Predicted vs. Actual Costs and Ridership - Urban Transport Projects

Transcription

1 Predicted vs. Actual Costs and Ridership - Urban Transport Projects Lessons from the US presented by Steve Lewis-Workman, Transport Economist East Asia Department, Transport Division ADB Transport Forum May 26, 2010

2 History of cost overruns Boston, USA Central Artery (aka Big Dig) Estimate $6 billion Actual $14.6 billion Tokyo, Japan Oedo Subway Estimate billion Actual 1,400 billion UK France Channel Tunnel Estimate 2.6 billion Actual 4.65 billion 143% over 105% over 79% over

3 History of over-estimated ridership LA Red Line Forecast 300,000/day Actual 130,000/day San Juan, PR Tren Urbano Forecast 110,000/day Actual 35,000/day Tokyo Oedo Subway Forecast 900,000/day Actual 700,000/day Bangkok Metro Forecast 400,000/day Actual 180,000/day 56% under 68% under 22% under 55% under

4 Reputation for poor performance Unreliable information Poor decisions Wasted resources Threatens financial sustainability in developing countries Produces culture of failure

5 Three Studies from US DOT 1990 Urban Rail Transit Study of Ridership and Costs 10 rail projects opened between 1975 and Predicted vs. Actual Impacts of New Starts 19 rail and busway projects opened between 1990 and Predicted vs. Actual Impacts of New Starts 21 rail and busway projects opened between 2002 and 2007 Detailed ridership forecast and cost estimation analysis required of every new project opened after 2007

6 Assembled data Average weekday project boardings/alightings by station Construction costs At 4 key points Alternatives analysis (select mode/alignment) Final EIS (approval to begin final design) Government funding contract approval After opening (latest available data for ridership) All data reviewed/approved by project sponsors

7 Adjusting to comparable year Adjusted actual ridership by average annual system-wide transit ridership growth to forecast year if still future Cost compared in real $ (no inflation) Remove financing charges Focus on project cost estimation Actual costs adjusted by removing prior year inflation impacts (ENR inflation by cost category)

8 Actual cost as % of estimates % % Average 28% overrun Median 20% overrun % % 50.00% 0.00% San Jose Tasman West Chicago SW Transitway Pittsburgh LRT San Diego El Cajon Extension St. Clair Co. MetroLink Extension Pittsburgh Stage II Reconstruction Salt Lake City I-15/State Street Chicago Douglas Branch Baltimore Central LRT Double-Track Salt Lake City University/Medical Ext Houston Southwest Transitway Miami Omni and Brickell Ext Metra SW Corridor Portland Interstate MAX LRT Tri-Rail Double Tracking Metra North Central Metra UP West Atlanta North Line Extension Sacramento South LRT I Dallas South Oak Cliff Baltimore Johns Hopkins Ext Largo Metrorail Extenstion Sacramento LRT Newark Rail Link MOS-1 Pittsburgh Airport Busway/ Wabash Denver Southwest Denver North I-25 St. Louis MetroLink Hiawatha Corridor LRT** BART Extension to SFO Mission Valley East LRT Extension Dallas North Central LRT Miami Metro Rail Baltimore BWI, Hunt Valley, Penn Sta. Denver Southeast Corridor Los Angeles Red Line San Jose Guadalupe Corridor Detroit People Mover South Boston Piers Transitway I Buffalo LRT Portland LRT Seattle Downtown Project Atlanta Metro System San Francisco BART Colma Jacksonville Skyway Express Baltimore Subway Memphis Med Center LRT Portland Westside Hillsboro Washington Metro System Hudson-Bergen MOS 1 & 2 Tren Urbano Miami People Mover

9 Actual ridership as % of forecast 160.0% 140.0% 120.0% Average 61% of forecast Median 60% of forecast 100.0% 80.0% 60.0% 40.0% 20.0% 0.0% Jacksonville ASE Detroit DPM Memphis Medical Center NJ Newark Elizabeth MOS 1 Buffalo LRT South Florida Tri-Rail Tren Urbano Miami HR Pittsburgh LRT Pittsburgh West B'Way San Jose Guadalupe Miami DPM Baltimore Metro BART to SFO Atlanta North Line LA Red Line Atlanta HR Miami Omni/Brickell Pittsburgh Stage II LRT Baltimore Central Double Track Chicago Orange Line Baltimore LRT Ext. Houston SW Transitway Washington Largo South Boston Piers Phase 1 Sacramento LRT NJ Hudson Bergen MOS 1&2 San Jose Tasman West Baltimore Johns Hopkins Portland LRT Washington HR San Diego Mission Valley East Sacramento South Phase 1 Portland Westside-Hillsboro BART Colma Chicago Douglas Branch Portland Interstate MAX Dallas South Oak Cliff Minneapolis Hiawatha St. Louis Initial System Denver Southeast LRT Salt Lake South LRT Denver SW LRT San Diego El Cajon St. Louis St. Clair Ext. Dallas North Central

10 Lies or incompetence? PROBABLY BOTH

11 Are forecasts getting better? Cost estimates as % of actual Study Average 50 th Percentile USDOT (FTA) % 151% USDOT (FTA) % 115% USDOT (FTA) % 122% Ridership forecasts as % of actual Study Average 50 th Percentile USDOT (FTA) % 41% USDOT (FTA) % 71% USDOT (FTA) % 64%

12 Distribution of errors Costs Sample means significantly different from % 70-90% % % % % 170%- Forecast errors significantly skewed Ridership % 50-70% 70-90% % % % 150%-

13 Do larger projects higher % overruns? Not really, but budgetary impact much larger $5,050.0 $4,550.0 $4,050.0 $3,550.0 As-built cost $3,050.0 $2,550.0 $2,050.0 $1,550.0 $1,050.0 R 2 = $550.0 $ % % % % % % % Actual cost as % of estimate

14 Some observations Difficult to determine specific causes for individual projects after the fact Intentions hard to determine Lack of archived data, no functioning travel model, costs in inconsistent formats 2003 study supplemented by detailed analysis of several case studies Ridership forecasts that did better tend to have higher population/employment than predicted

15 A problem only in US? Studies by Bent Flyvbjerg, et al. on costs International data set 258 projects 90% of projects overrun costs Rail projects - highest escalation (45% over) Road projects - lowest escalation (20% over) No improvement over time (70 years of data) Developing countries have HIGHER overruns

16 Does private ownership reduce cost overruns? No evidence that it does, though data sparse (Flyvbjerg, 2004) Average cost overruns on bridge/tunnel projects Private 34% State owned enterprise 110% Other public 23% Average cost overruns on high speed rail State owned enterprise 88% Other public 15% Average cost overruns on urban rail State owned enterprise 36% Other public 54%

17 How to improve? Funding partners must tolerate uncertainty Reference class forecasting (rely on data not engineers) Change incentives Force choices on decision-makers no free money Formula based multi-modal subsidy allocations Higher local government or private partner share Change subsidy schemes Pay for performance after completion Fixed subsidy amount at AA Don t use hurdle rates Shame

18 Thank You