Analyzing the Impact of Environmental and. with NHTS Data

Transcription

1 TRB Environment and Energy Research Conference, Raleigh, NC June 7, 2010 Analyzing the Impact of Environmental and Energy Policies on Travel Behavior with NHTS Data Presenting Author: Lei Zhang 1,* Co-Authors: Qing Shen 2, Yijing Lu 1, Jasmy Methipara 1, Xiang He 1, Nick Ferrari 1, Cory Krause 1, and Jin Hyun Hong 2 1.Department of Civil & Environmental Engineering University of Maryland College Park 2. Department of Urban Design and Planning University of Washington *Contact Information: Phone: , lei@umd.edu 1

2 Background Travel Survey Primary data source in the U.S. for travel behavior analysis and travel demand modeling at all levels National Household Travel Survey Socio-economic, demographic, location, vehicle ownership, and travel information at the household level Arguably the most comprehensive dataset for travel analysis and monitoring at the national level Travel Survey Trend Cross-sectional (Rotating) Panel Mail/Telephone-based methods GPS-based methods Decisions Decision-making processes 2

3 NHTS Data Processing and Integration Integration with EPA Nonattainment Areas Identifying whether or nor an NHTS household resides in a nonattainment area for each criteria pollutant over time. Integration with Energy/Environmental Data Fuel price at the state and county levels; Vehicle characteristics ti such as price, fuel type, pollution emission rates, etc.. Integration with Land Use Information Merging NHTS household longitude/latitude information with various land use characteristics at the parcel, grid cell, TAZ, and other geographic levels. 3

4 Research Questions Develop statistical models based on the NHTS data to answer the following environmental and energy policy questions: What is the impact of air quality control and EPA nonattainment designation on travel behavior and VMT? What is the impact of green transportation financing policies (e.g. green VMT fees, congestion pricing) on VMT, revenue, and equity at the national and state levels? What is the impact of land use policies (e.g. high density, mixed development, neighborhood h design) on travel behavior and VMT? Etc. 4

5 Environmental and Energy Policy Analysis #1 What is the impact of air quality control and EPA nonattainment designation on travel behavior and VMT? 5

6 Regression Model Results Dependent Var.: ln (VMT) Model 1 Model 2 Nonattainment Status: Own County * (0.060) ** (0.012) Nonattainment Status: Adjacent County * (0.066) Large urban area * * (0.079) (0.089) Small urban area ** ** (0.004) (0.007) Distance to Urban Center ** ** (0.048) (0.043) ln(population density) ** ** (0.000) 000) (0.000) 000) Number of transit trips taken ** ** (0.000) (0.000) Adjusted R-Square Household socio-economic and demographic variables not shown.

7 Impact of Nonattainment on VMT by County

8 Findings from Household-Level Analysis Average Impact of Nonattainment Designation 1.64% reduction in vehicle miles traveled Spatial Variation of the Impact 4.54% VMT reduction in counties that only have attainment surrounding counties 1.15% 15% VMT reduction in counties that only have nonattainment surrounding counties 8

9 Sensitivity Analysis: Time Period Selection Estimated Impact of Nonattainment Designation on % VMT Reduction 0.0% 0.2% 0.4% 0.6% 0.8% 1.0% 1.2% 1.4% 1.6% 1.8% 2.0% Time Period

10 Environmental and Energy Policy Analysis #2 What is the impact of green transportation financing policies (e.g. green VMT fees, congestion pricing) on VMT, revenue, and equity at the national and state levels? 10

11 Discrete-Continuous Choice Model Number of vehicles 0 vehicles 1 vehicle 2 vehicles 3 vehicles >3 vehicles CAR TRUCK VMT Both CARS Both TRUCKS CAR and TRUCK Feedback VMT VMT CAR CAR TRUCK TRUCK TRUCK CAR All CARS All TRUCKS VMT VMT 11

12 Results: Vehicle Use Equation Dependent Var.: ln (VMT) Coefficients P-Value ln(driving cost/mile) ln(income) ln(driving cost/mile)* ln(income) ln(driving cost/mile)* (vehilcle substitute) ln(vehicle count) Vehicle substitute Household socio-demographic variables not shown.

13 Effectiveness for Revenue Generation Gas Tax Increase federal tax by 10 cents/gallon to 28.4 cents/gallon. This 54.3% tax increase results in 50.5% 5% increase in federal tax revenue in the short run. Vehicle Miles Traveled Tax (VMT) To achieve the same level of revenue increase, the fixed permile charge needs to be 1.5 cents/mile. Green VMT Charge vehicles with > 20 mpg fuel efficiency 1 cent/mile. Charge vehicle with < 20 mpg fuel efficiency 2.1 cents/mile. Congestion Pricing Charge road users a per-mile fee that ranges from 1 to 3.4 cents/mile based on level of congestion in their respective metropolitan statistical area (MSA) or non-msa area. 13

14 VMT Reduction by Household Income duction % VMT Re 0% -2% -4% -6% -8% -10% -12% -14% Household Income Gas Tax Green VMT Congestion Pricing VMT 14

15 Changes in Consumer Surplus: By Income Group Cha nge in CS as % of Income 0.0% -0.2% -0.4% -0.6% -0.8% -1.0% -1.2% Household Income Gas Tax VMT Green VMT Congestion Pricing i 15

16 Percentage Change in State Revenue Green VMT Fee

17 Percentage Change in State Revenue: Congestion Pricing

18 Changes in Consumer Surplus: By Level of Urbanization 0 Gas Tax VMT Green VMT Congestion Pricing Red duction in CS ($) Large Urban with Rail Large Urban without Rail Small Urban Rural 18

19 Environmental and Energy Policy Analysis #3 What is the impact of land use policies (e.g. high density, mixed development, neighborhood design) on travel behavior and VMT? 19

20 Defining Land Use Policy Variables Density Built environment variables Residential density (building sqft/area) Commercial density (building sqft/area) Industrial density (building sqft/area) Office density (building sqft/area) Mixed Use Six (J=6) land use types are considered: residential, commercial, industrial, i office, government and others. Average Block Size Distance to CBD Etc.

21 Methodological Issues and Research Design Four Issues and Research Direction Methodological Issues Causality (self-selection) l Spatial auto-correlation Inter-trip dependency (tour) Geographic scale Research Design Address these issues with careful control for travel attitude in modeling Compare metropolitan areas that have different land use characteristics and policies, using the same analytical approach

22 Multi-level Regression Results Modeling - Multilevel linear regression 1 km TAZ Mean SD MEAN SD Number of stops(tour) Pro Car Pro Car Pro Transit Residential Density Office Density Industrial Density Entropy Distance to CBD 6.90E E E E 07 Average Block Size 4.10E E 0910E E 0830E E Household socio economic and demographic variables not shown.

23 Summary of Findings Preliminary findings Land use factors have significant impacts on VMT People living in more compact cities, well mixed neighborhoods, areas with small block sizes drive less than others The geographic scale of analysis is important

24 Conclusions Air quality control and EPA s nonattainment designation have a statistically significant negative correlation with vehicle miles traveled. Congestion pricing and green transportation financing policies will only result in significant VMT reduction among low-income households. They have about the same regressivity as the existing fuel taxes. Land use policies implemented at the proper geographic scale can effectively influence travel behavior and VMT. NHTS and similar travel surveys provide important and often necessary information for environmental and energy policy analysis at various levels. 24

25 Thank you! Additional Questions and Comments Lei Zhang Assistant Professor Department of Civil & Environmental Engineering g University of Maryland College Park lei@umd.edu, Acknowledgement This research is funded by the FHWA, and the USDOT UTC Program through the Center for Integrated Transportation Systems Management (CITSM) at the University of Maryland. 25