NSPA-2007_Report.pdf

US Department of Transportation Federal Transit Administration

PREDICTED AND ACTUAL IMPACTS OF

NEW STARTS PROJECTS - 2007

CAPITAL COST AND RIDERSHIP

DRAFT

Prepared by:

FTA Office of Planning and Environment with support from

Vanasse Hangen Brustlin, Inc.

April 2008

Acknowledgements

This report was primarily authored by Mr. Steven Lewis-Workman of the Federal Transit

Administration and Mr. Bryon White of VHB, Inc. Portions of this report were also written and edited by Ms. Stephanie McVey of the Federal Transit Administration and Mr. Frank Spielberg of VHB, Inc. The authors would like to thank all of the project sponsors and FTA Regional

Office staff who took the time to review and ensure the accuracy of the information contained in this study.

Table of Contents

1. OVERVIEW 1

1.1. REVIEW OF PAST STUDIES 2

1.2. METHODOLOGY 2

1.3. FINDINGS FOR CAPITAL COSTS 4

1.4. FINDINGS FOR RIDERSHIP 4

1.5. ORGANIZATION OF THIS REPORT 4

2. CAPITAL COSTS 7

2.1. CAPITAL COST ANALYSIS APPROACH 7

2.2. CAPITAL COST ANALYSIS RESULTS 8

2.3. COMPARISON TO NEW STARTS PROJECTS FROM PRIOR STUDIES 15

2.4. DURATION OF PROJECT DEVELOPMENT 16

3. RIDERSHIP 19

3.1. RIDERSHIP ANALYSIS APPROACH 19

3.2. FORECAST AND ACTUAL RIDERSHIP 20

3.2.1. AVERAGE WEEKDAY BOARDINGS 21

3.2.2. AVERAGE WEEKDAY BOARDINGS ADJUSTED TO FORECAST YEAR 22

3.3. COMPARISON TO NEW STARTS PROJECTS FROM PRIOR STUDIES 24

3.3.1. PREDICTED VS. ACTUAL – 2003 UPDATE 24

3.3.2. URBAN RAIL TRANSIT PROJECTS – 1990 UPDATE 25

3.4. KEY FINDINGS 26

APPENDIX: PROJECT PROFILES 32

BALTIMORE CENTRAL LRT DOUBLE-TRACKING 33

SOUTH BOSTON PIERS TRANSITWAY – PHASE 1 39

CHICAGO CTA DOUGLAS BRANCH RECONSTRUCTION 45

CHICAGO METRA NORTH CENTRAL CORRIDOR COMMUTER RAIL 49

CHICAGO METRA SOUTHWEST CORRIDOR COMMUTER RAIL 55

CHICAGO METRA UNION PACIFIC WEST LINE EXTENSION 61

DALLAS NORTH CENTRAL LIGHT RAIL EXTENSION 65

DENVER SOUTHEAST CORRIDOR LRT 71

FT. LAUDERDALE – MIAMI TRI-RAIL DOUBLE TRACKING 77

MEMPHIS MEDICAL CENTER RAIL EXTENSION 83

MINNEAPOLIS HIAWATHA CORRIDOR LRT 89

NEWARK-ELIZABETH RAIL LINK MOS-1 94

NEW JERSEY TRANSIT – HUDSON BERGEN LIGHT RAIL MOS I & II 101

PITTSBURGH STAGE II LIGHT RAIL RECONSTRUCTION 111

PORTLAND – INTERSTATE MAX LRT 117

SACRAMENTO SOUTH LRT PHASE 1 122

SAN DIEGO – MISSION VALLEY EAST LRT 127

SAN FRANCISCO - BART TO SFO 133

SAN JUAN – TREN URBANO 139

SALT LAKE CITY UNIVERSITY/MEDICAL CENTER EXTENSIONS 145

WASHINGTON METRO LARGO EXTENSION 151

Abbreviations

AA alternatives analysis

BRT bus rapid transit

CM construction management

CR commuter rail

DB design-build

DBB design-bid-build

DBOM design-build-operate-maintain

DEIS Draft Environmental Impact Statement

DPM Downtown People Mover

EA Environmental Assessment

Ext extension

FD final design

FEIS Final Environmental Impact Statement

FFGA Full Funding Grant Agreement

FTA Federal Transit Administration

HRT heavy rail transit

ITP Independent Testing Program

LF linear feet

LPA locally preferred alternative

LRT light rail transit

MOS minimum operable segment

MPO metropolitan planning organization

MIS Major Investment Study

NTD National Transit Database

NTP Notice to Proceed

PE preliminary engineering

PMP Project Management Plan

PMOC Project Management Oversight Contractor

ROD Record of Decision

ROW right-of-way

UMTA Urban Mass Transportation Administration

US Department of Transportation Page 1 of 155 Federal Transit Administration February 2008

1. OVERVIEW

The Federal Transit Administration’s (FTA) Office of Planning and Environment has conducted an analysis of the predicted and actual impacts of 21 recently opened major transit projects that have been constructed using funds under the New Starts program (49 USC 5309 et al). This report builds on a prior study by the Urban Mass Transportation Administration (UMTA –

FTA’s previous name) in 19901 and a more recent effort that FTA completed in 20032 to analyze the projects that have opened for revenue service between 1990 and 2002. With the publication of this report, FTA has completed a comprehensive analysis of the predicted and actual impacts of almost 50 New Starts projects built to date.

The analysis has three main purposes:

1. To provide an up-to-date assessment of the actual performance of projects compared to the forecasts made for those projects;

2. To consider the effectiveness of the procedures and technical methods used to develop information for decision-making by project sponsors and the FTA; and

3. To “fill the gap” between projects in the previous 2003 study and future projects which will all be subject to the New Starts Before and After Studies requirement [49 USC

5309(g)(2)(C)].

FTA based this analysis on an inventory of ridership forecasts and cost estimates prepared at various stages of the project planning and development process. The data sources included environmental documents, alternatives analyses (AA), Major Investment Studies (MIS), New

Starts application submissions, Full Funding Grant Agreements (FFGA), and Project

Management Oversight Contractor (PMOC) reports. This information was then compared to the actual results reported by the project sponsors for ridership and by the PMOCs for capital costs.

This analysis provides information about the predicted vs. actual performance of New Starts projects. While the analysis provides some limited insight into the performance of the forecasts prepared for major transit investments projects, it does not attempt to perform detailed forensics on major forecasting errors. Currently the ability to determine the cause of such errors is limited due to inadequate background documentation. In the future, Before and After Studies, which are now required by FTA of all projects that receive FFGAs, will provide information that will enable project sponsors and FTA to perform more detailed analyses of the causes of errors and notable successes. Preparing a Before and After Study will require project sponsors to archive detailed planning, engineering and ridership forecasting information that was not generally preserved for the projects covered in this and earlier studies.

1 Pickrell, Don H., Urban Rail Transit Projects: Forecast Versus Actual Ridership and Costs, DOT-T-91-04, Office of Grants Management, Urban Mass Transportation Administration, Washington, DC, October 1990.

2 This study was released in 2007 as an appendix to FTA’s Contractor Performance Assessment Report (2007) which can be accessed at http://www.fta.dot.gov/documents/CPAR_Final_Report_-_2007.pdf.

http://www.fta.dot.gov/ http://www.fta.dot.gov/about/offices/about_FTA_146.html#tpe http://www.fta.dot.gov/newstarts

US Department of Transportation Page 2 of 155

1.1. Review of Past Studies

UMTA first published an analysis of the predicted and actual impacts of 10 major capital transit projects in 1990. This study found that when ridership forecasts were compared to actual ridership, none of the 10 projects examined had achieved, at the time of the analysis, ridership greater than 72 percent of their forecasts. Nine of the 10 projects had achieved less than 50 percent of their forecasts. Capital cost estimates were also generally poor. Two projects were within 20 percent of the original cost estimate (adjusting for inflation), seven of 10 projects were between 30 and 100 percent higher than their original estimates, and one project was more than double (over 100 percent of) its cost estimate. On average, projects exceeded their inflation-adjusted cost estimates by about 50 percent.

FTA prepared a new study in 2003 examining 19 additional projects that had been completed between 1990 and 2002. FTA found that cost estimates had improved markedly since the 1990 report, but still systematically underestimated actual project costs. The actual capital cost of

New Starts projects were 20.9 percent greater, on average, than the inflation-adjusted estimate prepared during alternatives analysis, 13.5 percent greater than the estimate prepared before entering final design, and 7.3 percent greater than the FFGA cost estimate.

Ridership forecasts had also improved since the 1990 study with a number of projects’ actual ridership close to and even higher than predicted. However, New Starts project sponsors still systematically overestimated the actual ridership achieved by their projects. The results indicated that, as of 2002, three projects exceeded their initial ridership forecasts; three other projects exceeded 80 percent of their initial ridership forecasts. All told, eight of the 19 projects included in that study either achieved, or had a good chance of coming within a reasonable range

(±20 percent) of their initial planning level ridership forecasts while the other 11 projects remained well below their predicted ridership levels.

1.2. Methodology

For this current study (as well as the 2003 study), project scope, service levels, costs, and ridership were documented at four milestones:

1. Selection of the locally preferred alternative (LPA) or entry into preliminary engineering

(PE), usually signified by the completion of the AA and Draft Environmental Impact

Statement (DEIS);

2. Entry into final design, usually signified by the completion of a Final Environmental

Impact Statement (FEIS);

3. Signing of the FFGA; and

4. Project completion, for capital cost data, and two years after opening (if available) for ridership results.

US Department of Transportation Page 3 of 155

Data recorded for each milestone were:

Project scope (length, horizontal/vertical alignment, stations, parking, vehicles, ancillary facilities);

Service levels (headways and service hours);

Capital cost (construction, right-of-way, engineering, administration, vehicles, etc.);

Operating and maintenance (O&M) cost; and

Ridership (average weekday boardings on the project and system-wide boardings and/or unlinked trips).

The projections at each stage of planning and project development were taken from the various reports prepared by project sponsors and their consultants. These documents were collected from archived material at FTA or requested from the project sponsors. The FTA Office of

Planning and Environment and Regional Office staff collected the ridership data and prepared the ridership analysis, while FTA’s engineering consultant, VHB Inc., assisted FTA in preparing the analysis of capital costs. In most cases, the information in the published documents was not sufficient to use to examine specific sources of error in the cost estimates or ridership forecasts.

FTA and the contractor team condensed the relevant information that was collected for each project into Project Profiles. For each project, the Profiles contain a map and description;

summarize the planning and development; and document the predicted and actual scope, service levels, ridership, capital costs, and operating costs. These profiles were then sent to FTA

Regional Offices and to each project sponsor for review and validation. The information in the

Project Profiles, including the forecasts and actual data used for this analysis, has been reviewed for accuracy and validated by each project sponsor. In some cases, project sponsors provided some explanatory analyses of the source of cost overruns and ridership forecasting errors which are also included in the Project Profiles (see the Appendix).

No attempt was made to adjust costs or ridership estimates to reflect changes in the scope of a project. However, when changes in scope that would have a significant impact on expected ridership or capital costs were identified, those changes were noted in the text of the Project

Profiles. In three cases – the Portland Interstate MAX, Salt Lake City’s University and Medical

Center Ext., and Pittsburgh Stage II Reconstruction projects – the scope of the projects changed to such a large extent between AA and construction that including them in the summary analysis would significantly distort the results. Therefore, these projects were excluded from certain portions of the cost analysis. In cases where the actual project constructed was closer to an identifiable alternative included in the planning studies, actual costs and ridership were compared to forecasts from the alternative that most closely resembled the constructed project.

FTA’s previous study prepared in 2003 included a comparison of predicted and actual operating and maintenance (O&M) costs. Since all but two projects in the current study are extensions to existing systems, determining the actual operating costs for specific segments of a larger system is extremely difficult. FTA documented the O&M cost estimates for the projects in the current study and, from National Transit Database (NTD) data, attempted to allocate costs to the portion http://www.fta.dot.gov/regional_offices.html http://www.vhb.com/ http://www.ntdprogram.gov/

US Department of Transportation Page 4 of 155 of the total system that comprised each of the projects. In FTA’s view, this method did not produce meaningful results; therefore, the summary analysis of predicted vs. actual O&M costs has not been prepared for this study. However, the results of the O&M cost analysis for each project remain in the Project Profiles in the Appendix.

1.3. Findings for Capital Costs

On average, for the 21 projects completed between 2003 and 2007 actual construction costs exceeded the inflation-adjusted estimates developed in alternatives analysis by 40.2 percent, the final design entry cost estimates by 11.8 percent, and the FFGA estimates by

6.2 percent.

The sum of all the New Starts projects’ actual capital costs exceeded the sum of their original inflation-adjusted FFGA amounts by $1.54 billion. Over 60 percent ($948 million) of this increase was due to a single project (i.e. Tren Urbano). At the same time, 11 of the 21 projects had actual capital costs that were less than the FFGA inflation-adjusted estimate.

The average error in cost estimates for the projects in the current study is higher than the average error for the projects in the 2003 study. The cost estimates for the projects in both of the recent studies are more accurate than was found in the 1990 UMTA study.

For projects opened between 2003 and 2007, the average time between choosing the LPA and opening for revenue service was about 7.9 years. The length of time between selection of the LPA and project opening has not changed appreciably since 1990.

1.4. Findings for Ridership

FTA expects that eight out of 18 projects completed between 2003 and 2007, for which ridership data was available, will have actual ridership in excess of 80 percent of the forecast ridership developed in alternatives analysis.3 The same eight projects are expected to exceed 80 percent of the ridership forecasts developed before entering final design.

The projects in this study are expected to carry, on average, 74.5 percent of their alternatives analysis forecast ridership and 72.2 percent of the forecasts prepared before entering final design.

Ridership forecasts are not significantly better than the forecasts in FTA’s previous 2003 study, but forecast for the projects in both studies are significantly more accurate than the forecasts prepared for the projects in the 1990 UMTA study.

1.5. Organization of this Report

This report presents the predicted and actual impacts of 234 transit New Starts projects opened for full service between 2002 and 2007. These projects are listed in Table 1 below.

3 The most recent observed ridership figure was adjusted to the forecast year by the average annual rate of system-wide ridership growth observed over the period between 1996 and 2006.

4 FTA combined New Jersey Transit’s Hudson Bergen MOS 1 and MOS 2 projects as well as the Utah Transit

Authority’s University and Medical Center Extensions into single projects in the summary analysis because these

US Department of Transportation Page 5 of 155

Table 1: Projects Included in this Study, listed alphabetically by city name

City Project

AA/DEIS/MIS/EA

Year

FEIS

Year

FFGA

Year

Opening

Year

Forecast

Year

Baltimore Central LRT Double Tracking* 2000 NA 2001 2006 2020

Boston South Boston Piers Phase 1 1992 1993 1994 2004 2010

Chicago Metra UP West* 1998 NA 2001 2006 2020

Chicago Metra North Central* 1998 NA 2001 2006 2020

Chicago Metra Southwest* 1998 NA 2001 2006 2020

Chicago Douglas Branch Reconstruction* 2000 NA 2001 2005 2020

Dallas North Central LRT Extension 1996 1997 1999 2002 2010

Denver Southeast LRT 1997 1999 2000 2006 2020

Memphis Medical Center Extension* 1997 NA 2000 2004 2020

Miami South Florida Tri-Rail Upgrades 1998 1999 2000 2007 2015

Minneapolis Hiawatha LRT 1982 1985/1999 2001 2004 2020

Newark Newark Elizabeth MOS 1 1997 1998 2000 2006 2015

Northern

New Jersey Hudson Bergen MOS 1 & MOS 2** 1992 1996 1996/2000 2000-2006 2010

Pittsburgh Stage II LRT Reconstruction* 1996 NA 2001 2004 2005

Portland Interstate MAX LRT 1998 1999 2000 2004 2015

Sacramento South LRT Phase 1 1996 1997 1997 2003 2015

Salt Lake

City University & Medical Center Extensions** 1997 1999 2000/2001 2001/2003 2020

San Diego Mission Valley East LRT 1997 1998 2000 2005 2015

San

Francisco BART to SFO 1995 1996 1997 2003 2010

San Juan Tren Urbano 1995 1995 1996 2005 2010

Washington Largo Extension 1996 1999 2000 2004 2020

* These projects performed a single Environmental Assessment or were categorical exclusions.

** The Hudson Bergen projects and Salt Lake City projects represent four distinct FFGAs. In each case, they were planned and developed as single projects but were later divided for construction. These projects are considered single projects in this analysis.

The body of this report addresses the findings that have been gleaned from the project data. The details of any specific project are referenced only to illustrate points of interest. The following sections discuss the summary results for capital costs and ridership. Project Profiles in the

Appendix – one for each project – include more detailed information on each project’s development history, the scope of the project as conceived and executed and other information necessary to interpret the summary statistics.

individual projects were segments of larger projects. For this reason, 23 individual FFGA projects are listed as 21 projects.

US Department of Transportation Page 7 of 160

2. CAPITAL COSTS

This chapter presents the predicted and actual capital costs for FTA-funded New Starts transit projects that have fully opened for revenue service between 2002 and 2007. Reliable information about project costs is vital to the decision-making processes that determine transportation investment priorities at both the local and national levels. Local decision-makers need reliable information about the likely costs and benefits of proposed projects before they can make reasonable decisions to choose locally preferred alternatives. FTA also needs reliable estimates of costs and benefits before advancing projects in the project development process and before making funding decisions.

Cost estimates developed during an Alternatives Analysis study are used to support the local decision to choose the LPA and are generally the cost estimates that are presented to FTA when projects apply to enter preliminary engineering. The decision to adopt an LPA signifies that the local decision-makers have chosen the specific mode (highway, LRT, BRT, etc.) and general alignment of a project to address the identified problems and needs in a corridor. FTA considers this decision, made at the end of alternatives analysis, to be the most critical decision in the planning and project development process.

Cost estimates that were used to support the decision whether to allow the project to advance to final design are developed during PE and usually presented in the FEIS or EA. A final cost estimate is developed just before signing an FFGA and supports FTA’s final decision to proceed, prior to construction. Occasionally, costs change significantly during construction, necessitating an amended FFGA. The summary analysis contained in this chapter focuses on the original

FFGA because the amended FFGA does not support a major decision by the project sponsor or

FTA. The Project Profiles document any amendments that were incorporated after the start of construction.

2.1. Capital Cost Analysis Approach

The capital costs are compared in both year-of-expenditure (YOE) and inflation-adjusted (i.e.

constant) dollars in order to ascertain whether the cost estimates developed in the planning and development of New Starts projects have accurately predicted the actual costs of constructing them. FTA considers the comparison of predicted vs. actual costs in inflation-adjusted dollars to be the most illuminating comparison because inflation is not particularly relevant in choosing alternatives since the costs (and revenues) of all alternatives are affected by inflation. For this reason, the analysis focuses on the quality of the cost estimates in inflation-adjusted dollars

(using industry-accepted published inflation rates). For purposes of comparing estimated constant dollar costs to actual costs, the cost estimates were adjusted to midpoint-of-construction year dollars. While not a perfect comparison to actual costs, year-by-year expenditures were generally not available; therefore, the approximation of using the midpoint-of-construction year dollars was adopted.

For most of the projects, the available documents did not provide the level of detail that would be required to determine specific sources of cost estimating errors. Future projects that are subject

US Department of Transportation Page 8 of 160 to the Before and After Study requirement (i.e. FFGAs awarded after 2001) will be required to archive information in enough detail to assess the accuracy of the cost estimates over time and document the sources of error.

For each phase of project development, constant dollar capital costs were collected from the planning documents. Planning documents also generally include capital cost estimates in escalated dollars (either YOE or midpoint-of-construction year dollars) to reflect the expected impact of inflation before and during construction. However, for this analysis, the constant dollar costs are escalated by the actual annual inflation rates from the year in which the constant dollar costs are expressed to midpoint-of-construction year dollars. In order to accurately escalate to midpoint-of-construction year values, annual inflation factors were determined for each of the four components of the constant dollar cost estimates:

Construction Costs

Rolling Stock

Right-of-Way (ROW)

Design/Project Management/Other Professional Services

Each component was inflated using appropriate indices. The Engineering News Record

Construction Cost Index (20-city average) was used to determine the annual inflation rate for construction costs. The Producer Price Index’s “metals and metal products” sub-group, also from BLS, was used to determine the inflation rate for rolling stock. The national House Price

Index from the Office of Federal Housing Enterprise Oversight was used to inflate right-of-way costs. All costs associated with design, administration, project management or contingency were inflated using BLS’s Employment Cost Index’s private-sector professional sub-group.

No attempt was made to adjust costs to reflect changes in the scope of a project though, in three cases – Portland Interstate MAX LRT, Salt Lake City’s University and Medical Center Ext., and the Pittsburgh Stage II Reconstruction – the projects changed so much between planning and construction that cost estimates during planning and project development had to be excluded from the analysis. Significant changes in scope that could have had a significant impact on capital or operating costs were noted in the Project Profiles found in the Appendix. In cases where the actual project constructed was closer to an identifiable alternative included in the planning studies, actual costs and ridership were compared to forecasts from the alternative that most closely resembled the constructed project.

2.2. Capital Cost Analysis Results

Individual project costs (also referred to as capital costs) are shown in Table 2, along with each project’s estimated capital cost at three other project development stages – PE entry

(AA/DEIS/MIS completion), Final Design entry (FEIS/EA completion), and the FFGA. All the estimated costs are in inflation-adjusted dollars. If a data point in Table 2 indicates “NA,” there was no document available for that particular project.

http://enr.construction.com/features/conEco/default.asp http://enr.construction.com/features/conEco/default.asp http://www.bls.gov/ppi/home.htm http://www.ofheo.gov/hpi.aspx http://www.ofheo.gov/hpi.aspx http://www.bls.gov/ncs/ect/home.htm

US Department of Transportation Page 9 of 160

Table 2: Capital Costs for each Project, listed by as-built cost

Project Mode

Inflation-Adjusted Capital Cost (in Millions $)

AA/DEIS/MIS

(or PE Entry)

FEIS (or Final Design Entry)

Original

FFGA

Final

FFGA

As Built

Memphis Med Center

LRT***

LRT $36.0 $68.2 $73.3 $73.3 $58.1

Metra UP West CR $98.8 $140.4 $128.1 $128.1 $106.1

Baltimore Central LRT Double-Tracking

LRT $150.5 $150.1 $154.4 $210.0 $151.6

Metra SW Corridor CR $178.7 $217.7 $191.0 $191.0 $185.3

Salt Lake City University/Medical Ext.*

LRT NA $189.1 $204.5 $204.5 $192.1

Newark Rail Link MOS-1 LRT $181.3 $178.3 $215.4 $215.4 $207.7

Metra North Central CR $204.8 $237.0 $224.8 $224.8 $216.8

Sacramento South LRT (Phase 1)

LRT $201.6 $205.1 $219.7 $219.7 $218.6

Portland Interstate MAX LRT Extension*

LRT $803.8 $310.6 $321.5 $321.5 $323.6

Tri-Rail Double Tracking Segment 5

CR NA $330.2 $331.1 $338.8 $345.6

Pittsburgh Stage II Reconstruction*

LRT $400.7 $400.7 $363.2 $363.2 $385.0

Largo Metrorail Extenstion HR $375.0 $432.6 $412.6 $607.2 $426.4

Dallas North Central

LRT***

LRT $332.7 $406.0 $460.8 $460.8 $437.3

Chicago Douglas Branch HR $441.7 $477.7 $473.2 $473.2 $440.8

Mission Valley East LRT Extension

LRT $386.6 $386.6 $426.6 $426.6 $506.2

South Boston Piers Transitway - Phase 1

BRT $398.3 $477.3 $457.4 $600.2 $600.2

Hiawatha Corridor LRT** LRT $243.7 $540.6 $512.9 $708.4 $696.7

Denver Southeast Corridor***

LRT $585.0 $870.4 $867.8 $867.8 $850.8

BART Extension to SFO*** HR $1,193.9 $1,230.0 $1,185.7 $1,483.2 $1,551.6

Hudson-Bergen MOS 1 & 2***

LRT $930.4 $948.5 $1,842.0 $2,172.0 $1,756.2

Tren Urbano HR $1,085.6 $1,309.2 $1,280.6 $1,638.0 $2,228.4

* These projects had significant scope changes that reduced the total length of the projects by more than half between AA and actual construction so the AA/DEIS cost estimates for Portland, Salt Lake City, and Pittsburgh are not included in the summary analysis.

** The Hiawatha LRT project conducted AA in the early 1980s when the cost estimate reported in the AA column was developed and the LRT alternative chosen as the LPA. The project was not pursued until the late 1990s when it finally applied to enter PE.

*** These projects had scope and design changes during project development that had an effect on the as-built costs, but the mode and general alignment of these projects remained consistent throughout project development. These projects are included in the summary analysis.

Estimated capital costs tend to increase throughout the planning and project development process.

As shown in Table 3, the average as-built capital costs are about 40.2 percent higher than the

AA/DEIS (or PE entry) inflation-adjusted estimate, about 11.8 percent over the FEIS (or final design entry) inflation-adjusted estimate, and 6.2 percent over the FFGA inflation-adjusted estimate. In some cases (Portland, Salt Lake City, and Pittsburgh), the scope of the projects was reduced significantly during the planning and project development process so that comparing the

US Department of Transportation Page 10 of 160

AA/DEIS (or entry to PE) cost estimate to actual costs is not an accurate reflection of the quality of the cost estimates.

Table 3 shows that, by the time that the FFGA is executed, the as-built costs are generally close to the costs estimated for the original FFGA. However, there are notable exceptions. There were four projects that cost 30 percent more than estimated in the original FFGA. One small project – Memphis Medical Center Extension – was 20 percent under the FFGA inflation-adjusted budget.

US Department of Transportation Page 11 of 160

Table 3: As-built Capital Costs, as a Percentage of Predictions, listed by as-built cost

Project Transit Mode

As Built Capital Cost, as a percentage of Estimate (adjusted for inflation)

AA/DEIS/MIS

PE Entry

FEIS/EA

FD Entry

Original

FFGA

Memphis Med Center LRT LRT 161.4% 85.2% 79.3%

Metra UP West CR 107.4% 75.6% 82.8%

Baltimore Central LRT Double-Tracking LRT 100.7% 101.0% 98.2%

Metra SW Corridor CR 103.7% 85.1% 97.0%

Salt Lake City University/Medical Ext.3 LRT NA 101.6% 93.9%

Newark Rail Link MOS-1 LRT 114.6% 116.5% 96.4%

Metra North Central CR 105.9% 91.5% 96.4%

Sacramento South LRT (Phase 1) LRT 108.4% 106.6% 99.5%

Interstate MAX LRT Extension2 LRT NA 104.2% 100.7%

Pittsburgh Stage II Reconstruction4 LRT NA NA 106.0%

S. Florida Tri-Rail Double Tracking5 CR NA 104.7% 104.4%

Largo Metrorail Extenstion HR 113.7% 98.6% 103.3%

Dallas North Central LRT6 LRT 131.4% 107.7% 94.9%

Chicago Douglas Branch1 HR 99.8% 92.3% 93.2%

South Boston Piers Transitway - Phase 1 BRT 150.7% 125.7% 131.2%

Mission Valley East LRT Extension LRT 130.9% 130.9% 118.7%

Minneapolis Hiawatha Corridor LRT LRT 285.9% 128.9% 135.8%

Denver Southeast Corridor7 LRT 145.4% 102.9% 103.2%

BART Extension to SFO HR 130.0% 126.1% 130.9%

Hudson-Bergen MOS 1 & 2 LRT 188.8% 185.2% 95.3%

San Juan Tren Urbano HR 205.3% 170.2% 174.0%

Average of 21 projects 140.2% 111.8% 106.2%

NOTES:

1 The Douglas Branch project was a reconstruction of an existing line and did not have a planning study that documented the cost estimates prior to PE entry.

2 The Interstate MAX project scope was significantly reduced during PE so there is no valid comparison possible between the AA/DEIS and the actual project as constructed.

3 The Salt Lake City project scope was significantly reduced during PE so there is no valid comparison possible between the AA/DEIS and the actual project as constructed.

4 The Stage II Reconstruction project scope was significantly reduced during just before the FFGA so there is no valid comparison possible between the earlier estimates and the actual project as constructed. In this case, the scope was reduced because of cost overruns and funding difficulties. The actual cost of the reduced scope project was nearly equal to the planned project that was over twice as long.

5 Significant portions of the Tri-Rail project were already under construction when the project entered PE. There is no specific cost estimate for the scope of the actual FFGA project until this project entered final design.

6 The Dallas project increased in scope during project development by replacing planned single track segments with double track. This increase in scope was most likely responsible for the cost increase between AA and as-built. FTA decided to include the AA/DEIS cost estimate in the analysis because the nature of this scope change differed from the projects that were excluded because of scope changes. The excluded projects experienced major reductions in the length of their alignments while the length of the Dallas project remained fairly consistent.

7 The Denver project experienced scope and design changes in PE due to a major expansion in scope of the highway portion of this multi-modal project.

Table 4 describes the scope changes made in the planning process from the AA/DEIS to the

FFGA juxtaposed with the as-built costs as a percentage of both the AA/DEIS and FFGA.

Table 4: Scope Changes Between DEIS and As-built, listed by as-built cost

Project Major Scope Changes from DEIS to FFGA As Built Cost, as a percentage of

Inflation-Adjusted

US Department of Transportation Page 12 of 160

Estimate

AA/DEIS FFGA

Memphis Med Center LRT Project length decreased from 2.5 miles to 2.0 miles.

Number of stations decreased from 15 to 6.

161.4% 79.3%

Metra UP West Line Extension

Project length increased from 7 miles to 8.5 miles. Number of LRVs decreased from 1 locomotive and 8 cabs to 2 locomotives.

108.9% 82.8%

Baltimore LRT Double- Tracking

No major scope changes. 100.7% 98.2%

Metra SW Corridor New stations increased from 2 to 3. LRVs purchased decreased from 2 locomotives and 13 cabs to 3 locomotives.

103.7% 97.0%

Salt Lake City University/Medical Ext.

Project length reduced significantly; 5 LRVs added to scope. NA 93.9%

Newark Rail Link MOS-1 No major scope changes. 114.6% 96.4%

Metra North Central

Upgraded stations increased from 0 to 17. Double-tracking increased from 12 miles to 16.3 miles, with an additional 2.3 miles of triple tracking. One locomotive and 8 cabs was reduced to 2 locomotives.

105.9% 96.4%

Sacramento South LRT (Phase 1)

No major scope changes. 108.4% 99.5%

Interstate MAX LRT Extension

Project length decreased from 12 miles to 5.8 miles.

Increased stations from 9 to 10.

NA 100.7%

Pittsburgh Stage II Reconstruction

Project length decreased from 12 miles to 5.5 miles. NA 106.0%

S. Florida Tri-Rail Double Tracking

Number of renovated stations increased from 9 to 10.

Number of new stations decreased from 2 to 1.

NA 104.4%

Largo Metrorail Extenstion Number of vehicles decreased from 18 rail cars and 12 buses to 14 rail cars. After project completion, FFGA was amended to add 52 rail cars and system power upgrades.

113.7% 103.3%

Dallas North Central LRT Number of new stations increased from 8 to 10. Double tracking increased from 8.5 miles to 12.5 miles. LRVs purchased increased from 17 to 21.

131.4% 94.9%

Chicago Douglas Branch No major scope changes. NA 93.2%

S. Boston Piers Transitway - Phase 1

No major scope changes. 150.7% 131.2%

Mission Valley East LRT Extension

No major scope changes. 130.9% 118.7%

Minneapolis Hiawatha Corridor LRT

Project length increased from 11.5 miles to 12 miles.

Reconfigured station at Mall of America.

285.9% 135.8%

Denver Southeast Corridor

Changes caused by a major expansion in the highway portion of project.

145.4% 98.0%

BART Extension to SFO Increased track length from 6.4 miles to 8.7 miles.

Increased stations from 3 to 4.

130.0% 130.9%

Hudson-Bergen MOS 1 and 2

Various changes in alignment, parking, grade-crossing treatments.

220.6% 94.5%

San Juan Tren Urbano New stations increased from 14 to 16 in the amended FFGA.

LRVs purchased increased from 60 to 74.

205.3% 174.0%

Average of 21 projects 140.2% 106.2%

The as-built costs compared to the FFGA award amount are shown in Figure 1. The as-built costs on average were close to the FFGA award amount, but in inflation-adjusted dollars, all 21 projects combined exceeded the combined FFGA award amount by $1.54 billion. Eleven of the projects had actual capital costs below their FFGA award amount. However, there were four projects that far exceeded their original FFGA award amount:

BART to SFO Airport

US Department of Transportation Page 13 of 160

Tren Urbano

South Boston Piers Transitway Phase 1

Minneapolis Hiawatha Corridor LRT

Combined, these four projects exceeded the FFGA amount by about $1.64 billion. Accordingly, the as-built costs of the remaining 19 projects, in aggregate, were about $100 million less than their total FFGA award amounts.

Figure 1: As-built Costs Compared to Original FFGA Award Amounts

$947 .8

$365 .9

$183 .8

$142 .8

$79 .6

$21 .8 $14 .5 $13 .8 $2 .1

($1 .1 ) ($ 2 .8 ) ($ 5 .7 ) ($ 85 .8 )($ 32 .4 )($ 23 .5 )($ 22 .0 )($ 17 .0 )($ 15 .2 )($ 12 .4 )($ 8 .0 )($ 7 .7 )

($300 )

($200 )

($100 )

$0

$100

$200

$300

$400

$500

$600

$700

$800

$900

$1 ,000

T re n U rb a n o

B A

R T

E x te n s io n t o S

F O

H ia w a th a C o rr id o r L R

T

S o u th

B o s to n P ie rs

P h a s e

M is s io n V a ll e y E a s t

L R

T

P it ts b u rg h S ta g e I

I

T ri

R a il D o u b le

T ra c k in g

L a rg o M e tr o ra il

E x te n s ti o n

In te rs ta te M

A X

L R

T

S a c ra m e n to S o u th P h a s e

B a lt im o re C e n tr a l

D o u b le T ra c k

M e tr a S

W C o rr id o r

N e w a rk

R a il L in k M

O S

M e tr a N o rt h C e n tr a l

S a lt

L a k e C it y

U n iv

M e d ic a l

E x t.

M e m p h is M e d

C e n te r

L R

T

D e n v e r

S o u th e a s t

M e tr a U

P W e s t

D a ll a s N o rt h C e n tr a l

L R

T

C T

A D o u g la s B ra n c h

H u d s o n B e rg e n

US Department of Transportation Page 15 of 160

2.3. Comparison to New Starts Projects from Prior Studies

Two previous studies sponsored or conducted by FTA have examined the relationship between project capital costs estimates and the cost of the project as-built.5,6 Pickrell’s study in 1990 documented the forecast and actual capital outlays for 10 projects – four heavy rail projects, four light rail projects, and two Downtown People Movers (DPM). Planning and construction of those projects spanned the period 1969 to 1987. The total capital outlay for those projects, compared to the inflation-adjusted projected capital outlay for each project, ranged from 11 percent below to 83 percent over the estimate. On average the actual capital costs of the projects examined in the 1990 study were 50 percent greater than originally forecasted. The 2003 study examined 21 projects – five heavy rail, ten light rail, three bus/HOV, two DPMs and one trolley bus. For those projects the average as-built cost was 21 percent higher than the inflation-adjusted capital cost estimate from AA, and the as-built cost for the 50th percentile project was 15 percent higher than the estimate made during AA.

The data, although based on relatively small samples, appear to show that the more recent projects have smaller cost-estimating errors than the project in the 1990 study (see Table 5). The as-built cost as a percent of the AA estimate had, on average, been reduced from 150 percent in the 1990 study to 121 percent in the 2003 study but increased in the current study to about 140 percent. While the average appears to have increased again in the 2007 study, this change in the average is due to a small number of projects for which as-build costs were substantially larger than the estimates. For the “typical” project, as represented by the 50th percentile, the actual cost was in the range of 115 percent to 123 percent of the AA estimate – statistically equivalent given the sample size.

Whether this is due to better cost estimating techniques resulting from additional experience with major rail projects or to better cost containment by project sponsors during project development is difficult to discern. The projects examined in Pickrell’s 1990 study were planned at a time when there had been little construction of large-scale transit projects in the previous 30 years.

Cost estimation for projects considered in the 2003 and 2007 studies had the benefit of more recent construction experience for similar projects and of the findings of the 1990 study. In addition, the projects considered in the 2003 and 2007 studies were planned during a time when

FTA was exercising greater scrutiny of costs estimates through expanded Project Management

Oversight activities, and had imposed the discipline of the Full Funding Grant Agreement process.

5 Pickrell, Don H., Urban Rail Transit Projects: Forecast Versus Actual Ridership and Costs, DOT-T-91-04, Office of Grants Management, Urban Mass Transportation Administration, Washington DC, October 1990.

6 This study was recently released as an appendix to FTA’s Contractor Performance Assessment Report (2007) which can accessed at http://www.fta.dot.gov/documents/CPAR_Final_Report_-_2007.pdf.

US Department of Transportation Page 16 of 160

Table 5: As-Built Capital Cost Compared to Inflation-Adjusted AA/DEIS Estimate – 1990, 2003 and 2007

Studies

Study Average 50th Percentile

Pickrell Report 1990 150 % 151 %

FTA 2003 121 % 115 %

FTA 2007 140 % 122 %

The effectiveness of the Full Funding Grant Agreement and associated cost containment actions are illustrated in the relatively small differences between FFGA amounts and actual project costs observed in the 2003 and 2007 studies (see Table 6). On average, the actual costs range of six to seven percent over the FFGA cost estimate while the 50th percentile value, a measure that minimized the effect of outliers, ranges from two percent under to four percent over the FFGA cost estimate. Given the small size of the samples from which these measures are derived, the results from the 2003 and 2007 studies are approximately the same – the actual costs of most recent projects do not differ significantly from the FFGA amounts.

Table 6: Project As-built Costs as Percent of FFGA – 2003 and 2007 Studies

Study Average 50th Percentile

FTA 2003 107 % 104 %

FTA 2007 106 % 98 %

2.4. Duration of Project Development

New Starts transit projects are, by their nature, large and complex endeavors that must follow a reasonably well prescribed process of planning and project development. This process has been fairly consistent since the passage of the Intermodal Surface Transportation Efficiency Act of

1991 (ISTEA) and includes the selection of the LPA after a planning study followed by the federal New Starts process of PE, final design and construction managed by FTA. Almost all of the projects completed between 2003 and 2007 have been planned and developed under the rules and regulations reflecting ISTEA and its successor transportation laws.

The following chart shows the duration of project development, from the completion of the planning study (selection of the LPA) until the year of opening to revenue service for New Starts projects completed between 1990 and 2007 and includes several pre-ISTEA projects. The data shows that the project development period has remained remarkably consistent over time. There is no discernible trend in the duration of project development for projects that opened for revenue service between 1990 and 2007.7

7 The R2 value in Figure 2 confirms that there is no apparent trend in project development duration. The R2 value is the percent of the variation in project development duration explained by the year of opening (time trend) which is approximately zero.

US Department of Transportation Page 17 of 160

Figure 2: Duration of Project Development by Year of Opening

R = 0.0008

1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Year of Opening

P ro je c t

D e v e lo p m e n t

P e ri o d

Y e a rs

US Department of Transportation Page 19 of 160

3. RIDERSHIP

This chapter compares the ridership forecasts prepared to support key decisions in the planning and project development process for New Starts projects to the actual ridership performance of the projects. The planning level forecasts (typically prepared during an AA, MIS or an

AA/DEIS) were used to support the decision to choose the locally preferred alternative and were generally the forecasts that were presented to FTA when the project entered preliminary engineering. The ridership forecast that was used to support FTA’s decision to allow the project to advance to final design and construction is almost always prepared as part of the FEIS or EA

(the FFGA generally reports the FEIS ridership forecasts).

This study compares the ridership forecasts prepared for 18 of the 21 New Starts projects8 in this study at these two key decision points, to the actual ridership experienced by those projects after opening. The purpose of this exercise was to identify the extent to which the actual ridership on recent New Starts projects deviated from their original forecasts. The forecasts are not evaluated in enough detail to conclusively identify the specific reasons for any forecasting errors, though any obvious factors that influence ridership are noted where appropriate.

3.1. Ridership Analysis Approach

FTA’s approach to analyzing the accuracy of ridership forecasts focuses on the information available at key decision points. In FTA’s view, the most important decision point in the development of transportation projects occurs at the end of alternatives analysis when the local decision-makers select the locally preferred alternative. FTA believes that it is critical that this decision in particular, where the choice of a project’s mode and general alignment is made, be based on the most reliable information possible. When the LPA is a major transit capital investment, FTA uses the information developed during alternatives analysis to make a decision whether to allow a project to enter preliminary engineering to compete for federal funding. The next key decision point occurs after the completion of the FEIS (or EA) and FTA’s issuance of a

Record of Decision (or FONSI). FTA then uses the information in the final environmental document to decide whether to allow a project to advance to final design and construction.

While capital cost estimates are updated before a project is finally funded by FTA, ridership forecasts are generally not updated. The FEIS typically represents the final ridership forecast upon which key decisions depend.

8 The results for the three Chicago Metra commuter rail projects are excluded from the ridership analysis. The

MIS’s prepared for these projects only reported new transit riders in the forecast year relative to the baseline alternative. Neither project’s MIS reported how many boardings were expected on the Metra lines under the build or the baseline alternatives. Therefore, there is no way to compare actual boardings to the forecasts in a way that is comparable to the other projects.

US Department of Transportation Page 20 of 160

This study compares, to the extent possible, the predicted and actual ridership for each project at three points:

Selection of the LPA or entry into PE (usually from the MIS, AA or AA/DEIS), Entry into final design (usually from the FEIS or EA), and

Actual (2006/2007 actual and adjusted to the forecast year).

The primary ridership measure that was chosen as the basis of comparison is average weekday boardings. This measure was chosen primarily because nearly every ridership forecasting effort produces a forecast of average weekday boardings and every agency can supply estimates of actual average weekday boardings for their projects for most years. This measure was also chosen in FTA’s previous studies so the forecasting performance of the newer projects in this study can be compared to the results of the older study.

Projects that are extensions of existing systems require some care when interpreting station boarding data. The observed boardings on an extension to an existing system may be as little as half the total boardings attributable to the project system-wide. This is because counting average weekday boardings at the new stations will miss many boardings associated with the return trip that occur elsewhere on the system (including existing stations), thus undercounting ridership for the extension. For this reason, average weekday boardings are not the same as average weekday ridership for extensions of existing lines. The comparison of average weekday boardings is still useful for evaluating the performance of the forecasts as long as the actual station boardings are compared to forecast station boardings. Some forecasts present station boardings while other present boardings plus alightings9. FTA made every effort to compare the forecasts to the equivalent comparable measure of actual ridership. In each case, FTA compared the predicted ridership to the actual riderhip in equivalent measures. For specific details about the comparison made for each project, see the Project Profiles in the Appendix.

For this study, FTA defines a reasonably accurate forecast as one the can reasonably be expected to come within 20 percent of the actual ridership achieved by the project by the forecast year.

3.2. Forecast and Actual Ridership

Ridership forecasts are developed for a specific “forecast year.” In this study, the forecast year for each project except Pittsburgh Stage II Reconstruction remains in the future, either 2010, 2015, or 2020. In many cases, the comparison between current ridership and forecast year predictions is quite accurate because most of the ridership is captured in the early years after opening with often slow growth in ridership thereafter. However, there are some rapidly growing urban areas that have shown rapid ridership growth. To account for the ridership that is likely to occur by the forecast year, FTA also presents a comparison between predicted ridership and actual ridership adjusted by the average annual growth in system-wide transit boardings between 1996 and 2006. This adjustment allows the comparison of the forecast to a reasonable approximation of future ridership based on observed data and recent trends.

9 The ridership comparisons for BART to SFO, Portland Interstate MAX, and Denver Southeast use boardings plus alightings. For the Chicago Douglas Branch project, FTA doubled observed station boardings to approximate total ridership since no forecast of station boardings and alightings was produced.

US Department of Transportation Page 21 of 160

3.2.1. Average Weekday Boardings

Table 7 reports forecast and actual average weekday boardings (or boardings plus alightings) for each of the New Starts projects included in this study. The results indicate that, as of 2007, two projects have exceeded their AA/DEIS ridership forecast, six projects are between 60 and 80 percent of their AA/DEIS ridership forecasts, and the remaining 10 projects are well below their ridership planning level forecasts. Since the forecast years for several of these projects remain well in the future and many of these agencies have a history of growing ridership, most of the six projects that exceed 60 percent of their ridership forecasts will likely reach at least 80 percent of their predicted ridership by the forecast year. All told, slightly less than half of the New Starts projects included in this analysis (i.e. 8 of 18) either have achieved, or have a good chance of exceeding, 80 percent of their initial planning level ridership forecast.

As shown Table 7, only three ridership forecasts changed appreciably between planning and the completion of the FEIS. The Minneapolis Hiawatha forecasts were revised downward by a significant amount while Denver Southeast (due to technical model updates) and Dallas North

Central (due to a scope change resulting…