flexibility and design options analysis of a streetcar system in madison, wi

Flexibility and Design Options Analysis of a Streetcar System in Madison, WI

Paul Grogan

ESD.71 Application Portfolio

Massachusetts Institute of Technology

December 2, 2008

Paul Grogan, Dec. 2008Slide 1

Problem Statement• “Pedestrian Accelerator” proposed in Madison, WI

– Add sustainable commuting options

– Encourage economic development along route

• Modeled after similar success stories

– Portland, Oregon

– Little Rock, Arkansas

– Tampa, Florida

• Initial feasibility study finished October 2007

Paul Grogan, Dec. 2008Slide 2

Park Street Route (Purple)

Central Loop Route (Blue)

East Isthmus Route (Yellow)University of

Wisconsin Campus

Commercial Area

Residential Area

State Capital

Industrial Area

Student Residential Area

Convention Center

Residential Area

Commercial Area

Paul Grogan, Dec. 2008Slide 3

1 mi

Ridership Uncertainties• Expected initial ridership from feasibility study

– 2,828 daily riders throughout entire area

• Expected growth rate from feasibility study

– 0.6% per year increase without considering economic growth

– 1.5% per year increase considering economic growth

Paul Grogan, Dec. 2008Slide 4

RouteConstruction

CostPopulation Use Factor

Expected Ridership

Low Ridership(-20%)

High Ridership(+50%)

East Isthmus – Campus $82.6m 40% 1,131 905 1,697

Park Street $93.9m 45% 1,272 1,018 1,909

Central Loop $61.2m 30% 848 679 1,272

Route Design and Options• Option A: Large Scale, Fixed Design

– Implement all three routes at same time

– Save on construction costs ($237.7 million $200.7 million)

• Option B: Phased, Flexible Design

– Implement Central Loop route initially

– Decide whether to expand to East Isthmus – Campus or Park

Street expansion route after ten years

• Uncertainty includes initial ridership level and average ridership growth rate between years 10 and 40

Paul Grogan, Dec. 2008Slide 5

Paul Grogan, Dec. 2008Slide 6

Decision Tree Analysis of Route Flexibility and Expansion

• Strategy: build flexible route, do not expand

• Both design options have very negative net present values

– Ridership levels cannot support large initial investment

– Is NPV an appropriate valuation metric for a public service?

Paul Grogan, Dec. 2008Slide 7

Measure Fixed Flexible Expected NPV $ (181m) $ (55m)Maximum NPV $ (160m) $ (49m)Minimum NPV $ (200m) $ (61m)Initial CAPEX $ 201m $ 61mENPV / CAPEX $ (0.90) $ (0.90)ENPV / Rider $ (1,070) $ (827)

Decision Tree Analysis with Additional Ridership Levels

• Thought experiment: boost ridership levels 400% higher than expected to repay initial capital expenses

• Flexibility limits downside risk, but also limits upside potential (traditional result)

Paul Grogan, Dec. 2008Slide 8

Measure Fixed FlexibleExpected NPV $ 19m $ 26mMaximum NPV $ 129m $ 102mMinimum NPV $ (68m) $ (21m)Initial CAPEX $ 201m $ 61mENPV / CAPEX $ 0.09 $ 0.43ENPV / Rider $ 8 $ 27

Lattice Analysis Setup• Ridership growth model fitted to

Madison Metro Bus system ridership

from 1971 – 2007

• Low R-squared value (0.1337)

• Period in 1980’s of ridership increase

may be similar to late 2000’s

• Lattice analysis variables

• Ridership distribution after six years

Variable ValueV 0.52%σ 15.77%U 117.1%D 85.4%P 51.6%

Paul Grogan, Dec. 2008Slide 9

Lattice Analysis of Closing Option• Analyzing closing option in Central Loop route• Closing option is valued at $67,476

Paul Grogan, Dec. 2008Slide 10

Year 0 1 2 3 4 5 6

Expected Future NPV

----Decision

$728,108-

$1,154,294-

$1,526,719-

$1,776,965-

$1,834,112-

$1,624,138-

$1,052,779-

$180,640-

$478,653-

$751,209-

$918,785-

$898,597-

$621,415-

$(90,356)CLOSE

$69,498-

$252,706-

$369,287-

$306,718-

$(156,258)CLOSE

$(90,356)CLOSE

$(13,200)CLOSE

$77,134-

$(212,547)CLOSE

$(156,258)CLOSE

$(90,356)-

$(260,625)CLOSE

$(212,547)-

$(301,690)-

Path Enumeration and Valuation

Add PV cash flow from the first state in the path from a

lattice not accounting for option execution

Execute decision after current state?

Add PV one-time cost of executing option

(e.g. capital expense or penalty)

Add PV cash flows from remaining states in path from a

lattice accounting for option execution

Add PV cash flow from next state in path from a lattice not

accounting for option execution

Yes

No

In this analysis, the one-time cost is

assumed to be zero

In this analysis, the lattice accounting for option execution

is simply zero cash flows

Paul Grogan, Dec. 2008Slide 11

Lattice Valuation• Option is executed (close operations) in 34 of 64 paths

(50.1% chance overall)

– Good chance that revenue can’t cover operational costs

• Flexibility limits downside risk but does not affect upside (traditional result)

Paul Grogan, Dec. 2008Slide 12

Measure Fixed FlexibleExpected NPV $ 660,632 $ 728,108Maximum NPV $ 2,859,137 $ 2,859,137Minimum NPV $ (756,570) $ (17,393)

Conclusions• Expected ridership levels are not high enough to support

a streetcar system of the studied scale• Under expected ridership levels, a closing option is used a

very large percentage of the time (~50%)• Under higher ridership levels, a flexible, phased design

produces significantly higher net present value• Including a closing option increases net present value but

does not take into account the construction cost losses

Paul Grogan, Dec. 2008Slide 13