A simultaneous school choice and commuting mode choice model

1
A simultaneous school choice and commuting mode
choice model
Faculty of Transportation and Traffic Sciences
Friedrich List Institute of Logistics and
Aviation
Chair of Business Administration, Transport and
Logistics
Sven Müller, Stephan Brunow

• RGS/IBG Annual International Conference 2007
• London, UK August, 28.-31. 2007

2
Introduction

• declining/increasing enrolment yields school
  closures/openings

• modelling of school choice and travel-to-school
  mode choice
• influencing factors/variables
• choice probabilities

• BUT simultaneous instead of sequential to
  discover the coherence between mode and school
  choice

3
Approach
why nested logit (NL)?

• NL allows differential cross-elasticity among
  alternatives

• represents joint decision making (school and
  mode choice)

• allocates each alternative to a set of nests

• recent research
• Hensher and Reyes (2000) estimate nested logit
  for models of activity participation, trip
  chaining, and mode choice
• Wen and Koppelman (2001) found the NL was
  superior to other formulations for intraurban
  mode-choice problems

4
Methodology
nesting structure (tree)
Koppelmann and Sethi (2002) e.g.
nest level 3 (K)
structural parameter
nest level 2 (J)
nest level 1 (I)
111
112
212
312
122
113
213
123
223
elemental alternatives
5
Methodology
NL product of a series of multinomial logit
(MNL) choice models see McFadden (1978)
6
Data
7
Data
closest school of each profile-group enters the
choice set for every student
8
Data
.50
.50
walking density
cycling density
.25
.25
travel-time min
travel-time min
.25
.50
car density
public trans. density
.25
travel-time min
travel-time min
9
Application
level car availability
level mode choice
level school choice
10
Results
11
Summary

• simultaneous approach (nesting structure / NL)
  is adequate

• fair weather increases choice probability for
  cycling

• girls prefer safer modes like car/bus

• schools of profile group 2 (math, music) will
  show a higher percentage of students
  walking/cycling

• modes car/bus are most likely for public magnet
  schools offering 'languages'

• future work control for misspecification
  estimate a multinomial simultaneous structural
  equation system (SSES) with latent utility

12
Faculty of Transportation and Traffic Sciences
Friedrich List Institute for Transport and
Economics
Chair of Business Administration, Transport and
Logistics
Sven Mueller, M.A.Dresden University of
TechnologyInstitute of Logistics and
AviationChair of Business Administration,
Transport and Logisticssven.mueller_at_tu-dresden.d
e http//www.vbl-dresden.de
13
Data
"music" density
"magnet school" density
travel-time min
travel-time min
"language" density
"math" density
travel-time min
travel-time min