walker by liwenting

VIEWS: 16 PAGES: 26

									LESSONS FROM BEHAVIORAL ECONOMICS
FOR TRANSPORTATION PLANNING

JOAN L WALKER
UC BERKELEY




        Celebrating Ryuichi Kitamura, UC Davis 2009
    Example: Perceptions versus Reality
2


      Whatpercent of drivers thing they are above
      average?
        90%

      What  percentage of Professors think they are better
      than the average professor?
        94%
    Kitamura’s Legacy in TDM
5


       Emphasizing “behavior” as the forefront of
        behavioral modeling
         Heterogeneity

         Activity-based

         Dynamics

         Psychological     principles
           Attitudes,   lifestyle, prospect theory, behavioral change, …
       Approached with econometric rigor
        and with an eye to application
    Outline
6


       Introduction
       Experiments from Behavioral Economics
       Incorporating lessons in transportation
    Motivation
7


       Key to transport is to influence & predict behavior
         Methods   largely rooted in classic microeconomics
       Behavioral economics
         Cross between psychology and economics
         Focus on what really influences decisions
          as opposed to what we think influences them
         Clever experiments

       2008 books
         Dan  Arieli’s Predictably Irrational
         Richard Thaler and Cass Sunstein’s Nudge
    Behavioral Economics in Transportation
8




       Survey design
       Prospect theory
       Attitudes

       Lessons on time and cost
       Importance of experience
       Influence of information and social forces
    Outline
9


       Introduction
       Experiments from Behavioral Economics
       Incorporating lessons in transportation
     Relativity and Value of Time
10


        Two errands:
          Buy a new pen
           $15 where you are, $8 at store fifteen minutes away
           Do you go to other store? Most people say YES.
          Buy a new suit
           $455 where you are, $448 at store fifteen minutes away
           Do you go to other store? Most people say NO.
        Lessons
          Everythingis relative
          Impacts most important variable in transportation
                                             Kahnemann and Tversky (1981)
                                                            Ariely (2008)
     Cost and the Power of FREE!
11


        Chocolate sales
          Lindttruffles      $0.15  73%          $0.14  31%
          Hersheys kisses    $0.01  27%           FREE!  69%
        Lessons
          Zero  an emotional hot button,
            a source of irrational excitement.
          Alternative fuel vehicles?
           Eliminate registration & inspection fees, to create FREE!

        In general, easy to manipulate willingness to pay.
                                                              Ariely (2008)
     Experience and Memory
12

                                                     Preference
                  TREATMENT A
                                                     of B to A due
                                                     to peak-end
                                                     rule:
     PAIN LEVEL




                                                     weigh the
                                 TREATMENT B
                                                     peak and
                                                     end more
                                                     than the rest
                                                     of the
                                                     experience
                                         TIME


                  Kahneman, Fredrickson, Schreiber, Redelmeier (1993)
                                  Redelmeier, Katz, Kahneman (2003)
     Information and feedback
13


        Efforts of Southern California Edison to encourage
         energy conservation
          Emailsand text messages regarding energy use
            No effect
          Ambient Orb: red during high energy, green during low
            40% reduction during peak periods
                                                    Thomson (2007)
        Lessons
          Peoplehave a hard time understanding impacts
          Make them “visible”
     Conforming to Social Norms
14


        Minnesota tax compliance
          Taxes go to good works                INSIGNIFICANT
          Threats                               INSIGNIFICANT
          Access to help                        INSIGNIFICANT
          90% of Minnesotans complied           SIGNIFICANT
                                                       Coleman (1996)
                                            Thayler and Sunstein (2008)
        Lessons
                desire to conform to social norms
          Strong

          What we think are strong behavioral influences are not
     Social norms and Visualization
15


        Energy use study in CA                   (Schultz et al., 2007)
          Feedback approach 1:              High users reduce,
           Household energy use              Low users increase
           Avg energy use in neighborhood
          Feedback approach 2:              High users reduce more,
           Same as above, but with         Low users don’t change
        Lessons
          Don’t let people know they are better than social norm
          Arguments based on energy conservation, national
           security, economic growth, and environmental protection
     Influential people
16


        Littering problem in Texas
          Massivecampaign based on “Civic Duty”
            No effect
                  players & “Don’t mess with Texas!”
          Football
            29% reduction of visible litter in first year
            72% reduction in six years
        Lessons
         A  few influential people, offering strong signals about
           appropriate behavior, can have a strong effect.
                                               Thayler and Sunstein (2008)
     Chaos
17


        Artificial music market
          14,000  participants in 8 worlds
          Users download songs and rate songs

          Success of songs
            Variedacross worlds and was unpredictable
            Depended heavily on the choices of the first downloaders

        Lesson
          Small interventions at key stage can produce large
           variations
                             Salganik et al. (2006), Thaler and Sunstein (2008)
     Outline
18


        Introduction
        Experiments from Behavioral Economics
        Incorporating lessons in transportation
     Implications for Transportation
19


        Do the lessons hold for transportation decisions?
        What to do?
          Useto encourage sustainable behaviors
          Adapt models
                      tools do not require rationality
            Statistical
            Sometimes relatively simple adjustments suffice
                    Utility specification (e.g., non-linearities of time/cost,
                     inclusion of reference points, field effect for social influences)
                    Data solutions (e.g., data on experience)
            Sometimes        more complex
                    Modeling of constraints and latent variables
Neighborhood   Neighborhood
     A              B
     Residential Location Model with Lifestyle
21


        Hypothesis
          Lifestyle preferences exist (e.g., suburb vs. urban)
          Lifestyle differences lead to differences in considerations,
           criterion, and preferences for residential location choices
        Infer “lifestyle” preferences from choice behavior
         using latent class choice model
          Latent classes = lifestyle
          Choice model = location decisions
                                        (Alternative 1)        (Alternative 2)        (Alternative 3)         (Alternative 4)      (Alt. 5)

                                            Buy                   Buy                   Rent                    Rent
                                       Single Family          Multi-Family          Single Family            Multi-Family
22
               Type of Dwelling :        single house           apartment          duplex / row house         condominium
                Residence Size :        < 1,000 sq. ft.      500-1,000 sq. ft.    1,500 - 2,000 sq. ft.       < 500 sq. ft.        Move
                        Lot Size :      < 5,000 sq. ft.             n/a           5,000 - 7,500 sq. ft.            n/a              out
                         Parking :    street parking only   street parking only   driveway, no garage     reserved, uncovered      of the
         Price or Monthly Rents :          < $75K             $50K - $100K              > $1,200              $300 - $600          Metro
               Community Type :           mixed use             mixed use                 rural                   urban            Area
                    Housing Mix :    mostly single family   mostly multi-family   mostly multi-family      mostly multi-family
           Age of Development :          10-15 years             0-5 years            10-15 years              0 - 5 years
  Mix of Residential Ownership :         mostly own             mostly own             mostly rent             mostly own
 Shops/Services/Entertainment :       community square          basic shops        community square       basic, specialty shops
                     Local Parks :           none                   yes                   none                    none
                   Bicycle Paths :           none                   yes                    yes                     yes
                  School Quality :        very good             very good                  fair                    fair
          Neighborhood Safety :            average               average                average                 average
Shopping Prices Relative to Avg :         20% more              20% more                  same                 10% more
         Walking Time to Shops :        20-30 minutes         20-30 minutes          < 10 minutes            10 - 20 minutes
Bus Fare, Travel Time to Shops :     $1.00, 15-20 minutes   $1.00, > 20 minutes   $0.50, 5 - 10 minutes    $0.50, < 5 minutes
   Travel Time to Work by Auto :        > 20 minutes          15-20 minutes         15 - 20 minutes           < 10 minutes
 Travel Time to Work by Transit :       > 45 minutes          30-45 minutes         30 - 45 minutes          15 - 30 minutes
     Lifestyle Segmentation
23



                    Latent Segment 1 
          suburban, school, auto… (shopping)
            affluent, more established families



                   Latent Segment 2
                  transit, school… (suburban)
                  less affluent, younger families




                    Latent Segment 3 
           high density, urban activity… (car)
              older, non-family, professionals
       Modeling latent psychological
24
       constructs
                     Characteristics of Respondents S
                       Attributes of Alternatives Z



      Attitudinal     Attitudes          Perceptions      Perceptual
     Indicators IS       S*                  Z*          Indicators IZ



                              Preferences                    Stated
                                   U                    Preferences ySP




                               Revealed
                            Preferences yRP
     Summary
25


        Need to do experiments for transportation
        Potential for influencing behavior
          Information    and Feedback
            Powerful   and even more so with visualization
          Social   influences
            Humans   are easily influenced by others and like to conform
                 Obesity, happiness, study habits… transportation!
        Implications for modeling
          Random utility framework good starting point and can
           be expanded to reflect experimental findings
     Building on Kitamura’s Legacy
26


        Comprehensive Model System for Policy Analysis
          Activity-basedmodels
                                     Activity Space
          Behavioral realism
                                               Physical
                                                Activity
                            Person          Vehicle
                                           Ownership
                                                 Residential
                                                  Location
                                          Mode Choices

                                      Energy

								
To top