Multi Agent Car Modeling-lessanim.ppt

Document Sample
Multi Agent Car Modeling-lessanim.ppt Powered By Docstoc
					Multi-Agent based Discrete Choice Modeling of the
Swiss Car Market
Presented to You by Jia Yan
              Yann Blumer
              Nadja Gross
     Content


           Introduction
           HES-Framework
           Instruments for the Car market
           Model
           Results
           Conclusion




Titel, <Datum>                               2
     Introduction
                                 Opportunities

                                 Low-Cost
                    Mobility
                                                 Availability



                               Fossil Energy

                    Climate
                    Change                       Scarcity

                                 Dependency


                                   Threats

Titel, <Datum>                                                  3
     Cars are Important Contributors


      Second-largest sector of energy consumption (OECD)‫‏‬

      Consumption is still growing despite more efficient engines
                    + 15-30 % in Passenger-Kilometers until 2030 ‫‏‬
                    Larger weight


      Life cycle
                    Main emissions during use
                    High lifetime




Titel, <Datum>                                                        4
     Possible Solutions


      Reduced Demand
                    Cap in mobility
                    Not popular/realistic
      Energy efficiency
                    Enhanced engines
                    Eco drive               Scope of this Paper
      CO2 efficiency                         How do people buy fuel efficient cars ?
                    Renewable energies       What policy instruments are suited?
                    New Technologies




Titel, <Datum>                                                                       5
     HES-Framework


      Supports Policy Makers by
                    Helping to map the situation
                    Identifying crucial points of intervention




Titel, <Datum>                                                    6
                  Individual
                 Government


                       2
                               1




           3

                      4

Titel, <Datum>                 7
                               Governement




           2              CO2 reduction
                                                                   1
Information needed from
     Individual Level      Regulatory
                                                         Climate
                            measures
                                                         Change



                                         Law / Policy
                                        Implementation




Titel, <Datum>                                                     8
                                         Individual


                              3

                                    Mobility

     Need for understanding
            this step             Mean of Transport


           4

                                                Purchase




Titel, <Datum>                                             9
     Content


           Introduction
           HES-Framework
           Instruments for the Car market
           Model
           Results
           Conclusion




Titel, <Datum>                               10
     Policy instrument in Car market

   Definition:

   Feebate system : fee + rebate

   Refunded-tax policies: incentive payments + tax increase
   (partial feebate system )‫‏‬

   Abusolute CO2 emission

   Relative energy efficiency : (in relation to curb weight)‫‏‬

Titel, <Datum>                                                  11
     Current EU strategy


                    manufacturers agreement



                    EU strategy 1995

      consumer information            fiscal measures



Titel, <Datum>                                          12
                          direct change
                           of demand



                          Effects of
             short-term                     long-term
                          incentive
               impact                        impact
                           schemes


                          indirect change
                            of demand


Titel, <Datum>                                          13
       Possible benefits and drawbacks


     Advantages:


      Internalization of external costs of road transport, which
                 leads to economic gains
      Promotion of fuel-efficient vehicles (the direct, intended
                 effect).
      Encourage of environment-friendly individual behavior



Titel, <Datum>                                                      14
           Disadvantages:

            More governmental regulation with corresponding
                 administrative costs
            Risk of lowering the efficiency of the market segment
            Market actors experience adaptation costs
            Consumers may feel overloaded with information




Titel, <Datum>                                                       15
           Energy-efficiency
           Labeling




                               http://www.simplewisdom.co.uk/graphics/enlabel.jpg

Titel, <Datum>                                                                      16
     Acceptance of feebates by the public

                                        Acceptance of feebates by the public
                 Acceptance




                               3.4

                                                    2.4                              2.4




                              Feebate     Purchase of CO2 emission             Higher fuel Price




Titel, <Datum>                                                                                     17
     Content


           Introduction
           HES-Framework
           Instruments for the Car market
           Model
           Results
           Conclusion




Titel, <Datum>                               18
     The Model – how does it work?

                      Input data
                                   Swiss Car fleet    1’000’000 car sales are
                 n Agents
                                                       simulated with the reference
                                      Population       scenario…
                    Choice set Ci                     …and then rerun with an
                                Model Parameters
                                                       alternative scenario.
                                                  
                       Choice i Transaction behavior   The output is compared to
                                                       the reference scenario.


                  Total purchases n



Titel, <Datum>                                                                   19
     Model Inputs (I)‫‏‬


      Car fleet
                    2089 car types, available in 2005 in Switzerland
                    They differ by engine size, CO2 emissions, fuel type, transmission
                     type, etc.


      Population
                    40 agent types constructed with a dataset
                    Based on socio-demographic categories (income, gender, household
                     type, age)‫‏‬
                    Allocation to different language regions of Switzerland
                    To each agent type a car type is allocated, based on survey data


Titel, <Datum>                                                                        20
     Model Inputs (II)‫‏‬


      Model Parameters
                    Utility function for each car type (make, price, fuel costs, vehicle
                     length, luggage capacity, acceleration time, lower/higher medium to
                     the price, homemarket)‫‏‬
                    Random sample drawing: average set of 25 alternatives


      Transaction Behavior
                    Differentiation between consumers who are experiencing constraints
                     and those who are not
                    E.g. someone who has had an automatic transmission, is likely to
                     replace it with a car with automatic transmission


Titel, <Datum>                                                                         21
     Modeling the Decision Process


      Two staged process
      1. Choice-set composition
                    Bounded rationality
                    Situational constraints
                    Biased decision rules


      2. Choice
                    Agent Preferences
                    Perception of monetary values




Titel, <Datum>                                       22
     Scenarios


      Refunded Tax
                    3 % higher purchase tax, 770 € per sale
                    “A” cars get it refunded via cash payment

      Feebates
                    “G” cars pay a fee of 2000 € (15.3% of new registrations)‫‏‬
                    “A” cars get a cash rebate (14.7% of new registrations)‫‏‬


      For each incentive, one scenario based on an absolute and
       one on a relative energy efficiency is calculated
      Both scenarios cover administration costs and are revenue-
       neutral
Titel, <Datum>                                                                    23
     Suitability of the Model


      Validation of the model with real car market dataset gives
            very good results

      IIA assumption not testified (some alternatives may be
            dependent)‫‏‬
                    Nested logit model
      Choice constraints do not change over time
                    Conservative approach, effects of incentives are even stronger
      Future research
                    Include the fact that not all agents purchase new cars


Titel, <Datum>                                                                        24
     Results for refunded tax schemes




Titel, <Datum>                          25
                  A notable shift towards small
                 diesel engine
                  Market share of “A”-labeled
                 cars increases
                  Minor market share reductions
                 for categories “B” to “G”
                  Differences between relative
                 and absolute policy bases are
                 present but not dominant
Titel, <Datum>                                26
Titel, <Datum>   27
                  A clear drop in market
                 share for the “G” category
                  The resulting CO2 reduction
                 effect differs by 10%




Titel, <Datum>                                28
     Discussion of the Results


      No Clear Picture
                    Both lead to substancial reduction
                    Feebate: More Effective
                    Tax: More Efficient

      Acceptance is crucial to success of policy measures
                    Revenue Neutral
                    Little impact on market & car size

      Only a first step
                    Effects will only last for about 10 years
                    Icebreaker
                    Reduction of consumption/Car size needed

Titel, <Datum>                                                   29
     Discussion of the model


      It is better suited than aggregated models

      Limitation: Extrapolation into the future

      Room for further enhancement
                    Synthetic Population
                    More Input data




Titel, <Datum>
Thank you for your attention.


       Questions??
     Sources


      Mueller, M., De Haan, P., “How do people buy fuel-efficient
       cars? Agent-based microsimulation of consumer choice of
       new passenger cars, part I: Model structure, simulation of
       bounded rationality, and model validation”
      De Haan, P., Mueller, M., Scholz, R. “How do people buy
       fuel-efficient cars? Agent-based microsimulation of consumer
       choice of new passenger cars, part II: Forecasting effects of
       CO2 based feebates on new car registrations”
      Peters et al. “How do customers purchase cars?
       Psychological determinants of the importance of fuel
       consumption”
Titel, <Datum>                                                  32

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:15
posted:8/3/2011
language:English
pages:32