Transportation Engineering Laboratory, Hiroshima University 39-1 Development of Car Ownership and Use Model Considering Intra-household Interaction Junyi.

Transportation Engineering Laboratory, Hiroshima University 39-1

Development of Car Ownership and Use Model Considering Intra-household Interaction

Junyi ZHANG, Akimasa FUJIWARA

Masashi Kuwano

Transportation Engineering Laboratory (TEL)

Graduate School for International Development and Cooperation (IDEC), Hiroshima University, JAPAN

SAKURA Meeting, Japan-France Integrated Action Program, INRETS, July 2, 2004


Motivation

Household decision making

Car ownership in local cities in Japan30% (23,664,000 / 76,893000)

Car ownership in developing countries


Urban Air Quality Management Group

Urban EcosystemManagement Group

Socio-economicAssessment Team

Transportationindices

Environmentalindices

Land useindices

Capacityindices by

actor

Institutionindices

Carrying-Capacity

Emission production

Capacityindices by

actor

Capacityevaluation model

Land use andtransportation model

Environmental Management in Developing Country: 21 Century COE (Center of Excellence) Program


Development of Evaluation Framework for Land Use and Transportation Policies from the Environmental Perspective

GISy3 = h (Trip distribution, inter-zonal level-of-service)

y2 = g (Trip generation and attraction, inter-zonal travel utility）

y4 = i (Car traffic volume, link capacity, link free speed)

Trip generation and attraction

Trip distribution

Distribution of residence and.employment, transportation network

y1 = f (Population, car ownership, employment，travel pattern）

Link traffic volume and speed

Car traffic volume

Land use model

EnvironmentalIntensity

Trip generation and attraction model

Trip distribution model

Modal split model

Traffic assignment model

Travel accessibility

Travel level-of-service

Travel utility

Car ownership model

Emission productionEnergy consumption

Urban formpreference model

Population and economic synthesizers

Environmental evaluationModel

PolicyEvaluationPolicyEvaluation

GISy3 = h (Trip distribution, inter-zonal level-of-service)

y2 = g (Trip generation and attraction, inter-zonal travel utility）

y4 = i (Car traffic volume, link capacity, link free speed)

Trip generation and attraction

Trip distribution

Distribution of residence and.employment, transportation network

y1 = f (Population, car ownership, employment，travel pattern）

Link traffic volume and speed

Car traffic volume

Land use model

EnvironmentalIntensity

Trip generation and attraction model

Trip distribution model

Modal split model

Traffic assignment model

Travel accessibility

Travel level-of-service

Travel utility

Car ownership model

Emission productionEnergy consumption

Urban formpreference model

Population and economic synthesizersPopulation and economic synthesizers

Environmental evaluationModel

PolicyEvaluationPolicyEvaluation


Outline

1. Review of existing research

2. Methodological issues

3. Development of discrete choice models with inter-agent interaction

4. Summary of survey data

5. Estimation results and discussion

6. Conclusions and future research issues


1. Existing research

Non-IIA discrete choice model

Exogenous choice set Endogenous choice set

Flexible variance-covariance matrix

MNP, HEV, Mixed MNL & MNP, GEV

GenL

Context dependence Mother Logit, Dogit, SP, Context-sensitive spatial choice model

r_MNL, r_NL, r_QNL

Hierarchical or sequential decision structure

NL, GNL, CNL, OGEV, PD, NPCL, Network GEV, EBA

r_NL

Discrete choice models: examples



Focus on household, but not individual members

Focus on both household and its members

Husband & WifeHusband & Wife

Without intra-householdInteraction

With intra-householdInteraction

Type of household-related research


Thorndike, R.L. (1938) On what type of task will a group do well? Journal of Abnormal and Social Psychology 33, 409–413.




Arrow K.J. (1950) A Difficulty in the Concept of Social Welfare

Arrow K.J. (1951a) Mathematical Models in the Social Sciences

Arrow K.J. (1951b) Social Choice and Individual Values

Festinger L. (1954) A Theory of Social Comparison Processes

Harsanyi J.C. (1955) Cardinal Welfare, Individualistic Ethics, and Interpersonal Comparisons of Utility

Lorge I. and Solomon H. (1955) Two Models of Group Behavior in the Solution of Eureka-type Problems

Luce R.D. and Raiffa H. (1957) Games and Decisions

Nash J.F. (1950) The Bargaining Problem

Nash J.F. (1953) Two-person Cooperative Games

Samuelson P.A. (1956) Social Indifference Curves

Siegel S. (1957) Level of Aspiration and Group Decision Making

Simon H.A. (1955) A Behavioral Model of Rational Choice

Historical evidences: 1950s


Shopping behavior Wives did most of grocery shopping, with an awareness of products and brands that their families liked. Husbands and teenagers were frequently involved in new or different brands. The growing importance of men as buyers

Tourism Husband: mention initial idea to take a trip, suggest a

destination, and select an airline

Mutual decision: decision on where to go


Historical evidences: 1960s~1970s


Housing and automobile (1) Buying home: husband (price range, whether to move)

wife (no. of bedrooms and other house features)

(2) Automobile: for the make: husbands dominant households (60%)

for the color: husband dominant households (25%)

husband > wife

The growing involvement of women in family decisions


Historical evidences: 1960s~1970s



In marketing research, studies of husband-wife influence have been justified to

(1) select the proper respondent in consumer research surveys,

(2) determine the content of advertising messages, (3) select advertising media,

(4) guide product designers to include features that appeal to

those who are most influential in the purchase decision, and

(5) assist in the location of retail outlets

Applications of household decision-making mechanisms



Goals Strategy Ways of implementing

“Consensus”(Family members agree about goals)

Role structure “The Specialist”

Budgets “The controller”

Problem solving

“The expert”

“The better solution”

“The multiple Purchase”

“Accommodation”(Family members disagree about goals) Persuasion

“The irresponsible critic”

“Feminine intuition”

“Shopping Together”

“Coercion”

“Coalitions”

Bargaining

“The next purchase”

“The impulse purchase”

“The procrastinator”

By Davis H.L. (1976 )

Household Decision rules and heterogeneity


Kirchler E. (1988) Household Economic Decision Making, in Handbook of Economic Psychology, van Raaij W.F., van Veldhoven G.M. and Warneryd K.E. (eds.), Kluwer Academic Publishers.


A household will seek to minimize social and economic costs in decision situations by trying to make an optimal choice after passing through a commonly satisfying interaction process.

Household decision-making: Decision rules


Relative Influence

Young childless couple

Couple with child under 6

Couple with dependent child

Couple with independent child

Old childless couple

Husband

Wife

Child

100 %

By Cosenza R.M. and D.L. Davis (1981)


Household decision-making: Life cycle


1. Existing research (transportation)

Szalai (1972): The use of time: Daily activities of urban and suburban populations in twelve countries. The Hague: Mouton. 1) Large sample size: 30,000 time budgets drawn from 12 nations 2) 96 activities 3) 24-hour diary (types of activities (both primary and secondary activities), time, place and duration)

shows the relevance to household decision-making.


Household members interact in making decisions about the different activities that they perform and the related allocation of time.

(1) Joint activity participation(2) Household resource allocation(3) Task allocation(4) Role specification



Intra-household (temporal and spatial) interaction and inter-dependency among activities

Wife Husband

home restaurant

office

business

time Shared activity

Allocated activity

In-home activity

Independent activity

supermarket

Travel


1. Modeling approaches

(1) LISREL model (2) RUM (Nested-type logit model) (3) Mathematical programming model (4) Rule-based approach (5) Group decision-making theory


2. Survey methods

(1) Stated preference (2) Interactive agency choice experiments + Game theory


Wind (1976), Rogers (1976)

The view of consumers as individual decision makers is still very much alive despite commonsense observations that the family is the relevant decision-making unit and a growing research interest in the field of marketing research.

Current situation in the field of transportation: Most of the existing models typically assume an individual decision-making process.




h : group (e.g., a household) i : group member j : alternative

Gumbel distribution

j hnjhijj1h

hnjhijj1hhj )v,...,v,...,v(fexp

)v,...,v,...,v(fexpP

hj)v,...,v,...,v(f

)u,...,u,...,u(fGUFMaximize

hnjhijj1h

hnjhijj1h

Principle of random group utility maximization: A concept of meta-utility



(1) Mother logit model (McFadden, et al, 1977)

}vv{exp

}vv{expP

k j'j 'ikkik

j'ij 'jjijij

}v{exp

}v{exp

}v{exp

}v{expP

kt

0s siktks

t0s sijtjs

k ik

ijijt

j j'j t'ijt,ijt'ijjijt

j'j t'ijijtt'ijjijtijt )vv(wr

)vv(wrP

(2) Dynamic GEV model (Swait, et al, 2004)

(3) r_MNL model (Zhang et al, 2004)

Choice models based on meta-utility



Scale measure: 3, 5, 10-point scale

Husband decided Jointly decided Wife decided

-------------------------------------------------------------

Interaction-based measure: Power

SP survey

Game theory

Endogenous estimation based on the attributes of decision makers

Measurement of relative influence of household members



Car 1 Car 3

Member 1 Member 2

Car 2

Who makes the decisions?



Multi-linear type

ni i'i

n1i

)uuww(uwGUFi'ii'iii

Iso-elastic type

i

1iiuw

11

GUF

Group utility functions


ni i'i

n1i

)uuww(uwGUFi'ii'iii

1xx to Subject

xxU Maximize

21

21

0xxL

0xxL

)xx1(xxL

12

21

2121

2/1xx 21

i'iuu can be used to reflect household members’

concern for achieving equality of utilities


Multi-linear type of group utility functions


(1) Additive-type utility function

(2) Compromise-type utility function

(3) Capitulation-type utility function

n1i iiuwGUF

n1i i nuGUF

ii' i'1

in

1i ii

n1i ii

u)1n(u,uwGUF

uwGUF


Special cases ofmulti-linear household utility functions


i

1iiua

11

HUFMaximize

α → ∞ Max{minimum utility}

α ＞1 Max{small utility}

α → 1 Maximum Average

α ＝0 Average Maximum

α ＜0 Max{large utility}

I,...,2,1i|uminmaxHUF i

iwi iuHUF

i iiuwHUF


Iso-elastic type of group utility functions



Similarity and dissimilarity

Multi-linear type

ni i'i

n1i

)uuww(uwGUFi'ii'iii

Iso-elastic type

i

1iiuw

11

GUF

Comparisons of multi-linear andiso-elastic group utility functions


< Iso-elastic Group Utility Function >

3. Development of discrete Choice Model with inter-agency interaction

'j i1

'hij'hij

i1

hijhij

hjvw

11

exp

vw1

1exp

P

hji

1hijhij

i

1hijhij vw

11

uw1

1GUF

G_MNL model


< Multi-linear Group Utility Function >

j i 'i j'hij'ihijiji hijij

i 'i j'hij'ihijiji hijijhj vwvwvwexp

vwvwvwexpP


hji 'i j'hij'ihijiji hijij vwvwvwGUF

G_MNL model



G_NL model

'm'm'm

mm

'dm'd'd

dmdmm|ddm )'vv(e

)'vv(e)vv(e

)vv(ePPP

10and))vv(eln('v'd

m'd'dm


HouseholdHousehold

Social network

The second car

The first carBuy?

Buy what car?

New or used car

Use

ScrapRenewal

Buy?

Buy what car?

Use

ScrapRenewal

Future expectation

Analysis framework of household car ownership and use

G_MNL

G_NL



Hiroshima Region

Survey Area



Survey Area



Car46%

Transit28%

Motor cycle7%

Walk/bicycle19%

3,4times/week

15%

1,2times/week

15%

3,4times/mon.

4%

Almosteveryday

61%

severaltimes/year

2%

1,2times/mon.

3%

Car use frequency

Modal share

Survey Results



0% 20% 40% 60% 80% 100%

LEV

Other cars

Previous car just before buying the current car

Current car

<= 660cc

660cc~1000cc

1001cc~1500cc 1500cc~2000cc

2001cc~2500cc

>=2500cc

Survey Results


5. Estimation results and discussions

G_MNL model



The first car

Head ofHousehold

Other member

<=1500cc

> 1500cc

genderage -job -license

main user wifegenderage -job -license

v11 v21

w11

car price income

Inter-dependency ?

× 1-w11 ×+

Heterogeneity

G_MNL model



G_ NL model



The first car

Head ofHousehold

Other member

<=1500cc

> 1500cc

gender +age +joblicenseincome -

main user childgender +age +joblicenseincome -

v11 v21

w11 car price +

Inter-dependency -

× 1-w11 ×

The second car

<=1500cc

> 1500cc

gender +age +joblicenseincome -

main user childgender +age +joblicenseincome -

V12 v22

w12 car price +

Inter-dependency -

× 1-w12 ×

Inclusive value (0,1)

McFadden’s Rho-squared=0.4457

Sample size = 114

G_NL model


Conclusions andFuture Research Issues

1. Development of a new class of discrete choice models with inter-agency interaction (G_MNL, G_NL models) Multi-linear group utility Iso-elastic group utility

• Estimation of other types of G_models• Comprehensive model system for household car

ownership and use incorporating the influence of group decision-making mechanisms

Transportation Engineering Laboratory, Hiroshima University 39-1 Development of Car Ownership and Use Model Considering Intra-household Interaction Junyi.

Documents

hiroshima university39

transportation policies

existing research slide

related research slide

existing research arrow

existing research focus

use model

b social choice