Wang performance measures

Transportation Cost Index:A Comprehensive Performance Measure for Transportation and Land Use Systems

Liming Wang, Portland State University

In Collaboration with Jenny Liu, Huajie Yang, Shi Wei (PSU)

Bud Reiff (Metro), Brian Gregor (Oregon System Analytics)

Outline● Why yet another performance

measure (YAPM)?● Transportation cost index: the idea● Implementations and application● Ongoing and future work

Demand for Comprehensive Measures● As a supplement/replacement of

traffic-centric measures: LOS, travel delay

● MAP-21 emphasizes use of performance measures in transportation planning & operation

● State legislations: Oregon Job and Transportation Act (OJTA)

Existing Measures● Handy and Niemeier, 1997● Geurs and van Wee, 2004● NCHRP Report 311, 618, 694, 708 ...

Market Potential MeasuresEmployment accessible within 30 minutes by public transit during a.m. peak• Easy to interpret/understand• Opportunities, mode, time-of-

day and time budget specific

Source: University of Minnesota, Accessibility Observatory

Utility-based MeasuresLogsum as an accessibility measure • Elegant, composite measures

for all modes; possible to derive net user benefit between scenarios

• Hard to interpret by itself; unable to compare across regions/times

E(CS) = ln �𝑚𝑚′

exp 𝑈𝑈𝑚𝑚′𝑘𝑘𝑘𝑘 + 𝐶𝐶

Generalized Costs Indicator

per distance generalized costs for motorized trips• Easy to

interpret/understand; able to monitor trends and compare scenarios

• ignores land use system; mode, time-of-day specific

Source: Koopmans, et al, 2013

H+T® Affordability Index• Tracks out-of-pocket

monetary costs of transportation and adds them to housing costs as a location efficiency measure;

• Does not track the performance of transportation system except for Auto/Transit uses.

Source: Center for Neighborhood Technology (CNT)

Wish List for YAPM● A comprehensive indicator able to present an

overall picture;● Easy to interpret/understand;● Applicable to both trend monitoring and

scenario comparison;● Able to fill gaps in policy areas not adequately

covered by existing performance measures, e.g. the equity and compatibility aspects (Reiffand Gregor, 2005)

Consumer Price Index (CPI)Measures changes in the price level of a basket of consumer goods and services purchased by households:● Build a basket of goods and services from

pre-defined item groups● Track the prices of goods and services in the

basket

Transportation Cost Index (TCI)● Comprehensive measure of transportation

and land use systems;● Easy to interpret/understand;● If based on widely available data sources,

possible for use in trend monitoring and comparing scenario outcomes;

● Able to serve as an indicator for policy areas including transportation and land use system compatibility and balance.

Transportation Cost Index (TCI)Measures changes in the price level of a market basket of trips/destinations meeting households’ daily needs:● Identify a basket of trips/destinations

based on pre-defined groups (trip purpose categories);

● Track the costs of accessing destinations/trips in the basket.

Implementation A: Travel Survey-based MethodRelies primarily on input from household activity survey, e.g. Oregon Travel & Activity Survey (OTAS)• Construct travel baskets based on activity

diaries or a sample of trips/tours that are representative of regional travel pattern, potentially by trip purpose, household size, income group and geography;

• Easy for trend monitoring.

Implementation B: TDM-based MethodRelies on inputs from travel demand model− Data readily available for regions w/ TDM; − Straightforward for scenario comparison;− Theoretically can calculate the transportation

cost for every income group and for every TAZ.

− But we may not want to track it too closely –patterns from idiosyncrasy or true differences?

Implementation B: TDM-based Method

Calculate Travel Costs:Cost Estimate by Mode

𝐶𝐶 = 𝐶𝐶0 + 𝑘𝑘 � 𝑇𝑇𝑇𝑇 + 𝑤𝑤 � 𝑇𝑇𝑇𝑇𝐶𝐶0 - Constant𝑘𝑘 � 𝑇𝑇𝑇𝑇 - Monetary costs (Fuel and tire costs, Ownership costs, insurance, transit fare, etc) of travel𝑤𝑤 � 𝑇𝑇𝑇𝑇 - Time costs of travel

Applications and Demonstration

Generalized Costs by Household Income Level (Portland, 2011)

Low income

Mid income

High income

Generalized Costs by Household Size (Portland, 2011)

1

2

4+

3

Generalized Costs by Purpose & Income Level (Portland, 2011)

Generalized Costs by Purpose, Income Level and Transportation Districts (Portland, 2011)

Generalized Costs by Household Income Level (Portland)

2011 1994

Generalized Costs by Household Size (Portland)

2011 1994

Generalized Costs by Purpose and Income Level (Portland)

2011 1994

Generalized Costs Portland vs Salt Lake City

Df Sum Sq Mean Sq F value Pr(>F) msa 1 75870 75870 4.524 0.0334 *Res 11414 191411731 16770

Portland – Salt Lake CityMean difference: 5.3795% Confidence interval: [0.42, 10.32]

Generalized Costs by Household Income Level

Portland Salt Lake City

Generalized Costs by Purpose and Income Level

Portland Salt Lake City

Using NHTS Data (2009)summary(n.tcost.hh.msas.aov)

Df Sum Sq Mean Sq F value Pr(>F) msa 2 129454 64727 7.566 0.000539 ***Residuals 1435 12275919 8555

diff lwr upr p adjTampa Bay-Portland -21.63 -43.30 0.04 0.05Salt Lake City-Portland 12.59 -19.02 44.21 0.62Salt Lake City-Tampa 34.23 9.63 58.83 0.003

Using NHTS Data (2009)By household income group:

By household size:

Portland Salt Lake City Tampa Bay

Ongoing and Future Work● Under development/testing at

http://github.com/cities-lab/tci● Reconcile TCIs from the two methods;● Verify patterns of transportation costs with

information from alternative data sources, such as CES;

● Should external costs be included?

Ongoing and Future Work● Adopted by the Accessibility Indicator

Development Team (IDT) as one of indicators for the Oregon Mosaic project mandated by OJTA

● Test TCI usage in public engagement and policy making process

Acknowledgements

National Institute for Transportation and Communities

Oregon DOT

Extra Slides

Income LevelsTo be consistent with the classification used in Metro’s TDM, household income levels are classified with this scale (1994 dollars):• < $25K: Low Income • $25-50K: Mid Income• > $50K: High Income

Identify Activity Centers (Travel Market Basket)

Steps (Giulinao, 1991)

1. Calculate employment/size term density;2. Identify TAZs with densities greater than

density cutoff D and group contiguous TAZs identified into preliminary centers;

3. Calculate total employment or size terms for each center identified in step 2 and eliminate centers with total employment or size terms below total cutoff E from centers identified in step 2. The remaining are activity centers.

Determine Cutoffs

• Giulinao (1991) provides no guidance in selecting density cutoff (D) or total cutoff (E). They relied on expert knowledge

• Sensitivity Tests to determine cutoffs

Sensitivity Tests: HBW

Sensitivity Tests: HBS

Sensitivity Tests: HBS

Sensitivity Tests: HBO

Travel Costs Calculation:Cost Estimate by Mode

• Auto𝐶𝐶𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 = 𝐶𝐶𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎0 + 𝑘𝑘𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 � 𝑇𝑇𝑇𝑇𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 + 𝑤𝑤𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 � 𝑇𝑇𝑇𝑇𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎

– 𝐶𝐶𝑚𝑚0 - Constant– 𝑘𝑘𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 � 𝑇𝑇𝑇𝑇𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 - Monetary costs (Fuel and tire

costs, Ownership costs, insurance, etc) of driving– 𝑤𝑤𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 � 𝑇𝑇𝑇𝑇𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎- Time costs of driving

Travel Costs Calculation:Cost Estimate by Mode

• Public Transit:𝐶𝐶𝑝𝑝𝑎𝑎𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = fare + 𝑤𝑤𝑝𝑝𝑎𝑎𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 � 𝑇𝑇𝑇𝑇𝑝𝑝𝑎𝑎𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝

– Fare: Transit fares– 𝑤𝑤𝑚𝑚 � 𝑇𝑇𝑇𝑇𝑝𝑝𝑎𝑎𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝: Time costs of riding transit

• Non-motorized modes (bicycling and walking)𝐶𝐶𝑝𝑝𝑝𝑝𝑝𝑝𝑏𝑏𝑝𝑝𝑝𝑝𝑏𝑏 = 𝐶𝐶𝑝𝑝𝑝𝑝𝑝𝑝𝑏𝑏𝑝𝑝𝑝𝑝𝑏𝑏0 + 𝑤𝑤𝑝𝑝𝑝𝑝𝑝𝑝𝑏𝑏𝑝𝑝𝑝𝑝𝑏𝑏 � 𝑇𝑇𝑇𝑇𝑝𝑝𝑝𝑝𝑝𝑝𝑏𝑏𝑝𝑝𝑝𝑝𝑏𝑏

𝐶𝐶𝑤𝑤𝑎𝑎𝑝𝑝𝑘𝑘 = 𝑤𝑤𝑤𝑤𝑎𝑎𝑝𝑝𝑘𝑘 � 𝑇𝑇𝑇𝑇𝑤𝑤𝑎𝑎𝑝𝑝𝑘𝑘– Time costs of Bicycling and Walking