Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Wang performance measures

Liming Wang, Portland State University

  • Login to see the comments

Wang performance measures

  1. 1. 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)
  2. 2. Outline ● Why yet another performance measure (YAPM)? ● Transportation cost index: the idea ● Implementations and application ● Ongoing and future work
  3. 3. 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)
  4. 4. Existing Measures ● Handy and Niemeier, 1997 ● Geurs and van Wee, 2004 ● NCHRP Report 311, 618, 694, 708 ...
  5. 5. Market Potential Measures Employment 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
  6. 6. Utility-based Measures Logsum 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 𝑈𝑈 𝑚𝑚′ 𝑘𝑘𝑘𝑘 + 𝐶𝐶
  7. 7. 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
  8. 8. 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)
  9. 9. 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 (Reiff and Gregor, 2005)
  10. 10. 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
  11. 11. 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.
  12. 12. 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.
  13. 13. Implementation A: Travel Survey-based Method Relies 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.
  14. 14. Implementation B: TDM- based Method Relies 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?
  15. 15. Implementation B: TDM-based Method
  16. 16. Calculate Travel Costs: Cost Estimate by Mode 𝐶𝐶 = 𝐶𝐶0 + 𝑘𝑘 � 𝑇𝑇𝑇𝑇 + 𝑤𝑤 � 𝑇𝑇𝑇𝑇 𝐶𝐶0 - Constant 𝑘𝑘 � 𝑇𝑇𝑇𝑇 - Monetary costs (Fuel and tire costs, Ownership costs, insurance, transit fare, etc) of travel 𝑤𝑤 � 𝑇𝑇𝑇𝑇 - Time costs of travel
  17. 17. Applications and Demonstration
  18. 18. Generalized Costs by Household Income Level (Portland, 2011) Low income Mid income High income
  19. 19. Generalized Costs by Household Size (Portland, 2011) 1 2 4+ 3
  20. 20. Generalized Costs by Purpose & Income Level (Portland, 2011)
  21. 21. Generalized Costs by Purpose, Income Level and Transportation Districts (Portland, 2011)
  22. 22. Generalized Costs by Household Income Level (Portland) 2011 1994
  23. 23. Generalized Costs by Household Size (Portland) 2011 1994
  24. 24. Generalized Costs by Purpose and Income Level (Portland) 2011 1994
  25. 25. 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 City Mean difference: 5.37 95% Confidence interval: [0.42, 10.32]
  26. 26. Generalized Costs by Household Income Level Portland Salt Lake City
  27. 27. Generalized Costs by Purpose and Income Level Portland Salt Lake City
  28. 28. 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 adj Tampa Bay-Portland -21.63 -43.30 0.04 0.05 Salt Lake City-Portland 12.59 -19.02 44.21 0.62 Salt Lake City-Tampa 34.23 9.63 58.83 0.003
  29. 29. Using NHTS Data (2009) By household income group: By household size: Portland Salt Lake City Tampa Bay
  30. 30. 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?
  31. 31. 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
  32. 32. Acknowledgements National Institute for Transportation and Communities Oregon DOT
  33. 33. Extra Slides
  34. 34. Income Levels To 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
  35. 35. Identify Activity Centers (Travel Market Basket)
  36. 36. 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.
  37. 37. 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
  38. 38. Sensitivity Tests: HBW
  39. 39. Sensitivity Tests: HBS
  40. 40. Sensitivity Tests: HBS
  41. 41. Sensitivity Tests: HBO
  42. 42. Travel Costs Calculation: Cost Estimate by Mode • Auto 𝐶𝐶𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 = 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
  43. 43. Travel Costs Calculation: Cost Estimate by Mode • Public Transit: 𝐶𝐶𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 = fare + 𝑤𝑤𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 � 𝑇𝑇𝑇𝑇𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 – Fare: Transit fares – 𝑤𝑤𝑚𝑚 � 𝑇𝑇𝑇𝑇𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝: Time costs of riding transit • Non-motorized modes (bicycling and walking) 𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = 𝐶𝐶𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏0 + 𝑤𝑤𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 � 𝑇𝑇𝑇𝑇𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝐶𝐶𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 = 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 � 𝑇𝑇𝑇𝑇𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤 – Time costs of Bicycling and Walking

×