extends from Anacostia Station in Southeast Washington, D.C., to southern Prince George’s County, Maryland, at the interchange of the Capital Beltway and Branch Avenue (WRRRTS 1992). The five-station, 6.5-mile-long section of the Green Line was opened on January 13, 2001, after more than 30 years of planning. This Branch Avenue case study suggests that early plans can be very difficult to modify, that transit system plans have been based on indicator-based methods, and that geographic and social equity are critical political considerations for transit planning, so much so that they can outweigh basic measures of success like ridership and project cost. Figure I-8: Route Diagram for Branch Ave. Green Line Extension, Washington, DC I.6.1 Expanding Washington D.C. Rail Transit The Washington, D.C., region’s subway was initially designed in the 1950s when Congress authorized the National Capital Planning Commission and the National Capital Regional Planning Council to conduct a four-year Mass Transportation Survey (U.S. DOT 1975). In response to 1959 hearings on the planning survey, Congress formed the National Capital Transportation Agency that proposed an 83-mile rail system for the region in 1962. After the formation of the Washington Metropolitan Area Transit Authority in 1967, the region’s rail plan was revised to a 98-mile system that included an alignment to the southeast of the U.S. Capitol along Suitland Parkway that would terminate near the Branch Avenue interchange with the Capital Beltway. As of 1975, schedules called for the entire 98-mile system to be under construction as of July 1981 (U.S. DOT 1975). The Outer F segment of the proposed system, the Branch Avenue corridor,

I-42 was one of several segments considered worthy of further study and scheduled to be among the last segments constructed (WMATA 1993). Ultimately, system studies lasted into the middle of the 1980s with over 50 Outer F segment alternatives and 20 Outer F station and yard layout alternatives considered. A primary debate over the Outer F route related to a realignment proposed in 1976 (Peat, Marwick, Michell & Co. 1977). Due to the expense of crossing the Anacostia River and the land takings that would be required, Washington, D.C., officials proposed a new crossing and a corresponding new southerly route through a low-income community within the District.194 Eyeing an opportunity, landowners and politicians in Prince George’s County, Maryland, promoted the District’s proposal as well as a new terminus near a horse racing track long slated for redevelopment by the county.195 The debate led to a change in the officially adopted alignment, a lawsuit, and considerable re-analysis of alternatives. Environmental document experts were hired at WMATA to address some of the issues that had made the agency susceptible to the lawsuits, and a former U.S. Department of Transportation administrator was contracted to negotiate a resolution.196 Ultimately, a modified version of the original Branch Avenue route—one that passed through the low-income Congress Heights neighborhood—was selected in 1993 (WMATA 1993). The extension was 6.4 miles long and consisted of approximately equal-length subterranean, surface, and elevated tracks. Construction on the $900 million project began in late 1995 and the line was opened on January 13, 2001 (Washington Times 9/24/95, Washington Post 1/13/01). I.6.2 Branch Avenue Operations The 6.4-mile Branch Avenue extension includes the last five stations built along the 21-station, 23-mile Metrorail Green Line. Service runs the full extent of the line from Branch Avenue in the south to the Greenbelt Station in the north. Along the Green Line, 13 stations are located in the District of Columbia, four stations are located northeast of Washington, D.C., in north Prince George’s County, Maryland, and four stations of the Branch Avenue extension are located south of Washington, D.C., also inside Prince George’s County. North of L’Enfant Plaza Station, just south of the National Mall, the Green Line and Yellow Line merge and co-operate northward along the remainder of the route. The Green Line provides connections to the Orange and Blue Lines at L’Enfant Plaza Station and two connections to the Red Line, one at Gallery Place Station in the central business district and the other at Fort Totten Station in the north of the District. Service is operated throughout regular service hours (Open: 5 a.m. Monday-Friday, 7 a.m. Saturday-Sunday; Close: midnight Sunday-Thursday, 3 a.m. Friday-Saturday nights) and at 12- minute frequencies during most hours except 6-minute frequencies during weekday morning and afternoon peaks.197 Upon opening in 2001, the project experienced greater than anticipated ridership (Washington Post 1/19/01). Metro anticipated that after six months of service 18,000 daily riders would board at the stations along the new extension. However, on the second day of operations, ridership reached approximately 19,500 boardings. After only two weeks, ridership exceeded 30,000 (Washington 194 Interviewee BQ, telephone conversation, 7/11/12 195 Interviewee BR, telephone conversation, 7/19/12 196 Interviewee BQ, telephone conversation, 7/11/12 197 Metro Pocket Guide <http://www.wmata.com/pdfs/pocket_guides/english.pdf>

I-43 Post 1/25/01). Much of the difference in predicted and actual ridership was thought to be driven by free parking that had been offered temporarily at the new Green Line stations. Estimates had suggested 4,000 riders would switch from the Blue Line—where parking costs were $1.75 per day—but the actual number was closer to 12,000 riders. Though the Green Line is still known for its exceptional ridership, after several months and revisions to the parking policy, ridership normalized at levels in line with predictions. As found in the TCRP H-42 database, ridership was just over 25,000 as of 2009. I.6.3 Planning a Successful Transit Project The five-station Outer F alignment of the Washington, D.C., Metrorail system was initially proposed by the National Capital Planning Commission and the National Capital Regional Planning Council as part of a high-level rail transit system plan. Specific project plans were developed by National Capital Transportation Agency, predecessor to WMATA, in 1962. More than 30 years of detailed planning was carried out primarily by WMATA and its consultants with construction commencing in 1995. The FTA also played a role in moving the project through the federal environmental review process. The U.S. Congress, State of Maryland, and Prince George’s County were the primary government bodies involved in the planning of the project. The 98-mile system plan adopted in 1967 was motivated by the National Capital Regional Planning Council’s 1961 wedges and corridors concept presented in their Year 2000 Policies Plan. (U.S. DOT 1975) The transit services were intended to be competitive with automobile travel to alleviate congestion, address air quality concerns related to vehicle-miles traveled, and provide an enhanced experience relative to existing bus services, which were suffering from competition from private automobile travel. The wedges and corridors plan sought to focus urban growth every few miles at jointly located transit stations and roadway intersections along radial transportation corridors, thus allowing for the preservation of green space wedges between the corridors. A critical element of 1961’s Year 2000 Policies Plan was the development of a circumferential freeway that would connect the entire region via interchanges with the radial corridors. Notably, the plan defined radial corridors emanating in all directions from the central city and transit projects were proposed in each radial corridor to achieve an equitable allocation of benefits.198 While some corridors were aligned with fast-growing suburbs where it was imagined Metrorail would alleviate growth pains and allow for further economic development, in other corridors—like the Outer F alignment—Metrorail’s planners considered the benefit to be largely limited to improved transit performance for transit-dependent populations and the ability to attract middle-class riders that would appreciate central city parking savings.199 Local governments, like Prince George’s County, assumed that the economic development benefits associated with Metrorail were universal and system promoters, including WMATA and its consultants, did little to correct this notion because it benefited their causes. Early system plans focused on operating cost coverage as a key success metric.200 Carried out by a firm that has since become a global accounting firm, KPMG, the early plans were essentially financial forecasts based on “customer” patronage assumptions (i.e., ridership forecasts). A primary assumption of those plans was that many Metrorail patrons would access the system by private 198 Interviewee BQ, telephone conversation, 7/11/12 199 Interviewee BS, telephone conversation, 7/30/12 200 Interviewee BS, telephone conversation, 7/30/12

I-44 automobile. Surface park-and-ride facilities at each of the outlying stations were sized to accommodate 500 or 1,000 stalls according to the earliest plans (U.S. DOT 1975). During the late stages of system planning, it was determined that the Outer F alignment should reach the circumferential beltway for easy automobile access or not be built.201 In fact, later studies accommodated 3,000 stalls at the Branch Avenue Station terminus near the Branch Avenue and I- 495 interchange because a lack of parking would have dampened ridership forecasts.202 (Today, 3,074 stalls are located at the Branch Avenue Station and parkers pay a $4.50 daily fee.203) Also, in early iterations of the alignment, stations were consistently located at the intersection of the Suitland Parkway and major arterial roadways (e.g., Alabama Avenue) for ease of automobile access into stations without impacting residential streets (U.S. DOT 1975). Success, defined by system-wide operating cost coverage, was also heavily influenced by the financial windfall generated when system planners assumed long-distance commuter bus routes could be terminated.204 The bus services, which generally traveled along congested roadways and contributed to central city air quality issues, would be replaced by short-haul bus trips along less- congested suburban roadways that terminated at far-flung Metrorail stations serving as suburban bus transfer hubs. Each of the Outer F stations was originally slated to have between three and seven bus bays to accommodate rail-to-bus transfers (U.S. DOT 1975). Like parking stalls, this capacity was increased further during the planning process (WMATA 1993). (Today, there are 15 bus bays at the Branch Avenue Station.205) The concepts put forward by regional plans dictated that the Outer F alignment be co-located with Suitland Parkway, and assumptions made by project planners produced a proposal which relied heavily on travel time competitiveness and central city parking savings as drivers of ridership.206 A four-step transportation demand model developed by the regional council in the 1970s produced a ridership forecast that confirmed the feasibility of the early system plan (U.S. DOT 1975). With the exception of only a few stations and the route of the Branch Avenue alignment, the current WMATA system reflects the 1967 plan (Schrag 2006). Detailed studies of the Anacostia River crossing conducted in the mid-1970s led to a discussion of alternative alignments. The original crossing proposal required the line to pass under the Navy Yard between the Waterfront Station and the proposed Downtown Anacostia Station. Because of the cost of environmental cleanup within Navy Yard, the required demolition of several historical structures within Navy Yard, and concerns about construction impacts in Downtown Anacostia (along Good Hope Road) several alternative crossing concept plans were put forward in 1976 (Wallace, McHarg, Roberts and Todd 1976). The alternative that received the greatest attention shifted the Navy Yard Station considerably westward to allow the line to turn south and cross the Anacostia without passing through or under Navy Yard proper. This westward shift in alignment put the crossing on path with a relatively undeveloped linear greenway (as opposed to the Suitland Parkway corridor) that provided a fairly straight shot to a declining horse racing facility, the Rosecroft Racetrack, just south of the Capital Beltway. Analyses began to consider both a rerouting of the original alignment (still along Suitland Parkway) and the newly proposed route to the Rosecroft Racetrack. In the 1977 EIS, a third 201 Interviewee BS, telephone conversation, 7/30/12 202 Interviewee BS, telephone conversation, 7/30/12 203 Washington Metropolitan Area Transit Authority; http://www.wmata.com/rail/parking/; Accessed 10/22/12 204 Interviewee BT, telephone conversation, 7/24/12 205 Visual Survey of Satellite Image; Google Maps; Accessed 10/22/12. 206 Interviewee BQ, telephone conversation, 7/11/12

I-45 alignment was proposed that followed the Rosecroft proposal for half its distance and then turned northeastward along Southern Avenue (the border between the District of Columbia and Maryland) to Suitland Parkway where the route followed the original Branch Avenue alignment to its terminus at Branch Avenue Station (S-Curve alignment) (WMATA 1984). While this alternative was heavily studied, political interests in Maryland focused on the redevelopment opportunities at Rosecroft Racetrack. Proponents of the Rosecroft alternative included Prince George’s County officials who saw declining tax revenues from the racetrack and landowners who sought to redevelop the declining horse racing facility and surrounding property into a commercial center at the scale of the emerging Tyson’s Corner area or the contemporaneously proposed Reston Town Center development, both in the Virginia suburbs.207 The arguments were so convincing that the Prince George’s County Council approved the Rosecroft alignment and WMATA board, made up of four Virginia, four District, and four Maryland representatives, followed suit in 1978 (Washington Post 5/10/78). Environmental documents produced by WMATA in 1979 still considered the Branch Avenue route as an alternative, partly because of acrimony between Maryland officials over the alignments (WMATA 1979).208 Members of the Maryland House of Representatives suggested that the route was unduly selected and business owners along the Branch Avenue alignment sued in federal court over economic harm caused by the rerouting (Washington Post 2/27/80, Washington Post 10/14/80). The debate contributed to project delays because ongoing construction of other portions of the Green Line were dependent on decisions regarding the location of the Anacostia River Crossing. Because staff saw no way to resolve the conflict while it was being adjudicated, WMATA board members decided to set strict timetables for the cases to be settled and construction to proceed on the Rosecroft alignment by the end of 1984.209 The three viable alternatives—a meandering route following Suitland Parkway, a relatively direct route to Rosecroft Racetrack, and an S-Curve following a portion of the Rosecroft alignment but terminating at Branch Avenue—were the focus of studies and public meetings in the early 1980s.210 It is in the resolution of the debate that our research found a well-documented debate over definitions of transit project success. 207 Interviewee BR, telephone conversation, 7/19/12 208 Interviewee BQ, telephone conversation, 7/11/12 209 Interviewee BR, telephone conversation, 7/19/12 210 Interviewee BQ, telephone conversation, 7/11/12

I-46 Figure I-9: Alternative Alignments (Easternmost Infeasible Because of Navy Yard Station Location) (WMATA 1984) According to documents produced after the alignment debate was settled, the S-Curve to Branch Avenue alternative was officially considered preferable due to higher projected ridership, better transit service to transit-dependent populations, fewer displacements, greater secondary development potential, greater reduction in vehicle-miles traveled, and more regional air quality improvement (WMATA 1993). While there were technical analyses to quantify many of these measures of success, project planners found that politicians ignored data that did not corroborate this opinion and relied heavily on simple indicators of success that “ended up playing as much if not more in the final decision than the actual technical data.”211 211 Interviewee BS, telephone conversation, 7/30/12

I-47 Among all of the issues discussed during the process, one could argue that the environmental and historical impacts were the most influential.212 The impacts on Navy Yard’s historical structures, in addition to the environmental cleanup required at several sites within Navy Yard, were the impetus for discussing alignments outside of that defined by the 1967 plan. Later, when a toxic ash dump was found in the path of construction at the Elizabeth Hospital facility and when local environmental advocates became vocal about impacts to creeks, including Oxen Run Creek, the route was quickly altered again. From a political standpoint, these were indicators of cost, delay, and probable project failure that were to be avoided at all costs.213 Another very salient measure of project success was the number of transit-dependent households that would be served by Metrorail stations. Throughout the process, comparisons of alternatives included the number of census tracts in the station service areas that were defined as “very highly transit-dependent” (WMATA 1979). An analysis of 1980 census data determined that between 21,000 and 30,000 highly transit-dependent people would be served by various alignment alternatives. The Rosecroft alignment was projected to serve only 21,100 transit-dependent people, the lowest of any alternative. This was a very powerful political argument used against the Rosecroft alignment.214 According to at least one analysis, the Suitland Parkway alignment would have actually served the most transit-dependent people (WMATA 1984). However, District politicians had become wedded to the Congress Heights Station location that was part of the board-approved Rosecroft alignment.215 Congress Heights became the symbol of a transit-dependent community and any suggestion to not serve the area was considered an injustice. The service to Congress Heights was a strong argument for the Rosecroft or the S-Curve alignments rather than the Suitland Parkway alignment. Later, in spite of the data suggesting that the Suitland Parkway alignment would have stations closer to more transit-dependent individuals, the lack of rail transit in Congress Heights was identified as a “serious problem” because fewer than 40 percent of the adult residents in the Congress Heights area owned an automobile (WMATA 1993). This suggests two things. First, transit plans can be very sticky once a constituency identifies with a proposal. Second, sometimes the most salient indicator of success for a project is whether or not a particular location will be served directly by a station. In fact, this is the essence of the entire Rosecroft debate. Another social impact that was highly sited in the alignment debate was the number of displacements that would be required. Aside from the Navy’s resistance to the alignment passing under its facilities, another argument to move the Anacostia River crossing was the potential impact to businesses along the original route. The original route would have passed through a dense commercial street with predominantly African-American-owned businesses. That portion of the line would have required the taking of numerous commercial properties and the closure of the street for several years as cut-and-cover construction took place. As one planner phrased it, “We had considered putting the Anacostia station in the middle of [the commercial area] and it would have destroyed it.”216 Comparisons of the other alternatives showed that they would all require fewer takings than the original proposal. While each of the three alternatives would have required approximately the same 212 Interviewee BQ, telephone conversation, 7/11/12 213 Interviewee BR, telephone conversation, 7/19/12 214 Interviewee BQ, telephone conversation, 7/11/12 215 Interviewee BQ, telephone conversation, 7/11/12 216 Interviewee BQ, telephone conversation, 7/11/12

I-48 number of business and institutional takings, an important differentiator between the alternatives was the number of residential units that would be demolished. While other alignments would have required takings of greater acreage (including existing public parkland), the Rosecroft alternative was considered weakest because it would have required 125 residential units to be taken (versus 93 or 52 for the S-Curve and Suitland, respectively) (WMATA 1984). One of the strongest arguments for the Rosecroft alignment was the real estate redevelopment potential at the terminus station.217 However, according to data collected as part of the review process, there was as much or more development potential—defined by developable acres—within 2,000 feet of Prince George’s County stations along the S-Curve and Suitland alignments (WMATA 1984). That said, after several experiences with development occurring around stations in the Virginia suburbs and downtown, Metrorail was generally considered a motivation for economic development no matter where stations were built or what land uses surrounded them.218 Thus, an alignment like the S-Curve, which had one more station than the other alternatives, was perceived to have greater potential economic development impact.219 The size of developable parcels and the number of stations were statistics used by proponents of the Branch Avenue alignment to neutralize one of the major arguments for the Rosecroft alternative. Another argument for the Rosecroft alternative was its lower capital cost (WMATA 1984). The route was shorter, straighter, and had fewer points of conflict (e.g., stream crossings, roadway crossings). However, faulty logic was used by advocates of the Branch Avenue terminus to neutralize this argument.220 According to initial estimates, the Rosecroft and Suitland alignments were within $5 million of capital cost of one another (WMATA 1984). Thus, it was argued that cost was not a differentiator between the Rosecroft and Branch Avenue termini. Yet, the S-Curve alternative—which also terminated at Branch Avenue—was estimated to cost approximately $130 million more than either the Rosecroft or Suitland alternative. Nonetheless, the S-Curve alignment was considered on par with the other options and additional costs were attributed to the additional station at Congress Heights—the cost of serving transit-dependent populations (again, this was based on an argument to serve Congress Heights even though the Suitland alternative would serve more transit-dependent residents). Our TCRP H-42 research identifies ridership as a prime measure of transit project success. A frequently discussed indicator of ridership in the planning of the Outer F alignment was the existing federal employment centers located along the Suitland and S-Curve alternatives.221 For instance, the U.S. Census Bureau was located at the Suitland Federal Center. During the alignment debate, this was contrasted with the Rosecroft alignment that provided minimal direct access to any suburban employment. Likewise, the Rosecroft Racecourse was promoted as a major ridership generator. However, it was determined that transit would achieve little mode share because the horse races typically operated at night during off-peak transit service periods and, based on an informal survey conducted by planning consultants, most patrons owned cars.222 A recalibrated model produced a lower ridership forecast. 217Interviewee BQ, telephone conversation, 7/11/12 218 Interviewee BR, telephone conversation, 7/19/12 219 Interviewee BS, telephone conversation, 7/30/12 220 Interviewee BR, telephone conversation, 7/19/12 221 Interviewee BR, telephone conversation, 7/19/12 222 Interviewee BT, telephone conversation, 7/24/12

I-49 While early ridership estimates had been based on census track-level data and yielded very distinct ridership estimates, subsequent refinements contributed to model outputs that suggested the alternatives would experience similar ridership demand.223 Travel isochrones overlaid on detailed maps depicting individual single-family homes were used to recalculate the number of residents within stations’ service areas. When input into the models, this impacted the Rosecroft alternative because of the limited roadway infrastructure that existed in the area. Because the budget of the transit project could fund only a limited number of roadway improvements and Prince George’s County was not willing to commit to roadway construction, lower ridership projections for the Rosecroft alignment were maintained. Ultimately, ridership projections were not pivotal considerations in the Outer F alignment debates. Estimates conducted in the early 1980s suggested that the S-Curve and Suitland alternatives would have over 70,000 daily riders while the Rosecroft alignment would have just shy of 66,000 (WMATA 1984). As one project consultant noted, “Although the [ridership] data suggested Branch Avenue [was preferable], it was not so compelling a case that you would select Branch Avenue just by the data.”224 In fact, the ridership and operating characteristics of the alignments were so similar that the difference in projected annual net operating deficit of the proposals was less than 7% (WMATA 1984). In the end, the ridership figures were not used by officials to publish comparison benefit-cost measures of the alignments (WMATA 1984). Neither the operating deficits nor the capital cost figures were considered relative to patronage. Nor were costs considered relative to one of the most noted benefits of the project: rail access for transit-dependent people. Using 1984 comparative statistics to calculate such benefit-cost figures, the results (found in the table below) would have pointed to the Suitland alternative rather than the Rosecroft option (the alternative selected in 1978) or the S-Curve option (the route ultimately constructed). Table I-2: Benefit-Cost Calculations for Routes Under Consideration in 1984 (WMATA 1984) Benefit-cost measure Rosecroft S-Curve Suitland Total capital cost per trip (1990 ridership) $30.98 $33.89 $29.12 Operating deficit per trip (1990 ridership) $0.58 $0.54 $0.51 Capital cost per transit-dependent person in station catchments (1980) $35,180 $34,532 $25,826 Operating deficit (1990) per transit-dependent person in station catchments (1980) $654 $546 $450 Ultimately, the impasse was broken and the county, District, and WMATA selected the S-Curve alignment through Congress Heights and terminating at Branch Avenue. It met the demands of District politicians to serve a particular transit-dependent neighborhood, passed through major employment centers, avoided further lawsuits by businesses that had relied on the 1967 Metrorail plan to make investment decisions, and provided excellent automobile and bus access without considerable investment in new roadway infrastructure. After WMATA approved the S-Curve route and the court injunction was lifted in late 1984, construction commenced on the portion of the 223 Interviewee BT, telephone conversation, 7/24/12 224 Interviewee BS, telephone conversation, 7/30/12

I-50 Green Line from L’Enfant Plaza (the intersection with the Yellow/Blue Line) to Anacostia in 1985. Advocates of the Suitland Parkway alignment continued to agitate for shifting the alignment from the S-Curve throughout the late 1980s and early 1990s, but District interest in serving Congress Heights and fears of reopening the debate squelched any further realignment. The extension to Anacostia opened in 1991 and the segment of the Green Line in northern Prince George’s County opened in 1993. At that time, final plans were approved for the S-Curve alignment and a construction contract was signed in 1995 with Green Line Metrorail service to Branch Avenue Station commencing in January 2001. In spite of the difficulties and delays associated with the Outer F portion of the WMATA Green Line, WMATA staff currently considers it one of the most successful segments of the Metrorail system.225 Unlike the base system that was constructed in the 1970s and 1980s, planners of the Green Line had to prove their case for the line time and time again to Congress, to WMATA’s member jurisdictions, and to diverse groups who advocated for alternative alignments and to stop construction altogether. In the end, the line achieved the ridership projections while providing high- quality transit service to one of the most economically depressed parts of the Washington, D.C., metropolitan area. 225 Interviewee BU, telephone conversation, 8/27/12

I-51 I.7 Regional Contexts The following section provides brief overviews of the regional contexts of each case study. I.7.1 Charlotte Region The Charlotte-Gastonia-Rock Hill, NC-SC MSA had an estimated 2011 population of 1.8 million.226 The region includes five counties in North Carolina and one in South Carolina, and covers almost 3,200 square miles. Centered on the City of Charlotte (population: 751,087), the region is the largest in North Carolina, and 21st largest in the United States.227 Mecklenburg County, the county in which Charlotte is located, is 523 square miles and has a population of 919,628.228 The Charlotte region is located in the rolling hills of southwestern North Carolina’s Piedmont region just 85 miles southeast of the Appalachian Mountains, and 180 miles northwest of the Atlantic Ocean. Charlotte is the major banking center of the Southeastern United States and is the nation’s second-largest banking and financial hub. Bank of America’s headquarters and the east coast operations of Wells Fargo are among the major financial institutions located in Charlotte. The region is home to 273 Fortune 500 Companies, seven of which are headquartered in Mecklenburg County.229 Charlotte is served by two main freeways, Interstate 77 and Interstate 85, both of which connect the region to other major southeastern metropolitan areas. Most of the City of Charlotte lies within a beltway, I-485. The city’s central business district, Uptown, is encircled by the I-277 freeway. While the central city has a grid-based street pattern, the majority of the region is built around arterial roads that radiate out from the center city. Transit in the Charlotte region is operated by the Charlotte Area Transit System (CATS). The agency operates over 70 local and express bus routes and paratransit, in addition to the LYNX Blue Line, the Charlotte region’s only light rail line. The Charlotte Transportation Center (CTC) in Uptown Charlotte, the northernmost stop on the Blue Line, is the region’s multimodal transit hub. Local and express bus routes radiate out of central Charlotte in all directions, some reaching into neighboring South Carolina. Of the 344,436 workers commuting to work in 2010, 77.6% drove alone, 10.6% carpooled, 3.7% took public transportation, 2.2% walked, 0.8% used other means, and 5.2% worked at home.230 I.7.2 Dallas Region The Dallas-Fort Worth-Arlington, Texas MSA covers 9,286 square miles in 12 counties.231 The MSA, also called the Dallas-Fort Worth Metroplex, is the largest MSA in Texas, and the fourth- largest in the United States. It is also the 12th largest metropolitan economy (global scale) by 2005 226 Annual Estimates of the Population of Metropolitan and Micropolitan Statistical Areas: April 1, 2010 to July 1, 2011 <http://www.census.gov/popest/data/metro/totals/2011/tables/CBSA-EST2011-01.xls 227 U.S. Census – State & County Quick Facts 228 2010 U.S. Census 229 Chamber of Commerce; http://charlottechamber.com/eco-dev/charlotte-s-economy-demographics/ 230 U.S. Census Bureau – 2010 American Community Survey 1-Year Estimates, “Selected Economic Characteristics” 231 http://www.census.gov/popest/data/metro/totals/2011/tables/CBSA-EST2011-01.xls

I-52 GDP$232 and its 2011 population was estimated to be 6.56 million. The Dallas-Fort Worth- Arlington MSA contained 2,968,500 jobs in April 2012.233 The MSA includes the Dallas-Plano- Irving and Fort Worth-Arlington Metropolitan divisions and the Dallas-Plano-Irving MD contains 70% of the areas workforce. The MSA’s largest employment sector is trade, transportation and utilities.234 The Dallas-Fort Worth-Arlington MSA is characterized mostly by prairie land. Around Dallas is the blackland prairie—named for the fertile black soil and historically used to grow cotton.235 Around Fort Worth is the Fort Worth Prairie, which contains low fertility soil. Traditionally it was used for ranchland, but it is now the primary regional location for oil refining. The north-Dallas area suburbs are coined the “Silicon Prairie” because of the high number of technology firms and corporate offices in the region (AT&T, HP, Microsoft, etc.). The Richardson Chamber of Commerce went so far as to trademark “Telecom Corridor” to refer to their high-tech business community.236 The Dallas-Fort Worth region is served by two rail transit systems and a variety of bus and other transit services. DART operates the light rail system, and jointly (with the Fort Worth Transportation Authority) runs the area’s commuter rail service, the Trinity Railway Express (TRE). The DART light rail system consists of three color-coded lines totaling 58 stations and 77 miles of track, now the longest light rail system in the country.237 The TRE system adds another 10 stations and 34 miles. DART light rail serves over 71,000 passenger trips each weekday, while TRE serves 8,500 daily trips. The DART system also includes bus service on over 100 routes, serving over 125,000 weekday boardings. DART light rail is operated with modern light rail vehicles called Super Light Rail Vehicles, featuring level boarding and increased passenger capacity. DART light rail headways average about 15 minutes system-wide, but the Red Line and Blue Line have supplemental Orange Line service that increases frequency during peak hours to about 7 minutes. Of the 2,999,949 estimated workers in the DFW MSA, 81% commuted alone by auto, 10% carpooled, 1% took transit, 1% walked, 2% took a taxicab, motorcycle, bicycle, or other means, while 5% worked at home.238 I.7.3 Eugene Region The Eugene-Springfield, OR, MSA, which covers 4,722 square miles in one county (Lane), had an estimated 2011 population of 353,416.239 The region, centered on the cities of Eugene (population: 156,185) and Springfield (population: 59,403), is the third-largest in Oregon, and 144th-largest in the United States. Lane County’s population grew almost 9% between 2000 and 2010. Lane County stretches from the Pacific Ocean to the Cascade Mountain range in central Oregon. The center of the metropolitan area is located in the middle of the county, in the Willamette Valley. 232 Kessler, Dan. “Metropolian Transportation Update – International Right of Way Association North Texas Chapter.” North Central Texas Council of Governments. June 09 2009. 233 Work Area Profile Analysis. LEHD. On the Map. Census Bureau; http://onthemap.ces.census.gov/ 234 BLS “Dallas-Fort Worth Area Employment – April 2012.” Southwest Information Office. News Release April 2012. 235 Dallas Fort Worth Tourism; http://dfwtourism.com/demo/ 236 http://www.telecomcorridor.com/about-us/telecom-corridor-geneaology-project-1 237 http://www.dart.org/about/dartfacts.asp; Accessed 10/22/12 238 U.S. Census Bureau. “Means of Transportation to Work by Age.” 2010 ACS 1 year estimates. 239 http://www.census.gov/popest/data/metro/totals/2011/tables/CBSA-EST2011-01.xls; Accessed 7/28/12

I-53 Eugene and Springfield, the MSA’s two primary cities, are located on opposite sides of the Willamette River, in the southernmost corner of the valley, surrounded by mountains on three sides. The centers of the two cities are separated by only four miles. Eugene has the region’s largest central business district, and is home to the University of Oregon, which had nearly 25,000 students in 2011.240 Just across the Willamette River is Springfield, the region’s second-largest city, which has a smaller downtown. Much of the recent growth in employment has occurred at the fringe of the urban area, notably in office parks in the northwest portion of Springfield near the I-5/Randy Pape Beltline interchange. The region’s economy, originally heavily timber-based, has since diversified and now consists of manufacturing, high-tech and healthcare sectors. Interstate-5 bisects the Eugene-Springfield MSA, forming the border between the two cities but not serving either downtown. A spur, I-105 connects downtown Eugene to I-5 and areas east, but not directly to downtown Springfield. An incomplete Outer Loop (OR-569) and a short North-South freeway (Delta Highway) comprise the rest of the region’s limited access highway network. The Eugene-Springfield region is served by the Lane Transit District (LTD), which carries almost 39,000 weekday riders on 34 standard bus routes and its EmX BRT route.241 The network is, for the most part, a radial one, with the majority of routes fanning out from Eugene Station, a transit center in downtown Eugene. A handful of routes radiate out from Springfield Station in downtown Springfield. The system also features several outlying transit centers and almost 20 park-and-ride lots. According to 2006-2010 ACS estimates, just over 80% of Lane County residents commuted to work by auto. The next largest share of workers (8%) worked from home. Transit, cycling, and walking each captured roughly 4% of the commute share. I.7.4 Portland Region The Portland-Vancouver-Hillsboro, OR-WA MSA, which covers 6,684 square miles in six counties (four in Oregon and two in Washington), had an estimated 2011 population of 2.26 million.242 The region, centered on the city of Portland (population: 583,776), is the largest in Oregon, and 23rd-largest in the United States. The Portland region has an elected government body, Metro, that oversees long-range land use and transportation planning. Metro’s own analysis shows that employment in the region grew by 7.4% overall between 1996 and 2005, with the vast majority of that growth occurring in outlying Washington and Clark (WA) counties. Multnomah County, which contains Portland and its eastern suburbs, now holds roughly 36% of the region’s jobs.243 Portland centers on the Willamette River near its terminus at the Columbia River—which drains into the Pacific—and was the site of a 19th century seaport. Hydraulic power and wartime shipbuilding propelled growth in the 20th century. Today, Portland sits at the junction of two Interstate highways, I-84 (East-West) and I-5 (North-South). In addition to these trunk routes, the metropolitan area is served by two auxiliary routes, I-405, which forms half of a loop around Portland’s CBD, and I-205, an eastern bypass, as well as several shorter connecting limited access 240 http://admissions.uoregon.edu/profile.html; Accessed 10/22/12 241 http://www.ltd.org/about/history.html?SESSIONID=2177f5ac0d3a59aeab899b7df2c2bb81 242 http://www.census.gov/popest/data/metro/totals/2011/tables/CBSA-EST2011-01.xls 243 http://library.oregonmetro.gov/files/regionaltrendstravelfinal.pdf

I-54 highways. Portland was the first American city to tear down an existing limited access freeway when, in 1974, Harbor Drive was demolished and replaced with a park, reconnecting the central business district with the riverfront. The Portland region has a relatively extensive and well-developed bus-, light rail-, and streetcar- based transit system, operated by TriMet. The system includes the MAX light rail network, which started with the opening of Eastside MAX to Gresham in 1986. TriMet has consistently expanded the system, which now consists of four lines (Red, Blue, Green, and Yellow) radiating out of two dedicated alignments that cross one another perpendicularly in downtown Portland. In addition to the ever-expanding light rail system, TriMet operates a grid of frequent bus service throughout the metropolitan area. TriMet also participates in the operations of Portland’s downtown streetcar facilities which expanded outside of the downtown as of 2012. In 2000, 84% of Portland area workers commuted by auto, down from almost 90% in 1990. During that same period, public transportation’s share rose from 5.8% to 6.7%. 2010 ACS five-year estimates show transit’s share remaining flat, at 6.6%.244 According to TriMet, between 1990 and 2000, transit ridership “increased (58%) faster than population growth (24%) and overall growth in vehicle-miles traveled (35%).” The Portland region averages roughly 80 annual transit trips per capita, second only to New Orleans among American metropolitan regions of similar sizes.245 I.7.5 Salt Lake City Region The Salt Lake City-Ogden, UT, MSA has approximately 1.15 million246 residents. The region, centered on Salt Lake City, is the most populous in Utah and 48th-largest in the United States. A related larger regional geography is referred to as the Wasatch Front and consists of two MSAs (SLC-Ogden and Provo, UT) and has a combined population of over two million. Salt Lake City is the most populous city in the region, and its 109-square-mile area is bounded on two sides by mountain ranges and on a third side by the Great Salt Lake. The city lies at the junction of two cross-country Interstate highways, I-80 (extending east to New York City and west to San Francisco) and I-15 (extending north to Canada and south to Mexico). An incomplete belt route, I- 215, and a spur (Highway 201) comprise the rest of the highway network of the central metropolitan region. A grid of wide, regularly spaced arterial surface roads blanket the region. The Salt Lake City International Airport is five miles from downtown Salt Lake City and, as a hub for Delta airlines, is the 23rd-busiest airport in the nation.247 The top three employers in Salt Lake County are the University of Utah, Intermountain Health Care, and the State of Utah.248 The top employer in Weber County is the Internal Revenue Service, while the largest employer in Davis County is Hill Air Force Base. 249,250 Salt Lake City has become an attractive location for technology sector firms. Forbes listed Salt Lake City as the fourth-best city in the nation for tech jobs, citing Adobe, Electronic Arts, and Twitter.251 244 2006-2010 American Community Survey 5-Year Estimates 245 http://library.oregonmetro.gov/files/regionaltrendstravelfinal.pdf 246 Annual Estimates of the Population of Metropolitan and Micropolitan Statistical Areas: April 1, 2010 to July 1, 2011 <http://www.census.gov/popest/data/metro/totals/2011/tables/CBSA-EST2011-01.xls 247 Utah at a glance <http://www.edcutah.org/documents/Utah-At-A-Glance_000.pdf> 248 Salt Lake County Profile <http://www.edcutah.org/documents/SaltLakeCounty.pdf> 249 Weber County Profile <http://www.edcutah.org/documents/WeberCounty.pdf> 250 Davis County Profile < http://www.edcutah.org/documents/WeberCounty.pdf> 251 SLC ranks High as tech job hot spot <http://www.edcutah.org/documents/SLCTechJobHotSpot.pdf>

I-55 Public transportation in the Salt Lake City region is provided by the Utah Transit Authority (UTA), which operates bus, light rail and commuter rail routes throughout the entire region. UTA’s system averages over 150,000 daily boardings on 131 routes in six counties. The relatively new light rail network, TRAX, opened in 1999 with the Salt Lake City to Sandy (now Blue) line. The light rail system consists of three color-coded lines (Red, Blue and Green) with a new extension from downtown Salt Lake City to the airport slated to open in early 2013 and a 3.8-mile extension south from the current Sandy Blue Line terminus to open in 2014. A streetcar line in South Salt Lake is also anticipated to open in 2013 and will connect with the Blue, Red, and Green Lines at Central Pointe station—where the West Valley segment of the Green Line intersects with the North- South TRAX trunk line. The streetcar provides rail access to neighborhoods to the east of the trunk line and I-15 and just north of I-80. FrontRunner, a commuter rail line currently serving Ogden and points north of Salt Lake City, opened in 2008, and connects to the TRAX network at the Salt Lake City Intermodal Center. The FrontRunner will extend 45 miles south from the Salt Lake City Intermodal Center to Provo starting in late 2012. UTA's total system ridership in 2011 reached 41,553,315 with more than 22.6 million on UTA buses, 15.2 million on TRAX, and 1.6 million on FrontRunner.252 Of the 522,765 people commuting to work in the Salt Lake City, UT, MSA in 2010, 77.7% drove alone, 11.3% carpooled, 2.9% took public transportation, 2.3% walked, 1.9% arrived by other means, and 4.0% worked from home.253 I.7.6 Washington, D.C., Region The Washington-Arlington-Alexandria DC-VA-MD-WV MSA, which covers 5,564 square miles surrounding the nation’s capital, had an estimated 2011 population of 5.58 million.254 In addition to the District of Columbia, the MSA includes five counties in Maryland, nine counties in Virginia, and one county in West Virginia.255 The Washington, D.C., region is centered on the District of Columbia. The District is approximately 60 square miles with a 2010 population of approximately 600,000 people.256 As of January 2012, the region’s labor force comprised 3,174,984 people.257 The five largest employers were the U.S. Department of Defense, Fairfax County Public Schools, County of Fairfax, Prince William County School Board, and Booz, Allen and Hamilton.258 The technical services sector has historically been the largest employer in the region.259 The District of Columbia lies at the confluence of the Potomac and Anacostia rivers. Encircling the District, the I-495 Capital Beltway passes through Virginia and Maryland and intersects with I- 95 to the north and south, I-66 to the west, and I-270 to the northwest. A loop consisting of I-395, I- 695, and I-295 are the only direct Interstate connections inside the District, though five limited access parkways also enter the District’s borders. 252 UTA – 2011 Year in Review <http://www.rideuta.com/uploads/2011inreview.pdf> 253 2010 American Community Survey 1-Year Estimates – “Selected Economic Characteristics” 254 Annual Estimates of the Population of Metropolitan and Micropolitan Statistical Areas: April 1, 2010 to July 1, 2011<http://www.census.gov/popest/data/metro/totals/2011/tables/CBSA-EST2011-01.xls> 255 Economic Census Local Business Snapshot <http://www.census.gov/econ/census/snapshots_center/dc.html> 256 www.dc.gov 257 Bureau of Labor Statistics <http://www.bls.gov/xg_shells/ro3fx9512.htm> 258 Page 20 <http://virginialmi.com/report_center/community_profiles/5121S47890.pdf> 259 Economic Census: Local Snapshot <http://www.census.gov/econ/census/pdf/dc.pdf>

I-56 The Washington Metropolitan Area Transit Authority (WMATA) is the region’s dominant transit provider. In addition to its substantial Metrobus fleet, WMATA’s Metrorail serves 86 stations along 106 miles of track.260 During rush periods, the Green Line operates in intervals of 6 minutes between trains with a train size of 6-8 cars. The midday intervals between trains is 12 minutes and the evening intervals between trains is 20 minutes with a train size of six cars. Residents of the eight WMATA compact jurisdictions in D.C., Maryland, and Virginia generate 88% of weekday ridership.261 Riders outside of the WMATA service area have transit alternatives. In addition to the Metrobus and Metrorail, the Washington-Arlington-Alexandria region is served by several other bus services and two other commuter rails (MARC and VRE).262 Of the 2,931,890 commuters in 2010, 65.6% drove to work alone, 10.6% carpooled, 14% used public transportation, 3.5% walked, 1.5% arrived by other means, and 4.9% worked from home.263 260 Metro Facts <http://www.wmata.com/about_metro/docs/metrofacts.pdf> 261 Regional Transportation <http://www.wmata.com/getting_around/regional_transit.cfm> 262 Regional Transportation <http://www.wmata.com/getting_around/regional_transit.cfm> 263 U.S. Census Bureau “Commuting to Work” 2010 ACS 1 year estimates

J-1 APPENDIX J: Data Sources Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit 1. Costs 1a. Capital and Operaon Costs Public Transportaon Factbook (PTFB) American Public Transportaon Associaon (APTA) (non-profit transit agency industry organizaon) Capital and operang expenses 2010 (2008 data), and annually (2003-2010 available online). Some historical tables date to as early as 1902. Naonal and Canada Transit agency Naonal Transit Database (NTD) U.S. DOT, Federal Transit Administraon (FTA) Operang expenses Annual summaries available from 1996 to 2009. Time- series data files contain agency summaries from 1991 to 2009. Naonal Transit agency 1b. Discount Rates Stascs & Historical Data on H.15 Selected Interest Rates The Board of Governors of the Federal Reserve System Interest rates Daily 1954-present. Naonal Naonal 2. System and Financial 2a. Service Supplies TOD Database Center for Transit-Oriented Development; Center for Neighborhood Technology Locaons of U.S. fixed- guideway staons 2000 (employment 2002- 2008). Naonwide where fixed- guideway transit is present Fixed-guideway transit staons and surrounding area

J-2 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit Naonal Transportaon Atlas Database (NTAD) U.S. DOT, Bureau of Transportaon Stascs (BTS), Research and Innovave Technology Administraon (RITA), Presents locaons of fixed-guideway transit facilies 2010, 2009, 2008, etc. Naonal Facilies, staons and links Google General Transit Feed Specificaon (GTFS) Data Various transit operators Service informaon can be derived from staon informaon Varies by agency Numerous U.S. and Canadian transit systems Transit staon Public Transportaon Factbook (PTFB) American Public Transportaon Associaon (APTA) (non-profit transit agency industry organizaon) Public transit revenue miles and hours 2010 (2008 data), and annually (2003-2010 available online). Historical tables date to as early as 1902 depending on stasc. Most figures available for at least 10 years. Naonal (and some Canadian data) Transit agency Naonal Transit Database (NTD) U.S. DOT, Federal Transit Administraon (FTA) Transit vehicle revenue miles and hours and vehicle counts Annual summaries available from 1996 to 2009. Time- series data files contain agency summaries from 1991 to 2009. Naonal Transit agency 2b. Passenger Demands TOD Database Center for Transit-Oriented Development; Center for Neighborhood Technology Demographics surrounding fixed- guideway transit staons 2000 (employment 2002- 2008) Naonwide fixed- guideway transit system staon locaons Fixed-guideway transit staons and surrounding area Longitudinal Employer- Household Dynamics (LEHD) U.S. Census Bureau Employee home and work OD pairs 2002-2008 47 States Census Block Census Transportaon Planning Package (CTPP) U.S. Census Bureau / American Associaon of State Highway and Transportaon Officials (AASHTO) Origin-desnaon tables of employees Decennial Census: 1990, 2000. ACS: 2006-2008 Naonal Census Block Group

J-3 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit Public Transportaon Factbook (PTFB) American Public Transportaon Associaon (APTA) (non-profit transit agency industry organizaon) Unlinked transit trip summaries and passenger miles 2010 (2008 data), and annually (2003-2010 available online). Historical tables date to as early as 1902 depending on stasc. Most figures available for at least 10 years. Naonal and Canada Transit agency Naonal Transit Database (NTD) U.S. DOT, Federal Transit Administraon (FTA) Unlinked passenger trips and passenger miles traveled Annual summaries available from 1996 to 2009. Time- series data files contain agency summaries from 1991 to 2009. Naonal Transit agency Naonal Household Travel Survey (NHTS) and Naonwide Personal Transportaon Survey (NPTS) U.S. Department of Transportaon (DOT), Bureau of Transportaon Stascs (BTS), Federal Highway Administraon (FHWA) Comprehensive travel survey NHTS: 2009, 2001-2002, 2009; NPTS: 1995, 1990, 1983, 1977, 1969 Naonal Census Block Group & Household 2c. Revenues and Cross-Subsidies Public Transportaon Factbook (PTFB) American Public Transportaon Associaon (APTA) Fare collecon summaries 2010 (2008 data), and annually (2003-2010 available online). Historical tables date to as early as 1902 depending on stasc. Most figures available for at least 10 years. Naonal (and some Canadian data) Transit agency Naonal Transit Database (NTD) U.S. DOT, Federal Transit Administraon (FTA) Revenue by mode and service type, public resources for cross- subsidies Annual summaries available from 1996 to 2009. Time- series data files contain agency summaries from 1991 to 2009. Naonal Transit agency 3. Social Characteriscs 3a. Transportaon and Housing Affordability Housing +Transportaon (H+T) Affordability Index Center for Neighborhood Technology Housing and transportaon affordability indices 2008, 2000 Naonal (337 MSAs) Census Block Group

J-4 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit American Community Survey (ACS) U.S. Census Bureau Includes income and housing costs 1996-2009 Naonal Census Place American Housing Survey (AHS) U.S. Department of Housing and Urban Development (HUD) Includes income and housing costs 1973-2009 Naonal Census Tract ESRI Updated Demographics ESRI Esmates income and home valuaons Updated annually Naonal Census Block Group Regional, State, and City House Price Index (HPI) Data Federal Housing Finance Age (public) 1975-Present Year, Quarterly Naonal State & MSAs 3b. Public Health and Safety American Housing Survey (AHS) U.S. Department of Housing and Urban Development (HUD) Housing with lead paint informaon 1973-2009 Naonal Census Tract Public Transportaon Factbook (PTFB) American Public Transportaon Associaon (APTA) Includes transit fuel and energy use 2010 (2008 data), and annually (2003-2010 available online). Historical tables date to as early as 1902 depending on stasc. Most figures available for at least 10 years. Naonal (and some Canadian data) Transit agency Behavioral Risk Factor Surveillance System (BRFSS) U.S. Centers for Disease Control and Preven
on (CDC) Various indicators of public health Annual Survey (1984-Present Year); ArcGIS (2002-Present Year) Na
onal State & some MSAs Na
onal Health Interview Survey (NHIS) U.S. Na
onal Center for Health Sta
s
cs (NCHS), Centers for Disease Control and Preven
on (CDC) Survey of health condi
ons 1963-2009 Na
onal Households Fatality Analysis Repor
ng System (FARS) and Na
onal Automo
ve Sampling System General Es
mates System (NASS GES) U.S. Department of Transporta
on (DOT), Na
onal Highway Traffic Safety Administra
on (NHTSA) Traffic related fatali
es Annual 1975-Present (FARS) & 1998-Present Year (NHTSA) Na
onal County City & Class Trafficway (FARS), Geographic Region

J-5 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit 3c. Socioeconomic Diversity TOD Database Center for Transit-Oriented Development; Center for Neighborhood Technology Includes demographic informaon near transit staons 2000 (employment 2002- 2008) Naonwide fixed- guideway transit system staon locaons Transit staon and surrounding area Decennial Populaon and Housing Census U.S. Census Bureau Most complete source of demographic informaon 2000, 1990, 1980, etc. Naonal Census Block American Community Survey (ACS) U.S. Census Bureau Detailed populaon informaon at coarse units of analysis 1996-2009 Naonal Census Place GeoLycs 2001-2008 Demographic Data GeoLycs Annual demographic and socioeconomic informaon 2001-2008 Naonal Census Block Group ESRI Updated Demographics ESRI Esmates of demographic and socioeconomic characteriscs Updated annually Naonal Census Block Group 3d. Geographic Accessibility TOD Database Center for Transit-Oriented Development, Center for Neighborhood Technology Land use near staons indicates quanty of transit accessible land uses 2000 (employment 2002- 2008) Naonwide fixed- guideway transit system staon locaons Transit staon and surrounding area Housing +Transportaon (H+T) Affordability Index Center for Neighborhood Technology Includes measures of auto and transit usage 2008, 2000 Naonal (337 MSAs) Census Block Group Longitudinal Employer- Household Dynamics (LEHD) U.S. Census Bureau Employee home and work OD pairs 2002-2008 47 States Census Block

J-6 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit Naonal Dataset for Locaon Sustainability and Urban Form (5Ds & SLIs) Natural Resource Ecology Laboratory, Colorado State University Accessibility measures for auto and transit travel 2009 Naonal Census Block Group Census Transportaon Planning Package (CTPP) U.S. Census Bureau / American Associaon of State Highway and Transportaon Officials (AASHTO) Includes origin- desnaon informaon for U.S. workers and related travel mes Decennial Census: 1990, 2000. ACS: 2006-2008 Naonal Census Block Group Google General Transit Feed Specificaon (GTFS) Data Various transit operators GTFS informaon is the building block for transit trip roung Varies by agency Numerous U.S. and Canadian transit systems Staon and transit route Naonal Household Travel Survey (NHTS) and Naonwide Personal Transportaon Survey (NPTS) U.S. Department of Transportaon (DOT), Bureau of Transportaon Stascs (BTS), Federal Highway Administraon (FHWA) Comprehensive travel survey NHTS: 2009, 2001-2002, 2009; NPTS: 1995, 1990, 1983, 1977, 1969 Naonal Census Block Group & Household 4. GIS and Network Naonal Transportaon Atlas Database (NTAD) U.S. DOT, Bureau of Transportaon Stascs (BTS), Research and Innovave Technology Administraon (RITA), Includes GIS layers of transit staons and facilies 2010 (online or DVD), 2009 (DVD), 2008 (DVD) Naonal Facilies, staons and links Google General Transit Feed Specificaon (GTFS) Data Various transit operators Includes lat/long locaons of transit staons and service informaon Varies by agency Numerous U.S. and Canadian transit systems Staon and transit route U.S. Census Topologically Integrated Geographic Encoding and Referencing system (TIGER/Line) Shapefiles U.S. Census Bureau Census zones, streets, and other geographic features 2010, 2009, 2008, 2007, 2000, etc. Naonal Census block, streets, and point places

J-7 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit ESRI Updated Demographics ESRI Up-to-date demographic and household economic condion esmates Updated annually Naonal Census Block Group 5. Intermodal Characteris
cs 5a. Urban Mobility on Roadway Urban Mobility Report (UMR) Texas Transportaon Instute (TTI) Regional traffic congeson measurements 1982-2010 Naonal (selected MSAs) Metropolitan region Highway Stascs U.S. Department of Transportaon (DOT), Federal Highway Administraon (FHWA) Quanfies vehicle- miles of travel 1992-2009 Naonal Urbanized Areas 5b. Modal Compeveness Naonal Dataset for Locaon Sustainability and Urban Form (5Ds & SLIs) Natural Resource Ecology Laboratory, Colorado State University (Academic Instute) Accessibility measures for auto and transit 2009 Naonal Census Block Group Census Transportaon Planning Package (CTPP) U.S. Census Bureau / American Associaon of State Highway and Transportaon Officials (AASHTO) Includes mode and travel me for work trips Decennial Census: 1990, 2000. ACS: 2006-2008 Naonal Census Block Group Naonal Household Travel Survey (NHTS) and Naonwide Personal Transportaon Survey (NPTS) U.S. Department of Transportaon (DOT), Bureau of Transportaon Stascs (BTS), Federal Highway Administraon (FHWA) Mode and travel mes for household travel NHTS: 2009, 2001-2002, 2009; NPTS: 1995, 1990, 1983, 1977, 1969 Naonal Census Block Group & Household Urban Mobility Report (UMR) Texas Transportaon Instute (TTI) Indicaon of regional auto travel inconvenience 1982-2010 Naonal (selected MSAs) Metropolitan region 5c. Intermodal Connecvity Intermodal Passenger Connecvity Database Research and Innovave Technology Administraon (RITA)/Bureau of Transportaon Stascs (BTS) Intermodal facilies and relevant informaon 2011 Naonal Census Block Group and Facilies

J-8 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit Naonal Transportaon Atlas Database (NTAD) U.S. DOT, Bureau of Transportaon Stascs (BTS), Research and Innovave Technology Administraon (RITA), Intermodal facilies are included among other transportaon facilies 2010 (online or DVD), 2009 (DVD), 2008 (DVD) Naonal Facilies, staons and links 5d. Local Access Availability Google General Transit Feed Specificaon (GTFS) Data Various transit operators GTFS feeds are available for some U.S. BRT systems Varies by agency Numerous U.S. and Canadian transit systems Staon and transit route 6. Bus Rapid Transit (BRT) Naonal Transportaon Atlas Database (NTAD) U.S. DOT, Bureau of Transportaon Stascs (BTS), Research and Innovave Technology Administraon (RITA), Includes fixed- guideway transit facilies 2010 (online or DVD), 2009 (DVD), 2008 (DVD) Naonal Facilies, staons and links Google General Transit Feed Specificaon (GTFS) Data Various transit operators GTFS feeds are available for some U.S. BRT systems Varies by agency Numerous U.S. and Canadian transit systems Staon and transit route 7. Parking North America Central Business District Parking Rate Survey Colliers Internaonal Survey of center city parking prices in major MSAs 2001-2010 Naonal, selected cies Central business district “Parking in America” report Naonal Parking Associaon Survey of center city parking prices in major MSAs 2008-2010 Naonal, selected cies Central business district

J-9 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit 8. Urban Design 8a. Street Connecvity Naonal Dataset for Locaon Sustainability and Urban Form (5Ds & SLIs) Natural Resource Ecology Laboratory, Colorado State University Intersecon density 2009 Naonal Census Block Group Naonal Transportaon Atlas Database (NTAD) U.S. DOT, Bureau of Transportaon Stascs (BTS), Research and Innovave Technology Administraon (RITA), Street network GIS Shapefile 2010 (online or DVD), 2009 (DVD), 2008 (DVD) Naonal Street 9. Urban Development 9a. Residenal Locaon TOD Database Center for Transit-Oriented Development, Center for Neighborhood Technology Residenal occupaon in areas near transit staons 2000 (employment 2002- 2008) Naonwide fixed- guideway transit system staon locaons Transit staon and surrounding area Decennial Populaon and Housing Census U.S. Census Bureau Residenal demographic informaon 2000, 1990, 1980, etc. Naonal Census Block American Community Survey (ACS) U.S. Census Bureau Detailed populaon informaon at coarse units of analysis 1996-2009 Naonal Census Place GeoLycs 2001-2008 Demographic Data GeoLycs Private demographic data source 2001-2008 Naonal Census Block Group ESRI Updated Demographics ESRI Esmates of demographic and socioeconomic characteriscs Updated annually Naonal Census Block Group

J-10 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit 9b. Business Locaon TOD Database Center for Transit-Oriented Development, Center for Neighborhood Technology Employment indicates business acvity near transit staons 2000 (employment 2002- 2008) Naonwide fixed- guideway transit system staon locaons Transit staon and surrounding area County; Metro; & ZIP Code Business Paerns (CBP; MBP; & ZBP) U.S. Census Bureau Number of establishments, payroll, and employee counts 1986-2008 Na„onal ZIP Code (from 1994) Longitudinal Employer- Household Dynamics (LEHD) U.S. Census Bureau Employees, age, earnings, and industry by work loca„on 2002-2008 47 States Census Block Economic Census U.S. Census Bureau Business ac„vity and trade by geography 2007, 2002, 1997 Na„onal ZIP Code 9c. Mul„ple Integra„on Na„onal Dataset for Loca„on Sustainability and Urban Form (5Ds & SLIs) Natural Resource Ecology Laboratory, Colorado State University Includes various informa„on on factors theore„cally related to residen„al loca„on and transport mode share 2009 Na„onal Census Block Group TOD Database Center for Transit-Oriented Development, Center for Neighborhood Technology Employment indicates business ac„vity near transit sta„ons 2000 (employment 2002- 2008) Na„onwide fixed- guideway transit system sta„on loca„ons Transit sta„on and surrounding area

J-11 Aribute Source Provider Measure / Predictor Year Coverage Smallest Unit 9d. Property Transacon RealQuest Professional The FirstAmerica CoreLogic All property transacon prices and other a
ributes Long-term (custom order) Naonal Transacon Address DataQuick DataQuick All property transac on prices and other a ributes Long-term (custom order) Na onal Transac on Address Zillow.com Zillow, Inc. Housing property transac on prices and other a ributes Last few years Na onal Transac on Address HUD Aggregated USPS Administra ve Data On Address Vacancies U.S. Department of Housing and Urban Development (HUD) Business, residen al, and other property vacancy rates and absorp on days Quarterly Dec.2005 to Sep.2010. Na onal Census Tract Regional, State, and City House Price Index (HPI) Data Federal Housing Finance Agency (FHFA) Standardized housing price data 1975-Present Year, Quarterly Na onal State & MSAs

Abbreviations and acronyms used without definitions in TRB publications: A4A Airlines for America AAAE American Association of Airport Executives AASHO American Association of State Highway Officials AASHTO American Association of State Highway and Transportation Officials ACI–NA Airports Council International–North America ACRP Airport Cooperative Research Program ADA Americans with Disabilities Act APTA American Public Transportation Association ASCE American Society of Civil Engineers ASME American Society of Mechanical Engineers ASTM American Society for Testing and Materials ATA American Trucking Associations CTAA Community Transportation Association of America CTBSSP Commercial Truck and Bus Safety Synthesis Program DHS Department of Homeland Security DOE Department of Energy EPA Environmental Protection Agency FAA Federal Aviation Administration FHWA Federal Highway Administration FMCSA Federal Motor Carrier Safety Administration FRA Federal Railroad Administration FTA Federal Transit Administration HMCRP Hazardous Materials Cooperative Research Program IEEE Institute of Electrical and Electronics Engineers ISTEA Intermodal Surface Transportation Efficiency Act of 1991 ITE Institute of Transportation Engineers MAP-21 Moving Ahead for Progress in the 21st Century Act (2012) NASA National Aeronautics and Space Administration NASAO National Association of State Aviation Officials NCFRP National Cooperative Freight Research Program NCHRP National Cooperative Highway Research Program NHTSA National Highway Traffic Safety Administration NTSB National Transportation Safety Board PHMSA Pipeline and Hazardous Materials Safety Administration RITA Research and Innovative Technology Administration SAE Society of Automotive Engineers SAFETEA-LU Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (2005) TCRP Transit Cooperative Research Program TEA-21 Transportation Equity Act for the 21st Century (1998) TRB Transportation Research Board TSA Transportation Security Administration U.S.DOT United States Department of Transportation