You are on page 1of 35

1ho Lconomic 1hooiy of UiLan 1iaffic Congoslion

:
A Micioscopic Iosoaich Agonoa
Iichaio Ainoll
Dopailmonl of Lconomics
Loslon Collogo
Choslnul Hill. MA 02467
1-617-òò2-8674
iichaio.ainoll^Lc.oou
July 2001
Foi piosonlalion al CLSifo Vonico Summoi Inslilulo 2001. Woikshop on Lnyiionmonlal Lconomics
ano lho Lconomics of Congoslion: Coping wilh Lxloinalilios. Vonico Inloinalional Uniyoisily. San Soiyolo.
July 18-10. 2001. 1his papoi may Lo founo on-lino al hllp:´´FMWWW.Lc.oou´LC-V´Ainoll.fac.hlml
2
SO
1. Introduction
macroscopic
method
The Economic Theory of Urban Traffic Congestion:
A Microscopic Research Agenda
1ho lhomo of my lalk looay is lhal wo lho communily of uiLan lianspoil oconomisls
ioly oxcossiyoly on lho canonical moool of uiLan liaffic congoslion. which
has causoo us lo placo oxcossiyo omphasis on congoslion piicing as a policy lool lo impioyo
lho officioncy of uiLan liayol. UiLan liaffic congoslion is lho oulcomo of oocisions maoo on
many moio maigins of choico lhan oyon lho mosl sophislicaloo yaiianls of lho canonical
moools capluio. 1o incioaso lho usofulnoss of oui lhooiy in piaclical policy applicalion.
wo nooo lo ooyolop a poilfolio of moools lhal capluio lho omilloo maigins of choico.
Applioo miciooconomic lhooiy has Loon so succossful laigoly Locauso of ils .
which onlails woiking wilh simplo. concoplually consislonl moools Lasoo on maximizing
Lohayioi. 1his molhoo olucioalos Lasic piinciplos ano focusos on ossonlials. aLsliacling
fiom oisliacling oolail. Ioconl policy succossos of lhis appioach incluoo lho ooyolopmonl
of a maikol foi . lho oxpanooo uso of auclions foi iosouico allocalion wilh small
numLois of Luyois ano sollois. ano lho applicalion of inconliyo conliacling lo puLlic
ulililios (Laffonl ano 1iiola (1008).
Applicalion of lhis molhoo in lho conloxl of uiLan liaffic congoslion has loo uiLan
lianspoil oconomisls lo aoyocalo congoslion piicing iooally poifocl congoslion piic-
ing Lul ioalislically pailial oi impoifocl congoslion piicing. Lul oui ploas foi congoslion
piicing hayo Loon singulaily unsuccossful. Wilh lho aiguaLlo oxcoplion of Singapoio. con-
goslion piicing of uiLan liayol has nol Loon implomonloo anywhoio (Small ano Gomoz-
ILanoz (1008)). ano many juiisoiclions (o.g. CamLiiogo. Hong Kong (Loiins (1086)).
Slockholm (Ahlsliano (1081)).ano lho Ianoslaao) hayo Lackoo oown fiom plans lo inlio-
ouco congoslion piicing on oyon an oxpoiimonlal Lasis. Why congoslion piicing of uiLan
aulo liayol has Loon iocoiyoo Ly policy makois wilh so lilllo onlhusiasm has Loon much
2
,
on urban roads
alternatives to congestion pricing
microscopic
oiscussoo in lho liloialuio (Gomoz-ILanoz (1002). Jonos (1008)). My own yiow is lhal
wo uiLan lianspoil oconomisls hayo Loon oocoiyoo Ly lho simplicily of oui moools inlo
gioally unooioslimaling lho cosls ano piaclical oifficullios of implomonling congoslion
piicing . ano lhal policy makois hayo wisoly iosisloo ooing so. Lo lhal as
il may. lho social ioluins lo oui laLois will. I Lolioyo Lo gioaloi if wo ooyolo moio of oui
offoils lowaios oxamining . Ano lhis. I shall aiguo. will
ioouiio lho ooyolopmonl of moio moools of uiLan liayol. which pioyioo a
moio oolailoo oosciiplion of uiLan liaffic low ano uiLan liayol oocisions lhan is oono in
lho canonical macioscopic moool: whal I hayo in mino will hopofully Lo claiiloo in lho
lalk.
In soclion 2. I shall pioyioo a Liiof skolch of lho ooyolopmonl of uiLan lianspoil
oconomic lhooiy. as woll as ils cuiionl slalo. ano lhon pioyioo a ciiliouo of il. 1hon. in
soclion 8. I shall oiscuss a soloclion of iosoaich lopics. which will logolhoi illuslialo how
I lhink uiLan lianspoil oconomic lhooiy shoulo Lo io-oiionloo lo mako il moio usoful foi
policy makois.
8

Q,w
( ) ( )
( ) ( )
max ( ) ( ) ( )
D AC
Q,
MSC
MSC AC

p w
D p c Q, w ,
B Q , K w .
B Q Qc Q, w K w ,
2. Current Urban Transport Economic Theory and Its Applica-
tion
1ho ooyolopmonl of uiLan lianspoil oconomic lhooiy has onlailoo lho giaoual olaLoia-
lion of a canonical moool. 1ho Lasic moool (Lockmann. McGuiio. ano Winslon (10ò6))
oxaminos liayol on a poinl-inpul. poinl-oulpul ioao. Inoiyioual oiiyois aio ioonlical. ano
lho only oconomic oocision oach oiiyoi makos is liip fioouoncy. Congoslion is capluioo Ly
a congoslion cosl funclion which iolalos liip cosl lo liaffic yolumo ano capacily.
INSLI1 FIG. 1 HLIL
Figuio 1 giyos a oiagiammalic iopiosonlalion of lho Lasic moool (Wallois (1061).
Mohiing (1076 )). wilh capacily lxoo. is lho oomano cuiyo: iolalos oach oiiyoi`s
liip cosl lo liaffic yolumo. ano is yaiiously iofoiioo lo as liip cosl. ayoiago cosl.
usoi cosl. ano maiginal piiyalo cosl: ano is lho maiginal social cosl of a liip. In
lho aLsonco of goyoinmonl inloiyonlion. lho oouiliLiium occuis whoio oomano inloisocls
ayoiago cosl. 1ho oplimum occuis whoio oomano inloisocls maiginal social cosl. 1ho
yoilical oislanco Lolwoon ano is lho congoslion oxloinalily cosl. 1ho minimal
goyoinmonl inloiyonlion nooooo lo ooconlializo lho social oplimum is lho imposilion of a
congoslion loll ooual lo lho congoslion oxloinalily cosl. oyalualoo al lho social oplimum.
.
1ho samo moool may Lo oosciiLoo algoLiaically using oilhoi social suiplus oi social
wolfaio analysis (Mayoios ano Iioosl (1007)). Whoio is liip piico ano capacily.
lho oomano funclion is . (shoil-iun) ayoiago cosl lho social Lonoll func-
lion ano lho capacily consliuclion cosl funclion 1ho oiiocl. long-iun social
suiplus maximizalion pioLlom may lhon Lo wiillon as
(1)
4

Q,p,,w
p,w
Q,p,,n
: ( ) ( ) +
( )
= 0
: = 0
max ( ( ) ) + ( ( ))
= ( )
= + ( )
max ( ( )) ( ) ( ( ) ) ( )
= ( ) =
( )
max ( ) + ( )
= ( )
= ( )
= 0
Q B Q c Q, w Q
∂c Q, w
∂Q
w Q
∂c
∂w
K .
B Q Qp Q K w
Q D p
p c Q, w ,
B D p D p c D p , w K w .
Q D p p
c Q, w
B Q Qp Q K w
Q D p
p c Q, w

which giyos lho oplimalily conoilions:
(2a)
(2L)
Lo.(2a) slalos lhal oplimal liaffic yolumo is such lhal lho maiginal social Lonoll of a
liip oouals maiginal social cosl. which oouals shoil-iun ayoiago cosl plus lho congoslion
oxloinalily cosl. Lo.(2L) slalos lhal oplimal ioao wiolh is such lhal lho maiginal social
Lonoll fiom ioao oxpansion. lho ioouclion in liayol cosls holoing liaffic yolumo lxoo.
oouals lho maiginal consliuclion cosl. 1ho inoiiocl social suiplus maximizalion pioLlom.
whoio inoiyiouals oocioo on liip fioouoncy Lasoo on liip piico ano lho goyoinmonl oocioos
on lho congoslion loll. wilh liip piico ooualling ayoiago cosl plus lho loll. is
s.l. i) (8)
ii)
which iooucos lo
(4)
1ho coiiosponoing lisl-oiooi conoilions aio lho samo as lhoso of lho oiiocl maximizalion
pioLlom.
1ho no-loll oouiliLiium may Lo chaiacloiizoo as lho solulion lo ano
. oi as lho solulion lo lho following consliainoo maximizalion pioLlom
s.l. i)
ii)
iii)
ò
D, p, w M
M
route choice modal choice
congestion interaction
1ho Lasic moool has Loon oniichoo lo accounl foi olhoi maigins of choico. Laily on lho
moool was oxlonooo lo lioal ano . Wilh iospocl lo ioulo choico.
an inoiyioual choosos his ioulo on a nolwoik. fiom a giyon oiigin lo a giyon ooslinalion.
so as lo minimizo liip piico lho gonoializoo Waioiop piinciplo. Wilh iospocl lo mooal
choico. whon mooos aio poifocl suLslilulos in oomano. lho samo piinciplo applios. oxcopl
lhal lhoio is Lolwoon Lusos ano cais. Lul no congoslion inloiaclion
Lolwoon cais on oiffoionl links. Whon mooos aio nol poifocl suLslilulos in oomano. lho
maximizalion pioLlom is oxlonooo lo mulliplo mooos. wilh ano -oimonsional
yoclois. whoio is lho numLoi of mooos. 1ho moool was also oxlonooo oaily on lo
lioal usoi holoiogonoily (Sliolz (106ò )). Inoiyiouals fiom oiffoionl gioups hayo oiffoionl
Lonoll ano ayoiago cosl funclions. Ano whilo nol analylically nocossaiy. il is almosl
always assumoo lhal inoiyiouals fiom oiffoionl gioups onloi lho congoslion cosl funclions
symmoliically.
Unooi lhis assumplion. in lho moool oxlonooo lo lioal ioulo choico. mooal choico. ano
usoi holoiogonoily. lho full oplimum can Lo ooconlializoo Ly applying an anonymous loll
on oach link in lho nolwoik ooual lo lhal link`s congoslion oxloinalily cosl. Fuilhoimoio.
lho maiginal social Lonoll of capacily on oach link can Lo compuloo sliaighlfoiwaioly as
lho liayol cosl sayings on lhal link. liaffic lxoo. wilhoul consiooialion of how liayollois
swilch mooos ano ioulos in iosponso lo lho inciomonlal capacily oxpansion. 1hoso aio yoiy
impoilanl iosulls sinco lhoy inoicalo lhal. in loims of lho moool. yoiy lilllo infoimalion
is nooooo lo ooconlializo lho lisl-Losl oplimum. All lhal is ioouiioo is lo moasuio lho
link congoslion oxloinalily cosls. which ioouiios knowloogo of only lho link congoslion
funclions ano liaffic loyols. No infoimalion is nooooo on lho ioonlily of liayollois oi on
lhoii oomano funclions. Il is lhoiofoio oasy lo unooislano why oconomisls hayo pushoo
so haio foi poifocl congoslion piicing.
1ho aLoyo moools hayo Loon oxlonooo lo lioal fioighl liaffic. On lho assumplion
6
lhal a liuck conliiLulos lo congoslion in lho samo way as a lxoo numLoi of cais. lisl-
Losl congoslion lolling wilh liucks paying lho cai-oouiyalonls loll conlinuos lo
ooconlializo lho lisl-Losl oplimum.
1ho aLoyo moools hayo also Loon applioo lo lioal a iango of socono-Losl pioLloms.
Loyy-LamLoil (1068). Maichaio (1068). Shoiman (1071) ano Loiliano (1077) oxaminoo
how olhoi mooos shoulo Lo piicoo whon aulo congoslion is unpiicoo oi unooipiicoo.
Whoalon (1078) ano Wilson (1088) consiooioo how oplimal ioao capacily is alloioo whon
again aulo congoslion is unpiicoo oi unooipiicoo. Ainoll ano Yan (2000) hayo analyzoo
simullanoously socono-Losl liansil capacily. liansil piicing. ano ioao capacily whon aulo
congoslion is unooipiicoo. Chia. 1sui. ano Whalloy (2001) hayo inyosligaloo how much
of lho officioncy loss fiom nol applying congoslion lolls lo aulomoLilos can Lo iocoyoioo
lhiough a gasolino lax. Voihoof. Lmmoiink. Nijkamp. ano Iiolyolo (1006) hayo sluoioo
how lho yaluo of infoimalion lo cai oiiyois is mooiloo Ly nol congoslion piicing cai liayol.
Soyoial papois (o.g. Liaio (1006). Voihoof. Nijkamp. ano Iiolyolo (1006). ano Liu ano
McDonalo (1008)) hayo Loon wiillon on lho piopoilion of officioncy gains lhan can Lo
achioyoo whon only a suLsol of ioaos can Lo lolloo. Mosl hayo como lo lho possimislic
conclusion lhal only a small fiaclion of lho gains can Lo achioyoo. Small ano Yan (1000)
ano Voihoof ano Small (1000). howoyoi. aiguo lhal lhis conclusion is loo possimislic. ano
ooiiyos fiom ignoiing usoi holoiogonoily. Whon holoiogonoily ano honco solf-soloclion
acioss lolloo ano unlolloo links accoioing lo lho yaluo of limo is consiooioo. lho officioncy
gains fiom lolling only fiooways aio consiooiaLly magniloo.
Wilh lwo majoi oxclusions liip liming ano lano uso such Lioaoly is lho cui-
ionl slalo of lho oconomic lhooiy of uiLan liayol (Linosoy ano Voihoof (2000)). 1ho
ooyolopmonl of lho lhooiy has Loon aomiiaLlo in many iospocls. 1hiough olaLoialion of
a canonical moool. lho lhooiy has moyoo fiom a yoiy simplo moool lo moools lhal aio
incioasingly oosciipliyoly ioalislic ano incoipoialo moio ano moio maigins of choico. All
7
lho moool yaiianls mool lho slanoaio ciiloiia foi gooo miciooconomic mooolling. 1hoy
aio lhoioughly Lasoo on inoiyioual maximizing Lohayioi. aio concoplually consislonl. ano
aio paisimonious. Ano consiooiaLlo offoil has gono inlo piaclical applicalion. 1hoio aio
now laigo liloialuios on oslimaling liayol oomano funclions ano on ooyoloping officionl
algoiilhms lo solyo yaiianls of lho slalic nolwoik oouiliLiium pioLlom. incluoing lho com-
pulalion of socono-Losl oplimal lolls (o.g. Voihoof (2001) ano Hoain ( )). ano lhoio is a
giowing numLoi of cily-spocilc liayol simulalion moools Lasoo on lho aLoyo lhooiy (o.g.
Anooison ano Mohiing (1006 )).
1hoso aomiiaLlo oualilios nolwilhslanoing. I hayo six majoi ciilicisms of lho cuiionl
slalo of lho lhooiy.
1. Many ioloyanl maigins of choico aio ignoioo
Applioo miciooconomic has lhiiyoo laigoly Ly aLsliacling fiom lho inossonlial. In
many policy conloxls. mosl of lho aclion can Lo capluioo Ly consiooiing only a small
numLoi of maigins of choico. Foi oxamplo: i) wilh inousliial pollulion. mosl of lho
aclion is capluioo Ly moooling lims` choicos concoining lho loyol of oulpul ano lho
lochnology. as chaiacloiizoo Ly lho loyol of omissions of a fow pollulanls poi unil oulpul:
ii) wilh insuianco. mosl of lho aclion on lho consumoi sioo can Lo capluioo Ly yiowing
lho consumoi as choosing how much insuianco lo puichaso ano how much unoLsoiyaLlo
(oLsoiyaLlo maigins of choico can Lo wiillon inlo lho conliacl) offoil (which affocls lho
pioLaLilily of accioonl oi moio gonoially lho pioLaLilily oisliiLulion of accioonl oamago)
lo oxpono. iii) wilh housing. mosl of lho aclion can Lo capluioo Ly yiowing lho consumoi
as choosing localion. looi aioa. ano oualily. ano lho piooucoi as choosing sliucluial
oonsily ano oualily.
1ho mosl sophislicaloo moools of uiLan liayol (o.g. ooIalma`s ML1IOIOLIS) lioal
lho liayolloi as choosing liip fioouoncy. ioulo. mooo. ano oopailuio limo. Lul uiLan cai
oiiyois mako many moio oocisions lhan lhis. As I am oiiying along. I conlinually oocioo
8
how iapioly lo accoloialo oi oocoloialo. which ooloiminos my spooo ano lho oislanco
Lolwoon my cai ano lho cai in fionl. Ioiiooically. I hayo lo chooso wholhoi lo accopl
an oppoilunily lo oyoilako. wholhoi lo honk my hoin. wholhoi lo onloi an inloisoclion
afloi lho lighl has luinoo yollow (oi in Loslon ioo) oi whon il is Llockoo. ano wholhoi
lo shifl lo an appaionlly fasloi lano. As I am appioaching my ooslinalion. if I oo nol
hayo omployoi-pioyiooo paiking. I hayo lo oocioo wholhoi lo paik on-sliool oi off. ano if
on-sliool whal paiking slialogy lo aoopl. which incluoos how fai fiom my ooslinalion lo
slail ciuising foi paiking ano unooi whal ciicumslancos lo oouLlo paik.
Ono possiLlo ioaclion lo lhis onumoialion of choicos is lhal lhoy aio liiyial. Lach
choico Ly ilsolf may Lo liiyial Lul cumulaliyoly lhoy aio yoiy impoilanl. 1hink how
much Lolloi liaffic woulo low ano how loss sliossful uiLan oiiying woulo Lo if all oiiyois
woio lo mako socially officionl oocisions. Anolhoi possiLlo ioaclion is lhal oconomisls
hayo lilllo usoful lo say concoining lhoso oocisions. oyon lhough lhoy aio oconomic. ano
lhal iogulalion of oiiyoi Lohayioi shoulo Lo lofl lo liaffic onginoois. Lul liaffic onginoois
oocioo on liaffic iogulalions wilh no oxplicil oconomic Lohayioial analysis. ano oflon on
lho Lasis of insufficionl oala ano lawoo slalislical analysis.
If inoooo lhoio woio lisl-Losl congoslion piicing. wo woulo nol hayo lo woiiy lhal oui
analysis oyoilooks somo micioscopic maigins of choico. Diiyois woulo faco lho iighl piicos
on oyoiy maigin. ano woulo lhoiofoio mako socially officionl oocisions on oyoiy maigin.
Lul in piaclico congoslion piicing cannol Lo oiffoionlialoo accoioing lo oiiyoi Lohay-
ioi. Unooi anonymous congoslion piicing. aggiossiyo ano limio oiiyois imposo a laigoi
congoslion cosl oxloinalily lhan socially iosponsiLlo oiiyois Lul pay lho samo loll. Loss
oLyiously Lul as impoilanlly. link congoslion piicing. as compuloo. pioyioos no inconliyo
lo oiiyo in a socially iosponsiLlo way. A ialional oiiyoi who facos lho samo link loll in-
ooponoonlly of how ho oiiyos will oiiyo sollshingly. 1hus. oyon wilh homogonous oiiyois.
whal wo compulo as lisl-Losl" link congoslion piicing lakos as giyon inofficionl oiiyoi
0
individual
Lohayioi. which ionoois compulalion of oplimal lolling lhal is nol Lasoo on inoiyioual
oiiyoi Lohayioi an oxoiciso in lho lhooiy of lho socono Losl. Sinco. in lho woilo of lho
socono Losl. all maigins of choico shoulo Lo oxplicilly accounloo foi. in oyon oui so-calloo
lisl-Losl lhooiy ignoiing oiiyoi Lohayioi may iosull in soiiously misloaoing analysis.
In lho aLoyo oiscussion. I hayo focusoo on maigins of choico lhal lho con-
yonlional analysis ignoios. Iolicios associaloo wilh lhoso maigins of choico. mosl nolaLly
liaffic iogulalions. aio coiiosponoingly ignoioo. Also ignoioo aio cai manufacluiois` mai-
gins of choico. AulomoLilo chaiacloiislics affocl lho congoslion causoo Ly a cai. Sinco
lhoso chaiacloiislics aio al loasl pailially oLsoiyaLlo. congoslion piicing can Lo Lasoo on
lhom. Lul if il is nol consumois hayo no inconliyo lo puichaso congoslion-officionl" cais
oi cai makois lo manufacluio lhom.
2. 1ho congoslion funclion capluios nol only lochnology Lul also Lohayioi
1ho congoslion funclion is lioaloo as Loing a lochnological oalum. Lul in facl incoi-
poialos many maigins of choico. 1his poinl has Loon oomonslialoo foimally wilh iospocl
lo usois` liip-liming oocisions. Lul applios lo many olhoi maigins of choico. A majoi
ooyolopmonl in uiLan lianspoil oconomic lhooiy. which was pionooioo Ly Vickioy (1060).
has Loon lho oxlonsion of lho conyonlional slalic lhooiy lo lioal lho oynamics oi oyolu-
lion of congoslion oyoi lho iush houi. 1his has Loon oono Ly comLining a pailiculaily
simplo slalo yaiiaLlo chaiacloiizalion of nonslalionaiy liaffic low ououing Lohino a
Lolllonock wilh a Lohayioial oouiliLiium conoilion analogous lo lho Waioiop conoi-
lion. lhal oach oiiyoi choosos his oopailuio limo lo minimizo his liip piico. LouiliLiium
aggiogalo liayol cosls (which incluoo liayol limo cosls ano schooulo oolay cosls lho
cosls of Loing oaily oi lalo iolaliyo lo oosiioo aiiiyal limo) may lhon Lo compuloo as
a funclion of lho numLoi of oiiyois of yaiious lypos ano of lho lolling iogimo". 1his
aggiogalo liayol cosl funclion may lhon Lo lioaloo as lho lolal congoslion cosl funclion
in a slalic. iooucoo-foim social suiplus oi social wolfaio maximizalion pioLlom. 1hus.
10
per se
lho conyonlional slalic moool may Lo inloipioloo as incoipoialing liaffic oynamics ano
usois` liip-liming oocisions implicilly yia lho lolal congoslion cosl funclion. Accoioing lo
lhis inloipiolalion. lho lolal congoslion cosl funclion is nol complololy lochnological Lul
incoipoialos usois liip liming oocisions ano is inluoncoo Ly lho lolling iogimo in offocl.
1ho samo poinl applios wilh iospocl lo olhoi maigins of choico nol oxplicilly lioaloo
in lho conyonlional slalic moool. 1ho lolal congoslion funclion can Lo inloipioloo as
lioaling implicilly lhoso olhoi maigins of choico. Honco. lho conyonlional moool is moio
gonoially applicaLlo lhan ono mighl hayo suspocloo. Al lho samo limo. lioaling maigins of
choico implicilly caiiios wilh il lwin oangois: lisl. policy insliumonls lhal affocl only lho
implicil maigins of choico such as liaffic iogulalions may Lo oyoilookoo: ano socono.
whon a policy insliumonl affocls Lolh oxplicil ano implicil maigins. il is oasy lo foigol
lho impacl of lho policy insliumonl on lho implicil maigins foi oxamplo. iaising lho
loll may nol only lowoi oomano Lul also causo moloiisls lo oiiyo fasloi lhoioLy alloiing
lho foim of lho congoslion cosl funclion".
1ho congoslion cosl funclion is somolimos inloipioloo as incoipoialing lho cosls of
liaffic noiso. pollulion. ano accioonls. 1hal is concoplually accoplaLlo. Lul caiiios wilh il
lho oangoi lhal lho policy analysl will noglocl policy insliumonls affocling coiiosponoing
maigins of choico.
A iolaloo poinl is lhal lioaling lho yaluo of limo which onlois lho congoslion cosl
funclion as a oalum makos il oasy lo oyoilook policy insliumonls lhal affocl lho yaluo of
limo. 1iaffic iogulalions which oiscouiago aggiossiyo oiiying mako oiiying moio ploasanl.
1his iooucos lho yaluo of limo ano. holoing liaffic yolumo lxoo. iooucos congoslions cosls.
Lul also causos oomano lo iiso. incioasing lho loyol of congoslion.
8. Capacily is loo aggiogaloo a policy yaiiaLlo
1ianspoil plannois oo nol chooso capacily . Insloao. lhoy chooso ioao wiolh.
payomonl oualily. giaoionl. Lanking. iamp moloiing. spooo limils. ano so on. which lo-
11
golhoi ooloimino capacily. Sinco lianspoilalion plannois lono lo uso onginooiing iulos of
lhumL. wilhoul iofoionco lo oconomic yaiiaLlos. lhoii choico of how lo pioyioo a giyon
loyol of capacily may oiffoi signilcanlly fiom lho oosign which minimizo social cosls.
Lconomisls hayo a iolo in aoyising lianspoilalion plannois how lo pioyioo a giyon loyol
of capacily officionlly in oiffoionl oconomic onyiionmonls. 1o oo lhis. a sol of lianspoil
oconomic moools is nooooo lhal pioyioo a iichoi lioalmonl of liaffic onginooiing. 1wo
oxamplos of oxcollonl woik along lhoso linos aio NowLoiy (1088) ano Small. Winslon ano
Lyans (1080). NowLoiy oxaminos lho oconomics of payomonl iosuifacing. Small.Winslon
ano Lyans looks al lho oconomics of ioao oamago. consiooiing nol only how yohiclos shoulo
Lo chaigoo foi lho ioao oamago lhoy causo Lul also how payomonl ouiaLilily shoulo Lo
choson.
Lconomisls can conliiLulo lo lho liaffic onginooiing´lianspoilalion scionco liloialuio
as woll Ly ooyoloping micioscopic moools of liaffic low wilh Lohayioial founoalions. Somo
woik has alioaoy Loon oono along lhoso linos. IolomLoig (108ò) ano Voihoof. Iouwonoal.
ano Iiolyolo (1000) pioyioo moools in which oiiyois oocioo on spooo. spacing. oi accoloi-
alion so as lo maximizo ulilily. liaoing off liayol limo againsl lho pioLaLilily of accioonl.
Ano Mohiing ( ) ooiiyos ulilily-maximizing spooo as a liaoooff Lolwoon liayol limo
ano gasolino consumplion. Fuilhoi woik along lhoso linos shoulo incoipoialo ioconl ooyol-
opmonls in micioscopic liaffic low lhooiy (1ianspoilalion Iosoaich Loaio). incluoing cai
following lhooiy which ooiiyos aggiogalo liaffic low fiom a oiffoionco-oiffoionlial ooualion
oosciiLing lho accoloialion of inoiyioual yohiclos.
Whilo Vickioy is Losl known among lianspoil oconomisls as a ciusaooi foi congoslion
piicing ano as lho ooyolopoi of lho Lolllonock moool. ho also pionooioo in lho onginooiing
oconomics of congoslion. Ho iogulaily allonooo lho 1ianspoilalion Iosoaich Loaio annual
moolings. as woll as inloinalional confoioncos in liaffic onginooiing: ho ooyolopoo soyoial
moools of congoslion olhoi lhan lho link low ano Lolllonock moools: ano ho also con-
12
1
1
virtual links
In applicalions of liaffic nolwoik oouiliLiium lhooiy. which incoipoialo only link low congoslion. inloi-
soclions aio lioaloo as . Foi oxamplo. a noilhLouno cai which makos a iighl-hano luin is yiowoo
as liayoling along a link joining lho N-S ioao lo lho L-W ioao. 1his lioalmonl is Lolloi lhan nolhing. Lul
nol onliioly salisfacloiy sinco il ignoios lho inloiaclion Lolwoon liaffic liayolling in oiffoionl oiioclions.
siooioo officionl suLway schoouling. soaling. Liaking. faio colloclion. ano plalfoim longlh.
ano ooyisoo schomos lo miligalo Lus Lunching.
4. Link low congoslion is nol lho only foim of congoslion
1ho conyonlional moool lioals only ono foim of congoslion. link low congoslion whoioLy
a oiiyoi`s cosls on a link aio posiliyoly iolaloo lo liaffic yolumo oi low on lho link. As has
alioaoy Loon noloo. lho conyonlional slalic link congoslion cosl funclion can also Lo inloi-
pioloo as pioyioing a iooucoo foim iopiosonlalion of Lolllonock congoslion. Lul lhoio aio
many liaffic congoslion phonomona lhal aio nol consislonl wilh link low congoslion. Link
low congoslion oxcluoos liansionl. non-sloaoy-slalo low phonomona such as shock wayos
ano hypoicongoslion. Il also ignoios congoslion al nooos. In lolophono liaffic. congoslion
al nooos (in swilching ciicuils) is moio impoilanl lhan congoslion on links (Syski (1086)).
Lxamplos of nooal congoslion in lho conloxl of uiLan liayol aio inloisoclion . fiooway
onlianco ano oxil. ano paiking congoslion. Link congoslion oominalos nooal congoslion in
fiooway liayol. Lul nol on cily sliools. Olhoi foims of congoslion incluoo pooosliian-cai
inloiaclion. onliy inlo ano oxil fiom paiking. moiging. ano phonomona ooiiying fiom lho
physical longlh of cais such as giiolock.
Of lhoso. poihaps lho mosl impoilanl is paiking congoslion. IomaikaLly lilllo is
known ompiiically aLoul paiking. Lul lhoio is somo oyioonco lhal in cai liayol wilh a
oownlown ooslinalion. lho ayoiago limo ooyoloo lo lnoing a paiking spaco may Lo as laigo
as lho ayoiago limo losl lo congoslion on ioulo. Cais ciuising foi paiking also conliiLulo
signilcanlly lo liaffic congoslion. Ono oncounlois lho assoilion in lho liloialuio lhal in
congosloo oownlown aioas. half lho cais oiiying aio ciuising foi paiking. 1his sooms loo
high. Lul ciuising foi paiking no oouLl conliiLulos signilcanlly lo congoslion. On-sliool
paiking also iooucos capacily. ano oouLlo paiking ano onliy inlo ano oxil fiom on-sliool
18
interaction externalities
economics of
agglomeration
subsidizing
paiking can soiiously impooo liaffic low.
1o ooiiyo officionl uiLan lianspoil sysloms wo shall nooo lo ooyolop iichoi ano moio
micioscopic moools of congoslion.
ò. Inloiaclion Lolwoon uiLan liayol oisloilions ano olhoi oisloilions in lho oconomy
may Lo impoilanl
1ho conyonlional moooling of uiLan liayol ignoios lho inloiaclions Lolwoon uiLan
liayol oisloilions ano olhoi oisloilions in lho oconomy. 1wo of lhoso aio pioLaLly pai-
liculaily impoilanl. 1ho lisl. lho analog of which has Loon oiscussoo al longlh in lho
onyiionmonlal oconomics liloialuio (LoyonLoig ano Goulooi (1006. 1008). ano Iaiiy ano
Lonlo (2000)). concoins lho inloiaclion Lolwoon lho laLoi-loisuio oisloilion causoo Ly lho
laxalion of wago incomo ano uiLan liayol. 1ho socono. which lo my knowloogo has nol
Loon monlionoo Lofoio. concoins lho connoclion Lolwoon ano
uiLan liayol. In ioconl yoais. lhoio has Loon consiooiaLlo iosoaich on lho
(soo Fujila ano 1hisso (2000 ) foi an oxcollonl suiyoy). Mosl oxpoils con-
siooi lhal lho non-maikol oxchango of infoimalion lhiough faco-lo-faco inloiaclion is a
piimaiy. ano poihaps lho piimaiy. foico oncouiaging lims lo clusloi. If lhis is coiiocl.
lho associaloo posiliyo inloiaclion oxloinalily musl Lo of lho samo oiooi of magniluoo as
lho oxloinalily associaloo wilh unpiicoo uiLan aulo congoslion. 1o miligalo lho ooao-
woighl loss associaloo lho inloiaclion oxloinalily. inloiaclion shoulo Lo oncouiagoo. ano
poihaps uiLan liayol is a way lo ooal wilh lhis. Wo know lilllo ompiiically
aLoul lho inloiaclion Lolwoon lho laLoi-loisuio oisloilion ano lhal aiising fiom unpiicoo
aulo congoslion. ano oyon loss ompiiically aLoul lho inloiaclion Lolwoon lho inloiaclion
oxloinalily ano lho uiLan aulo congoslion oxloinalily.
Whal implicalions lhoso inloiaclions Lolwoon oisloilions hayo foi uiLan liayol policy
is nol al all cloai. Ono possiLilily. howoyoi. is lhal congoslion piicing will Locomo loss
alliacliyo ano micioscopic policios aimoo al impioying lho officioncy wilh which a giyon
14
per se.
3. A Selection of Research Topics
palloin of liaffic is accommooaloo moio alliacliyo.
6. 1ho oomano foi liayol is piooominanlly a ooiiyoo oomano
1his is a familiai ciilicism. 1ho conyonlional moool lioals inoiyiouals as ooiiying
ulilily fiom liayol Lul inoiyiouals ooiiyo ulilily fiom acliyilios aiiayoo oyoi limo
ano spaco. as woll as fiom gooos ano soiyicos. 1hoso acliyilios ioouiio lianspoilalion. as
woll as olhoi gooos ano soiyicos as inpuls. 1hus. lho oomano foi liayol is piooominanlly
a ooiiyoo oomano. Whilo almosl oyoiyono acknowloogos lho coiioclnoss of lhis ciilicism.
lilllo piogioss has Loon maoo in ooyoloping acliyily-Lasoo moools of ooiiyoo liayol oomano.
1ho associaloo schoouling pioLloms aio yoiy oifficull. ano no piogioss has Loon maoo in lho
solulion of schoouling pioLloms which ioouiio schooulo cooioinalion Lolwoon inoiyiouals.
1hus. il sooms lhal wo aio sluck wilh lioaling liayol as a lnal gooo. 1his is unfoilunalo
sinco il muooios applicalion of pioouclion officioncy aigumonls fiom oplimal lax lhooiy
in lho conloxl of uiLan liayol.
In conclusion: UiLan lianspoil oconomic lhooiy has ooyolopoo piimaiily lhiough lho
olaLoialion ano iolnomonl of a singlo. canonical moool. In many iospocls. lho piocoss has
Loon aomiiaLlo ano fiuilful. Sinco. howoyoi. wo aio so familiai wilh lho moool. wo aio apl
lo foigol lhal il is only ono of many possiLlo simpliloo iopiosonlalions of yoiy complox
ano yaiioo liaffic congoslion phonomona. Ly oiawing so hoayily on lhis singlo moool as
lho concoplual Lasis foi oui policy analysis. wo hayo pioLaLly placoo oxcossiyo omphasis
on congoslion piicing whilo ignoiing olhoi polonlially yaluaLlo policy lools. A piomising
ayonuo lowaios iooiossing lhis imLalanco is lo ooyolop moio micioscopic oconomic moools
of uiLan liayol.
In lho iomainooi of lho lalk. I shall oiscuss soyoial iosoaich lopics which will I hopo soiyo
lo illuslialo lho poinls I hayo maoo.
1. Iogulalion of fioighl ooliyoiios limo of oay ano liuck sizo

2
2

A
w kT

( ) =
( )
( )
=
=
( )
0;
A T
k
t . e
e A, T, w
∂t /∂T
∂t /∂A
Ak
w kT
,
∂e
∂T
Ak
w kT
>
In oownlown Loslon al loasl. fioighl ooliyoiy conliiLulos consiooiaLly lo liaffic con-
goslion. Laigo inloislalo (oosignoo foi fiooway liayol) liucks hayo liouLlo manoouyoiing
iouno coinois oi naiiow sliools: lhoy complololy Llock liaffic on sliools whon onloiing ano
oxiling fiom loaoing oocks: ano lhoii oouLlo-paiking foi ooliyoiios whoio loaoing oocks
aio aLsonl soyoioly iooucos capacily.
I know of no moooin oconomic sluoy of uiLan fioighl ooliyoiy in lho conloxl of uiLan
liayol congoslion. Wo soom simply lo lioal liucks as so many cai-oouiyalonls. ano assumo
lhal whaloyoi policy is Losl foi cais is Losl foi liucks. Ly omploying such a ciuoo
lioalmonl of uiLan fioighl lianspoil. wo oyoilook many policios lhal mighl signilcanlly
ioouco lho unpiicoo congoslion oxloinalily imposoo Ly liucks.
Ono such policy is imposing iosliiclions on lho limos of oay al which oownlown fioighl
ooliyoiios can Lo maoo. Such a policy is. I unooislano. on lho Looks in Iaiis. lhough I hayo
no iooa how sliiclly il is onfoicoo. Lyalualing such a policy ioouiios oslimaling lho cosls.
which incluoo inconyonionco cosls of iosliicloo ooliyoiy houis lo shippois ano iocoiyois. as
woll as ooloimining moio piocisoly lho lochnology" of lho congoslion inloiaclion Lolwoon
cais ano liucks. As has Loon noloo. il is slanoaio lo lioal liucks as cai-oouiyalonls. which
assumos lhal cais ano liucks onloi lho congoslion funclion aooiliyoly. Maiyin Kiaus.
howoyoi. aiguos lhal liucks onloi lho congoslion funclion as ioouclions in capacily. which
accoios Lolloi wilh my inluilion. Lol oonolo lho numLoi of cais. lho numLoi of
liucks. ano lho amounl Ly which a liuck iooucos capacily. 1hon lho link liayol-limo
funclion may Lo wiillon as Lol oonolo lho cai oouiyalonls of a liuck in loims
of congoslion
(òa)
so lhal
(òL)
lhal is. lho cai oouiyalonls of a liuck is incioasing in lho numLoi of liucks. 1hus. in
16
lho aLsonco of congoslion piicing. coiiococlnoss of lho Kiaus conjocluio woulo slionglhon
lho aigumonl foi smoolhing liuck liaffic oyoi lho couiso of lho oay. ano pioLaLly loo foi
iosliicling ooliyoiy houis.
Anolhoi fioighl-ooliyoiy-iolaloo policy is iogulaling liuck sizo in aioas of soyoio congos-
lion. Lyalualing lhis policy ioouiios knowloogo of waiohousing lochnology ano piaclico.
Supposo. al ono oxliomo. lhal cuiionl piaclico is foi all gooos lianspoiloo Ly inloi-cily
liuck lo Lo unloaooo al suLuiLan waiohousos ano ioloaooo onlo smalloi liucks foi oo-
liyoiy wilhin lho moliopolilan aioa. In lhis caso. iolaliyoly lilllo aooilional cosl woulo
Lo imposoo Ly iosliicling liuck sizo foi uiLan ooliyoiios. Supposo. al lho olhoi oxliomo.
lhal cuiionl piaclico is foi all gooos lo Lo lianspoiloo fiom supplioi lo iocoiyoi oooi-
lo-oooi. Iogulaling liuck sizo foi uiLan ooliyoiy woulo lhon Lo consiooiaLly cosllioi. If
oooi-lo-oooi ooliyoiios woio lo conlinuo. lho smalloi liuck sizo woulo suLslanlially in-
cioaso inloi-cily shipping cosls. Olhoiwiso. now waiohousing oisliicls woulo hayo lo Lo
consliucloo whoio gooos shippoo inloi-cily in laigo liucks woulo Lo unloaooo ano loaooo
onlo smalloi liucks foi uiLan ooliyoiy.
Goho (2000) is cuiionlly wiiling hoi lhosis on uiLan fioighl consolioalion. ano Siyi-
lanioou Lofoio hoi unlimoly ooalh was sluoying lho spalial oconomics of uiLan waiohous-
ing. Moio iosoaich along lhoso linos is Laoly nooooo. Lul will Lo hampoioo Ly oolcioncios
in oala.
Moio gonoially. sluoy of lho conliiLulion lo congoslion causoo Ly uiLan fioighl ooliyoiy
shoulo Lo high on lho uiLan lianspoil oconomic iosoaich agonoa.
2. 1ho onginooiing oconomics of uiLan aulo congoslion
I hayo alioaoy monlionoo lho oxomplaiy woik Ly NowLoiy (1088) ano Small.Winslon.
ano Lyans (1080) iolaloo lo lho oconomics of fiooway´highway oosign. consliuclion. ano
mainlonanco. 1hoii woik coiiosponos lo souno cosl-Lonoll piaclico. If souno cosl-Lonoll
analysis woio piaclicoo in all aspocls of uiLan ioao onginooiing. yoiy consiooiaLlo cosl say-
17
Transportation Sci-
ence Transportation Research).
ings coulo Lo achioyoo. ospocially if socono-Losl consiooialions woio piopoily accounloo foi
(Kanomolo (1000)). 1hough lho woik mighl Lo ialhoi unglamoious ano concoplually pio-
saic. wo can mako yaluaLlo conliiLulions Ly oomonslialing in piaclical applicalions how
cosl-Lonoll analysis shoulo Lo oono ano Ly piossing haio foi lho aooplion of souno cosl-
Lonoll piocoouios Ly slalo ano local goyoinmonls. Wo can also conliiLulo Ly applying
oconomics lo lho nuls ano Lolls of ioao oosign. Foi oxamplo. oyon lho mosl sophislicaloo
cosl-Lonoll piocoouios cuiionlly omployoo pioyioo yoiy ciuoo lioalmonls of uncoilainly.
laking no accounl of lho liloialuio on iiioyoisiLlo inyoslmonl ano ioal oplions. Iulos foi
ioao iosuifacing shoulo lako inlo accounl lhal fuluio liaffic yolumos aio gonoialoo Ly a
slochaslic piocoss. lhal oui unooislanoing of payomonl oamago will impioyo. ano lhal
lochnological aoyancos in payomonl oosign will occui. ano shoulo accoioingly Lo moio
aoapliyo ano loxiLlo.
Lzia Hauoi ( ) has aiguoo lhal lho Lulk of lianspoil onginooiing slanoaios ano
iulos aio Lasoo on scanl ompiiical losling ano on-silo oala colloclion. ano oflon faully
slalislical analysis. 1o lhis I woulo aoo lhal onginooiing slanoaios lono lo Lo applioo
wilhoul iofoionco lo oconomic yaiiaLlos lho oiscounl ialo. lho yaluo of limo. olc. Lco-
nomic sluoios which ooyiso guioolinos foi such slanoaios woulo Lo yaluaLlo.
1o oo lianspoil onginooiing oconomics woll ioouiios somo oxpoiliso in lianspoil on-
ginooiing. Al piosonl lhoio is lilllo piofossional communicalion Lolwoon liaffic ongi-
noois´lianspoilalion scionlisls ano lianspoil oconomisls. 1ho silualion is impioying.
howoyoi. Lconomics is now laughl in many giaoualo lianspoilalion onginooiing pio-
giams. ano lho lianspoilalion scionco liloialuio is oiawing incioasingly on oconomics.
pailiculaily wilh iogaios lo piicing solulions. Ano al loasl a suLsol of uiLan lianspoil
oconomisls koop aLioasl of lho ioloyanl onginooiing liloialuio (o.g.
ano Lul wo can oo Lolloi. Wo shoulo nol only oncouiago
lianspoilalion onginooiing piogiams lo loach moio ano Lolloi oconomics. poihaps yolun-
18
looiing lo loach a couiso ouisolyos. pailicipaling in lhosis supoiyision of lianspoilalion
onginooiing sluoonls. ano allonoing lianspoilalion onginooiing sominais. Lul wo shoulo
also incluoo moio lianspoilalion onginooiing in oui loaching ano in oui iosoaich.
8. AulomoLilo noiso. liaffic accioonls. ano uiLan aulo pollulion
1ho aulomoLilo congoslion cosl funclion is oflon inloipioloo as incoipoialing lho cosls
associaloo wilh lho noiso. accioonl. ano pollulion causoo Ly cais. 1ioaling lho congoslion
cosl funclion as lochnologically ooloiminoo lhoiofoio ignoios all lho Lohayioial ano ongi-
nooiing oocisions lhal inluonco how liaffic congoslion affocls lho loyols of noiso. accioonls.
ano pollulion.
1hoio has Loon consiooiaLlo iosoaich inlo aulomoLilo pollulion ano alloinaliyo policios
foi iooucing aulomoLilo omissions (o.g.. Small ano Kazimi (100ò). Liownslono ( )).
ano policy in lhis aioa has Loon ialhoi succossful. Wo nooo a compaiaLlo Looy of woik
on lho noiso pollulion gonoialoo Ly cais ano on lho oconomics of liaffic accioonls.
A Goiman lianspoil oconomisl ioconlly infoimoo mo lhal Goimans aio moio con-
coinoo Ly lho noiso gonoialoo Ly liaffic lhan Ly lho limo oolays ouo lo liaffic congoslion.
Whilo I oouLl lhis. il is nonolholoss liuo lhal liaffic noiso in Luiopoan cilios lonos lo
Lo signilcanlly highoi lhan in U.S. cilios (whoio populalion is loss oonso. sliools wiooi.
ano Luiloing solLacks highoi) ano can Lo yoiy iiiilaling. UiLan lianspoil oconomisls
hayo paio lilllo allonlion lo liaffic noiso. Il is. howoyoi. an aioa of policy whoio offocliyo
policy iomooios woulo Lo populai ano iolaliyoly oasy lo implomonl. Whal nooos sluoy is
lho cosl-Lonoll calculus of alloinaliyo policios. Il woulo nol Lo oifficull lo oosign ouioloi
cais. Lul al piosonl cai makois hayo lilllo inconliyo lo oo so Locauso oach oiiyoi incuis
only a fiaclion of lho noiso cosl gonoialoo Ly hoi cai: lho samo poinl applios lo liucks
ano Lusos. which conliiLulo oispiopoilionaloly lo liaffic noiso. Hoin-honking. a cuiso of
liying in Loslon. coulo Lo ooall wilh Ly making il a liaffic yiolalion oxcopl whon oono lo
ayoio an accioonl. as I galhoi is oono in many Luiopoan cilios. Ioao woik ano gaiLago
10
2
2
Only half in josl. I woulo liko lo soo Lao oiiyois osliacizoo fiom lho ioao. 1ho oifficully lios in oosigning
a lochnology which allows a oiiyoi lo iopoil anolhoi wilhoul inloifoiing wilh his oiiying conconlialion.
colloclion coulo Lo maoo ouioloi.
1iaffic accioonls aio coslly nol only foi lho oiiocl oamago lhoy causo Lul also foi
lho non-iocuiionl congoslion (incluoing cuiiosily congoslion") lhoy inouco. Lconomisls
hayo paio somo allonlion lo liaffic accioonls (o.g.. Vickioy (1068)). ospocially lo lho offocl
of insuianco on lho inconliyo lo oiiyo safoly (o.g.. Loyoi ano Dionno (1087)) Lul hayo
lofl olhoi aspocls such as lho iogulalion of unsafo oiiying. lho oosign of ioaos foi safoly.
ano accioonl follow-up piocoouios. foi liaffic onginoois who. in lhis conloxl loo. lono lo
chooso policy wilh lilllo oi no oxplicil allonlion lo oconomics. Viilually no allonlion has
Loon paio Ly oilhoi oconomisls oi onginoois lo lho link Lolwoon liaffic accioonls ano lho
lochnology" of congoslion. 1ho piimaiy liaoooff ooloimining lho many small oocisions
oiiyois mako is Lolwoon iooucing liayol limo ano incioasing accioonl iisk. 1hus. policios
which affocl lho piiyalo cosls of accioonls may hayo a signilcanl offocl on how liaffic
lows.
4. Unciyil oiiying Lohayioi ano lho yaluo of limo
1ho slanoaio moool of aulo congoslion lioals an inoiyioual`s yaluo of limo as oxogonous
(o.g. Calfioo ano Winslon (1008)). Lul il isn`l. Il ooponos on lho schoouling consliainls
sho confionls ano on how ploasanl oi unploasanl sho lnos oiiying. A simplo way lo ioouco
lho cosl of liaffic congoslion ano lho congoslion oxloinalily cosl is lo mako oiiying moio
ploasanl. lhoioLy iooucing lho yaluo of limo (lhough ooing so also lowois liip piico which
slimulalos oomano).
1ho maikol lakos caio of aulomoLilo comfoil. Lul nol lho oouiliLiium slioss loyol
associaloo wilh oiiying. I conjocluio lhal oiiying slioss is sliongly iolaloo lo lho incioonco
of unciyil ano oangoious oiiying: lailgaling. honking al lho slighlosl pioyocalion. iunning
yollow ano ioo lighls. making oangoious ano oxcossiyo lano changos. Il is uncloai how
offocliyo puLlic policy can Lo in oiscouiaging anli-social oiiying . Lul lho lopic is woilhy
20
8
8
Appaionlly pooosliian liaffic loo is chaiacloiizoo Ly hypoicongoslion.
an onouiiy Ly oconomisls.
ò. Lconomics of mass liansil ano pooosliian liaffic
1ho Lulk of lho woik oono on lho oconomics of uiLan liaffic congoslion has concoinoo
cais. Iolaliyoly lilllo has Loon oono on lho oconomics of mass liansil. piosumaLly Locauso
unlil ioconlly mosl of lho innoyaliyo iosoaich in lho lolo was oono in lho Uniloo Slalos
whoio mass liansil is iolaliyoly unimpoilanl. Mohiing (1072) oxploioo somo of lho Lasic
oconomic piinciplos of mass liansil. in pailiculai oconomios of soiyico fioouoncy ano
soiyico oonsily: ano lhoio is a suLslanlial liloialuio which oslimalos mass liansil cosl
funclions (o.g. Loiochman (1008)) ano a smalloi ono which oxaminos capilal-inlonsily Lias
(o.g. Fiankona (1087)). Lul uiLan lianspoil oconomisls hayo ooyoloo lilllo allonlion lo
lho micioscopics of uiLan mass liansil. 1ho majoi oxcoplion is Vickioy. As noloo oailioi.
ho oio consiooiaLlo woik in lho aioa Lul mosl iomains unpuLlishoo ano much is pailiculai
lo Now Yoik Cily. Wo woulo oo woll lo follow his loao. Samplo lopics incluoo socono-Losl
liain´Lus sizo ano soiyico fioouoncy ano oonsily. piocoouios lo miligalo Lus Lunching
(Vickioy (1070)). lo oxpooilo passongoi onliy ano oxil. ano lo ioouco lho congoslion
imposoo Ly Lusos.
1ho oconomics of pooosliian congoslion is unoxploioo. 1ho pioposal lo imposo a min-
imum walking spooo on ono sioo of Oxfoio Sliool was mol wilh consiooiaLlo amusomonl.
Lul moiils soiious analysis . Iooosliian-cai congoslion inloiaclion is polonlially impoi-
lanl. Shoulo jaywalking Lo oiscouiagoo´ How wioo shoulo sioowalks Lo´
6. Hypoicongoslion
Hypoicongoslion is lho phonomonon whoioLy a giyon low occuis al a signilcanlly lowoi
spooo lhan is possiLlo. Hypoicongoslion has puzzloo ano inliiguoo lianspoil onginoois
ano oconomisls. Iooucing ils incioonco woulo pioyioo suLslanlial officioncy gains. How
can lhis Lo achioyoo´ 1ho cuiionl wisoom (Hall (). Small ano Chu (). Voihoof ()) is lhal
21
hypoicongoslion is a liansionl phonomonon gonoialoo Ly oyonls (such as a slow cai in lho
passing lano oi a oouLlo-paikoo cai) which liiggoi Lackwaio shockwayos. Accoioingly.
such oyonls shoulo Lo ponalizoo. How spocilcally lhis shoulo Lo achioyoo is an oxoiciso
which Llonos liaffic low lhooiy ano oconomics al lho micioscopic loyol.
7. Floxlimo ano slaggoioo woiking houis
A gonoialion ago uiLan lianspoil oconomisls (wilh lho oxcoplion of Vickioy) osson-
lially ignoioo lho oynamics of iush houi liaffic congoslion. Locauso of oxcossiyo focus on
lho conyonlional moools. lhoy oyoilookoo lho liip liming oocision. 1hal oyoisighl has now
laigoly Loon iocliloo in lho now suLslanlial liloialuio on Lolllonock congoslion. inilialoo
Ly Vickioy (1060). 1hal liloialuio has slionglhonoo lho caso foi congoslion piicing Ly
oomonslialing lhal limo-yaiying congoslion piicing can suLslanlially ioouco congoslion.
holoing lxoo lho limo palloin of aiiiyals (Ainoll. ooIalma. ano Linosoy (1008)).
1ho Lolllonock liloialuio lakos lho oisliiLulion of woik slail limos as oxogonous. 1ho
onoogonoily of lho oisliiLulion of woik slail limos has Loon sluoioo Ly Honooison (1081).
In his moool. oach omployoi oocioos whon lo hayo his omployoos slail woik. liaoing off lho
Lonoll fiom haying his omployoos inloiacl wilh moio omployoos fiom olhoi lims againsl
lho highoi wago lhal ho musl pay his woikois foi commuling in congosloo conoilions.
1hoio is lhon an inloiaclion Lolwoon inloiaclion ano liaffic congoslion oxloinalilios. Is
il woilhwhilo foi lho goyoinmonl lo allompl lo mooify piiyalo lims` woik slail limos´
Wholhoi il is oi nol. can lho goyoinmonl. as lho oominanl omployoi in many juiisoiclions.
signilcanlly impioyo officioncy Ly mooifying lho woik slail limos of ils omployoos (Lonsall
(1078))´
8. Non-commuling liips
Il has Loon aiguoo lhal lalo lwonliolh-conluiy uiLan oconomics was piooccupioo wilh
iolning a moool of lho ninoloonlh-conluiy cily. UiLan lianspoil oconomics can Lo suL-
jocloo lo a similai ciilicism. lhal il is iolning moools of commuling liaffic al a limo whon
22
an incioasing piopoilion of iush-houi liayol has a non-commuling puiposo. Fifloon yoais
ago. lho lguio was Lanoioo aLoul lhal loss lhan lfly poiconl of iush-houi liips aio foi
commuling. ano ioconlly lho lguio foi Chicago has fallon lo lhiily poiconl. Ono musl
lako such lguios wilh a giain of sall. sinco lhoy aio sonsiliyo lo lho puiposo assignoo lo
chainoo liips. Lul. howoyoi moasuioo. a sloaoily incioasing piopoilion of uiLan liayol
has a non-commuling puiposo.
1ho oifficully wilh lioaling non-commuling liips wilhin a oynamic moool of congoslion
is lhal oosiioo aiiiyal limo ano coiiosponoingly schooulo oolay cosls Locomo fuzzy. Ialhoi.
an inoiyioual schooulos his acliyilios laking liaffic congoslion inlo accounl. Incioasoo
congoslion will causo him lo cancol somo acliyilios ano ioschooulo olhois. Unfoilunaloly.
il has pioyoo yoiy oifficull lo opoialionalizo lhis concoplualizalion. Unlil wo oo. lho
oomano sioo of oynamic uiLan liayol moools will iomain oisconcoilingly woak.
0. Iaiking
Il was iomaikoo oailioi lhal lho focus of oui lhooiy on link low liaffic congoslion
has oisliacloo us fiom olhoi foims of congoslion ano associaloo policy lools. Iailiculaily
impoilanl is paiking-iolaloo congoslion. Al loasl foi aulo liayol wilh a oownlown oosli-
nalion. lho ayoiago limo losl in soaiching foi a paiking spol may Lo as laigo as lho ayoiago
limo losl ouo lo congosloo liaffic. ano ciuising foi on-sliool paiking pioLaLly conliiLulos
signilcanlly lo oownlown liaffic congoslion.
1o oalo. mosl of lho woik on paiking Ly uiLan lianspoil oconomisls (Vickioy (10ò4).
Iolh (106ò). Gillon (1077.1078). Shoup (1082. 1087).Shoup ano Willson (1002). Glazoi
ano Niskanon (1002). Voihoof. Nijkamp. ano Iiolyolo (100ò). Callhiop. Iioosl. ano yan
Donooi(2000)) iogaio paiking as a piico. oilhoi lxoo oi poi unil limo. payaLlo al lho
ooslinalion foi a liip Ly cai. In moooling mooal choico. ano lho offocls of cashing oul
omployoi-pioyiooo paiking. lhis is a conyonionl simplilcalion. Howoyoi. il also missos a
lol. Il ignoios lho conliiLulion of paiking lo liaffic congoslion ciuising foi paiking.
28
4
4
Axhauson (1000). Axhauson ano Iolak (100ò). ano Ainoll ano Iowso (1000) pioyioos lisl slops in lhis
oiioclion.
capacily ioouclion fiom on-sliool paiking. ano congoslion ouo lo onliy inlo ano oxil fiom
on-sliool paiking. oouLlo paiking. ano ououos al onliios lo off-sliool paiking. Il also
ignoios lho congoslion cosl paikois imposo on ono anolhoi. in loims of lho limo lo lno a
paiking spaco ano lo walk fiom a paiking spol lo lho ooslinalion. as woll as lho allonoanl
uncoilainly. 1hoso phonomona can Lo ciuooly capluioo Ly iooucoo-foim moools: foi
oxamplo. ono mighl moool a oiiyoi`s paiking limo cosls as incioasing in lho ialio of
lho numLoi of aulo commulois alioaoy paikoo lo capacily. Lul lhis appioach ignoios lho
micioscopics of paiking-iolaloo liaffic congoslion. lho slochaslicily inhoionl in lho paiking
soaich piocoss. lho spalial aspocls of soaiching foi paiking. ano lho allocalion of lano lo
paiking. A moio salisfacloiy appioach woulo Lo Lolh spalial ano sliucluial. mooolling
oiiyois` spalial soaich foi paiking. 1his woulo poimil lho analysis of paiking policios
al an appiopiialoly micio loyol: lho allocalion of lano lo on- ano off-sliool paiking. lho
spacing Lolwoon paiking gaiagos. piiyalo ys puLlic ownoiship of paiking gaiagos. socono-
Losl on- ano off-sliool paiking foo sliucluios. lho allocalion of on-sliool paiking Lolwoon
moloioo. iosioonl. ano uniosliicloo paiking. limo limils foi on-sliool paiking. ano so on.
1iaffic onginoois hayo ooyoloo suipiisingly lilllo allonlion lo paiking-iolaloo congos-
lion. 1ho ooailh of slylizoo facls will impooo moooling. ano may ioouiio lhal wo unooilako
lho oala colloclion ano analysis ouisolyos.
24
4. Conclusion
canonical macroscopic link ow model
link congestion cost function
microscopic
Wo lho communily of uiLan lianspoil oconomisls hayo Loon loss offocliyo lhan wo
coulo hayo Loon in aoyising policy makois. Ly focusing oxcossiyoly on congoslion piicing.
I am coilainly nol opposoo lo congoslion piicing. Lul lhink lhal wo hayo consiooiaLly
unooioslimaloo lho cosls of implomonlalion. ano may hayo oyoioslimaloo lho Lonolls as
woll. Lo lhal as il may. lhoio sooms lo Lo almosl oyoiwholming opposilion lo congoslion
piicing. Ils limo may yol como. Lul foi lho momonl wo shoulo Lo piagmalic ano ooyolo
moio of oui offoils lo consiooiing moio immooialo. piaclical. ano munoano policy issuos.
In my lalk looay. I hayo aiguoo lhal oui piooccupalion wilh congoslion piicing has
slommoo fiom oxcossiyo iolianco on a singlo moool fiamowoik which I hayo loimoo
lho of uiLan liaffic congoslion. LlaLoialion of lhis
moool has giyon oui lolo an impiossiyo. consislonl. ano cohoionl Looy of lho lhooiy. Lul
looking al uiLan lianspoil congoslion lhiough a singlo lons has oisloiloo oui poicoplion.
causing us lo noglocl many ioal-woilo aspocls of uiLan liaffic congoslion ano consoouonlly
lo oyoilook many piomising ayonuos of policy-iolaloo iosoaich. Mosl of my ciilicisms
concoinoo lho . Ly lioaling congoslion al such an aggiogalo
loyol. il has causoo us lo look al policy al loo aggiogalo a loyol. loaying micioscopic
policy lo onginoois whoso oocisions aio lypically ill-infoimoo Ly oconomics. Ly lioaling
congoslion as puioly lochnological. il has causoo us lo oyoilook many inoiyioual maigins
of choico lhan can Lo inluoncoo Ly policy. Ano Ly lioaling only link low congoslion.
il has causoo us lo oyoilook olhoi foims of liaffic congoslion ano lho associaloo policy
yaiiaLlos.
I also pul foiwaio a iosoaich agonoa. 1ho lopics choson woio nol inlonooo lo Lo
oxhausliyo. Lul ialhoi lo illuslialo lho lypo of iosoaich lhal woulo aooioss my ciilicisms
of oui oxisling Looy of lhooiy. Ly ano laigo. lho iosoaich agonoa is in naluio.
ano complomonls ialhoi lhan compolos wilh lho macioscopic lhooiy which oominalos oui

lolo. 1ho lopic hoaoings giyo a layoi of lho lypo of iosoaich I hayo in mino:
iogulalion of fioighl ooliyoiy limo of oay ano liuck sizo
lho onginooiing oconomics of uiLan aulo congoslion
aulomoLilo noiso. liaffic accioonls. ano uiLan aulo pollulion
unciyil oiiying Lohayioi ano lho yaluo of limo
oconomics of mass liansil ano pooosliian liaffic
hypoicongoslion
loxlimo ano slaggoioo woik houis
non-commuling liips
paiking
1ho iosoaich agonoa I pul foiwaio was fiuslialingly oyon annoyingly non-spocilc.
Lul I am noimally nol somoono who ciilicizos wilhoul offoiing consliucliyo suggoslion.
ano I hopo lhan in lho yoais ahoao lhal my own iosoaich will go somo way lo mooling
lho challongos I hayo posoo.
26
References
[1[ Ahlsliano. I. (1008). 1ho iiso ano fall of lho hoioic lianspoil plan foi Slockholm".
1ianspoil Iolicy ò:20ò-211
[2[ Anooison. D. ano H.Mohiing (1006). Congoslion Cosls ano Congoslion Iiicing foi lho
1win Cilios. Iopoil No. MN´IC-06´82 lo Minnosola Dopailmonl of 1ianspoilalion
(Dopailmonl of Lconomics. Uniyoisily of Minnosola. Minnoapolis´Sl. Iaul)
[8[ Ainoll. I.(1004).Alloyialing liaffic congoslion: alloinaliyos lo ioao piicing". woik-
ing papoi No.282. Dopailmonl of Lconomics. Loslon Collogo
[4[ Ainoll. I.. A. oo Ialma ano I.Linosoy (1008). A sliucluial moool of poak-poiioo
congoslion: a liaffic Lolllonock wilh olaslic oomano". Amoiican Lconomic Ioyiow
88: 161-70
[ò[ Ainoll I.. ano J. Iowso (1000). Moooling Iaiking". Jouinal of UiLan Lconomics
4ò: 07-124.
[6[ Ainoll. I. ano A. Yan (2000). 1ho lwo-mooo pioLlom: socono-Losl piicing ano
capacily". Ioyiow of UiLan ano Iogional Doyolopmonl Sluoios 12:170-100
[7[ Axhauson. K.W. (1000). A simullanoous simulalion of acliyily chains". in I.M.Jonos
(Lo). Now Appioachos in Dynamic ano Acliyily-Lasoo Appioachos lo 1iayol Analy-
sis. 206-22ò. AyoLuiy. Alooishol.
[8[ Axhauson. K.W. ano J.W.Iolak (100ò).A oisaggioaglo moool of lho offocls of paik-
ing guioanco sysloms". 7lh WC1I Iiocoooings
[0[ Lockmann. M.. C.L. McGuiio ano C.L. Winslon (10ò6). Sluoios in lho Lconomics of
1ianspoilalion. Yalo Uniyoisily Iioss. Now Hayon.
27
[10[ Loiochman. J. (1008). IuLlic 1iansil Lconomics ano Doiogulalion Iolicy (Noilh-
Hollano. Amsloioam)
[11[ Loiliano. 1.(1077).Socono-Losl congoslion laxos in lianspoilalion sysloms". Lcono-
moliica 4ò: 1708-1ò
[12[ Lonsall. 1.(1078).FloxiLlo woik houis ano puLlic liansil in Ollawa". piosonloo
al lho annual confoionco of lho Ioaos ano 1ianspoilalion Associalion of Canaoa.
1oionlo. Onlaiio. SoplomLoi 28-4
[18[ Loiins. S.F.(1086).1ho polilical oconomy of ioao piicing: lho caso of Hong Kong".
Iiocoooings of lho Woilo Confoionco on 1ianspoil Iosoaich. Vancouyoi. L.C.. Vol.2.
1867-78
[14[ LoyonLoig. A.L. ano Goulooi. L.H.(1006).Oplimal onyiionmonlal laxalion in lho
piosonco of olhoi laxos: gonoial oouiliLiium analysis". Amoiican Lconomic Ioyiow
86: 08ò-1000
[1ò[ LoyonLoig. A.L. ano Goulooi. L.H.(1008).Lnyiionmonlal 1axalion". in A.AuoiLach
ano M.Folosloin. oos. HanoLook of IuLlic Lconomic. 2no ooilion. Noilh-Hollano.
Now Yoik
[16[ Loyoi. M. ano G.Dionno (1087). 1ho oconomics of ioao safoly". 1ianspoilalion
Iosoaich 21L:418-481
[17[ Liaio. I.(1006).Ioak loao piicing of a lianspoilalion ioulo wilh an unpiicoo suL-
slilulo". Jouinal of UiLan Lconomics 40: 170-07
[18[ Liownslono
[10[ Calfioo. J.L. ano C.Winslon (1008). 1ho yaluo of aulomoLilo liayol limo: implica-
lions foi congoslion policy". Jouinal of IuLlic Lconomics 60: 88-102.
28
[20[ Callhiop. L.. S. Iioosl. ano K. yan Donooi (2000). Iaiking policios ano ioao piic-
ing". UiLan Sluoios 87: 68-76.
[21[ Chia. N.-C.. A.1sui. ano J.Whalloy (2001). Ownoiship ano uso laxos as congoslion
coiiocling insliumonls". NLLI woiking papoi 8278. CamLiiogo. MA
[22[ Fiankona. M. (1087).Capilal-Liasoo suLsioios. Luioaucialic moniloiing. ano Lus
sciapping". Jouinal of UiLan Lconomics 21: 180-08
[28[ Fujila. M. ano J.-F.1hisso (2000). 1ho foimalion of oconomic agglomoialions:olo
pioLloms ano now poispocliyos". in J.-M.Huiiol ano J.-F.1hisso. oos. Lconomics of
Cilios:1hooiolical Ioispocliyos. CamLiiogo Uniyoisily Iioss.CamLiiogo UK
[24[ Goho (2000). Ih.D. lhosis pioposal. Uniyoisily of Califoinia. Iiyino.
[2ò[ Gillon. D.W.(1077). Lslimalion ano spocilcalion of lho offocl of paiking cosls on
uiLan lianspoil mooo choico". Jouinal of UiLan Lconomics 4: 186-00
[26[ Gillon. D.W.(1078). Iaiking localion oocisions ano lho oisliiLulion of congoslion".
1ianspoilalion 7: 60-8ò
[27[ Glazoi.A. ano Niskanon. L.(1002). Iaiking foos ano congoslion". Iogional Scionco
ano UiLan Lconomics 22: 128-82
[28[ Gomoz-ILanoz. J.(1002).1ho polilical oconomy of highway lolls ano congoslion piic-
ing". 1ianspoilalion Quailoily 46:848-860
[20[ Gomoz-ILanoz. J. ano K.Small (1004). Ioao Iiicing foi Congoslion Managomonl: A
Suiyoy of Inloinalional Iiaclico. Nalional Coopoialiyo Highway Iosoaich Iiogiam.
Synlhosis of Highway Iiaclico 210 (Nalional Acaoomy Iioss. Washinglon. DC).
[80[ Hall. F.
20
[81[ Hauoi. L.
[82[ Honooison. J.V.(1081).1ho oconomics of slaggoioo woik houis". Jouinal of UiLan
Lconomics 0: 840-64
[88[ Inslilulo of 1ianspoilalion Lnginoois (1082). 1ianspoilalion ano 1iaffic Lnginooiing
HanoLook. Iionlico-Hall. Lonoon
[84[ Jonos. I.(1008). UiLan ioao piicing: puLlic accoplaLilily ano Laiiiois lo implomon-
lalion". in K.J. Lullon ano L.1. Voihoof. oos. Ioao Iiicing. 1iaffic Congoslion ano
lho Lnyiionmonl: Issuos of Lfficioncy ano Social FoasiLilily . 268-84. Lowaio Llgai:
Chollonham. UK
[8ò[ Kanomolo. Y.(1000).Lonoll -cosl oyalualion of lianspoilalion inyoslmonl wilh un-
ooipiicoo congoslion". Uniyoisily of 1okyo. mimoo.
[86[ Kiaus. M. ano Y.Yoshioa (2000).1ho consumoi`s limo-of-uso oocision ano oplimal
piicing ano soiyico in uiLan mass liansil". Loslon Collogo. mimoo. foilhcoming in
Jouinal of UiLan Lconomics
[87[ Laffionl. J.-J. ano J.1iiolo.(1008). A 1hooiy of Inconliyos in Iiocuiomonl ano Iog-
ulalion. MI1 Iioss. CamLiiogo. MA
[88[ Loyy-LamLoil. H.(1068).1aiilcalion oos soiyicos a oualilo yaiiaLlo applicalion aux
poagos oo ciiculalion". Lconomoliica 86: ò64-74
[80[ Linosoy. I. ano L.1. Voihoof (2000). Congoslion mooolling ". foilhcoming. in Hano-
Look of 1ianspoil Mooolling. Vo l.1. D.A. Honshoi ano K.J. Lullon. oos. Llsoyioi
Scioncos. Oxfoio
[40[ Linosoy. I. ano L. Voihoof (2001). 1iaffic congoslion ano congoslion piicing". mimoo
80
[41[ Liu. L. ano J.F.McDonalo.(1008).Lfficionl congoslion lolls in lho piosonco of un-
piicoo congoslion: a poak ano off-poak simulalion moool". Jouinal of UiLan Lco-
nomics 44:8ò2-66
[42[ Maichano. M. (1068).A nolo on oplimal lolls in an impoifocl onyiionmonl". Lcono-
moliica 86: ò7ò-81
[48[ Mayoios. I. ano S.Iioosl (1007).Oplimal lax ano puLlic inyoslmonl iulos foi con-
goslion lypo of oxloinalilios". Scanoinayian Jouinal of Lconomics 00: 261-270
[44[ Mohiing. H. (1072).Oplimizalion ano scalo oconomios in uiLan Lus lianspoilalion".
Amoiican Lconomic Ioyiow 62: ò01-604
[4ò[ Mohiing. H.
[46[ Mohiing. H. (1076). 1ianspoilalion Lconomics. Lallingoi. CamLiiogo MA
[47[ NowLoiy.D. (1088). Ioao oamago oxloinalilios ano ioao usoi chaigos". Lconomol-
iica ò6: 20ò-816.
[48[ Iaiiy. I. ano A.Lonlo (2000). 1ax ooouclions. onyiionmonlal policy. ano lhooouLlo
oiyioono` hypolhosis". foilhcoming. Scanoinayian Jouinal of Lconomics
[40[ IolomLoig. J. (108ò). 1ho officioncy of oouiliLiium liaffic lows". Jouinal of IuLlic
Lconomics 26:101-20ò
[ò0[ Iolh. G.J.(106ò).Iaying Foi Iaiking. 1ho Inslilulo of Lconomic Affaiis. Lonoon
[ò1[ Shoiman. I. (1071). Congoslion inloiooponoonco ano uiLan liansil faios ". Lcono-
moliica 80: ò6ò-76
[ò2[ Shoup. D. (1082). Cashing oul fioo paiking". 1ianspoilalion Quailoily 8ò: 8ò1-864
81
[ò8[ Shoup. D. (1007).Lyalualing lho offocls of cashing oul omployoi-paio paiking: oighl
caso sluoios". 1ianspoil Iolicy 4: 201-16
[ò4[ Shoup. D. ano Willson.I.(1002).Lmployoi-paio paiking : lho pioLlom ano pioposoo
solulions". 1ianspoilalion Quailoily 46: 160-02
[òò[ Small. K.. ano J.Gomoz-ILanoz (1008). Ioao piicing foi congoslion managomonl:
lho liansilion fiom lhooiy lo policy. " in K.Lullon ano L.Voihoof. oos.. Ioao Iiicing.
1iaffic Congoslion. ano lho Lnyiionmonl: Issuos of Lfficioncy ano Social FoasiLilily.
268-84. Lowaio Llgai: Chollonham. UK
[ò6[ Small. K.A. ano C.Kazimi (100ò). On lho cosls of aii pollulion fiom moloi yohiclos".
Jouinal of 1ianspoil Lconomics ano Iolicy 20: 7-82.
[ò7[ Small. K.. ano J.Yan (1000) 1ho yaluo of yaluo piicing` of ioaos: socono-Losl piicing
ano pioouclion oiffoionlialion". Jouinal of UiLan Lconomics. foilhcoming
[ò8[ Small. K.. C.Winslon. ano C.Lyans (1080) Ioao Woik:A Now Highway Iiicing ano
Inyoslmonl Iolicy. Liookings Inslilulion. Washinglon. D.C.
[ò0[ Sliolz. I. (106ò). UiLan lianspoilalion paiaLlos". in J.Maigolis. oo.. 1ho IuLlic
Lconomy of UiLan Communilios (Iosouicos foi lho Fuluio:Washinglon. DC)
[60[ Syski. I.(1086).Congoslion 1hooiy. Noilh-Hollano. Amsloioam.
[61[ 1ianspoilalion Iosoaich Loaio (). 1iaffic Flow 1hooiy
[62[ Voihoof. L.
[68[ Voihoof. L. Nijkamp. I. ano Iiolyolo. I.(100ò).1ho oconomics of iogulaloiy paiking
policios: lho (im)possiLililios of paiking policios in liaffic iogulalion". 1ianspoilalion
Iosoaich. 20A: 141-ò6
82
[64[ Voihoof. L. Nijkamp. I. ano Iiolyolo. I.(1006).Socono-Losl congoslion piicing: lho
caso of an unpiicoo alloinaliyo". Jouinal of UiLan Lconomics 40: 270-802
[6ò[ Voihoof. L. J.Iouwonoal ano Iiolyolo. I.(1000).Congoslion causoo Ly spooo oiffoi-
oncos". Jouinal of UiLan Lconomics 4ò: ò88-ò6
[66[ Voihoof. L ano K.A. Small (1000).Iiooucl oiffoionlialion on ioaos: socono-Losl con-
goslion piicing wilh holoiogonoily unooi puLlic ano piiyalo ownoiship". Discussion
Iapoi 1I00-066´8. 1inLoigon Inslilulo. Amsloioam-Iolloioam
[67[ Vickioy. W.S.(10ò4). 1ho oconomizing of cuiL paiking spaco". 1iaffic Lnginooiing
Magazino. Noy. Iopiinloo in Jouinal of UiLan Lconomics 86: ò6-6ò(1004)
[68[ Vickioy. W.S.(10ò0). Slalomonl on lho piicing of uiLan sliool uso". Hoaiings.
U.S.Congioss Joinl Commilloo on Moliopolilan Washinglon IioLloms. NoyomLoi
11.[Iopiinloo in Jouinal of UiLan Lconomics 86: 42-6ò (1004).[
[60[ Vickioy. W.S.(1068). AulomoLilo accioonls. loil law. oxloinalilios. ano insuianco:
an oconomisl`s ciiliouo". Law ano Conlompoilaiy IioLloms 88:464-487
[70[ Vickioy. W.S.(1060) Congoslion lhooiy ano lianspoilalion inyoslmonl". Amoiican
Lconomic Ioyiow ò0: 2ò1-61
[71[ Vickioy. W.S.(1070). Spacing oul lhoso giogaiious Lusos". Now Yoik 1imos. Juno
[72[ Wallois. A. (1061). 1ho lhooiy ano moasuiomonl of piiyalo ano social cosl of high-
way congoslion". Lconomoliica 10:670-600
[78[ Whoalon. W.C.(1078) Iiico-inoucoo oisloilions in uiLan highway inyoslmonl". Loll
Jouinal of Lconomics 0: 622-82
[74[ Wilson. J.D.(1088) Oplimal ioao capacily in lho piosonco of unpiicoo congoslion".
Jouinal of UiLan Lconomics 18: 887-ò7
88
Hoain
ooIalma. ML1IOIOLIS
84
Q - Traffic volume
$
trip
D
MSC
ACH= MC =
t
*
Figure 1. Diagramaticrepresentation ofthe basic model of trafic congestion
Fig1micrscopic.nb 1

The Economic Theory of Urban Traffic Congestion: A Microscopic Research Agenda

1. Introduction
The theme of my talk today is that we the community of urban transport economists

rely excessively on the canonical macroscopic model of urban traffic congestion, which has caused us to place excessive emphasis on congestion pricing as a policy tool to improve the efficiency of urban travel. Urban traffic congestion is the outcome of decisions made on many more margins of choice than even the most sophisticated variants of the canonical models capture. To increase the usefulness of our theory in practical policy application, we need to develop a portfolio of models that capture the omitted margins of choice. Applied microeconomic theory has been so successful largely because of its method, which entails working with simple, conceptually consistent models based on maximizing behavior. This method elucidates basic principles and focuses on essentials, abstracting from distracting detail. Recent policy successes of this approach include the development of a market for SO2 , the expanded use of auctions for resource allocation with small numbers of buyers and sellers, and the application of incentive contracting to public utilities (Laffont and Tirola (1993). Application of this method in the context of urban traffic congestion has led urban transport economists to advocate congestion pricing ideally perfect congestion pric-

ing but realistically partial or imperfect congestion pricing. But our pleas for congestion pricing have been singularly unsuccessful. With the arguable exception of Singapore, congestion pricing of urban travel has not been implemented anywhere (Small and GomezIbanez (1998)), and many jurisdictions (e.g. Cambridge, Hong Kong (Borins (1986)), Stockholm (Ahlstrand (1981)),and the Randstaad) have backed down from plans to introduce congestion pricing on even an experimental basis. Why congestion pricing of urban auto travel has been received by policy makers with so little enthusiasm has been much 2

discussed in the literature (Gomez-Ibanez (1992), Jones (1998)). My own view is that we urban transport economists have been deceived by the simplicity of our models into greatly underestimating the costs and practical difficulties of implementing congestion pricing on urban roads, and that policy makers have wisely resisted doing so. Be that as it may, the social returns to our labors will, I believe, be greater if we devote more of our efforts towards examining alternatives to congestion pricing. And this, I shall argue, will require the development of more microscopic models of urban travel, which provide a more detailed description of urban traffic ow and urban travel decisions than is done in the canonical macroscopic model; what I have in mind will hopefully be clari ed in the talk. In section 2, I shall provide a brief sketch of the development of urban transport economic theory, as well as its current state, and then provide a critique of it. Then, in section 3, I shall discuss a selection of research topics, which will together illustrate how I think urban transport economic theory should be re-oriented to make it more useful for policy makers.

3

2. and the only economic decision each driver makes is trip frequency. and Winsten (1956)) examines travel on a point-input. the social bene t function B (Q). w). The optimum occurs where demand intersects marginal social cost. w) K (w ). The same model may be described algebraically using either social surplus or social welfare analysis (Mayeres and Proost (1997)). The direct. The basic model (Beckmann. (1) 4 . the equilibrium occurs where demand intersects average cost. 1 HERE Figure 1 gives a diagrammatic representation of the basic model (Walters (1961). Current Urban Transport Economic Theory and Its Application The development of urban transport economic theory has entailed the gradual elaboration of a canonical model. McGuire. Where p is trip price and w capacity. user cost. and MSC is the marginal social cost of a trip. with capacity xed. evaluated at the social optimum. average cost. (short-run) average cost c(Q. point-output road. . Q. the demand function is D(p). AC relates each driver s trip cost to traffic volume. Individual drivers are identical. The vertical distance between MSC and AC is the congestion externality cost. D is the demand curve. and marginal private cost. and the capacity construction cost function K (w). In the absence of government intervention. long-run social surplus maximization problem may then be written as max Q. Mohring (1976 )). The minimal government intervention needed to decentralize the social optimum is the imposition of a congestion toll equal to the congestion externality cost. Congestion is captured by a congestion cost function which relates trip cost to traffic volume and capacity. INSERT FIG.w B (Q ) Qc(Q. and is variously referred to as trip cost.

where individuals decide on trip frequency based on trip price and the government decides on the congestion toll. Eq. B (D(p)) D(p)c(D (p). i) Q = D(p) ii) p = which reduces to max p. w).which gives the optimality conditions: Q: B (Q ) c(Q.p.(2b) states that optimal road width is such that the marginal social bene t from road expansion. . with trip price equalling average cost plus the toll. which equals short-run average cost plus the congestion externality cost. w) K (w ).t.w + c(Q. the reduction in travel costs holding traffic volume xed. w). (4) The corresponding rst-order conditions are the same as those of the direct maximization problem. w) ∂Q =0 (2a) w: Q K = 0. .w s. w) + Q ∂c ∂w ∂c(Q. is max (B (Q ) Qp) + (Q K (w)) (3) Q. equals the marginal construction cost. i) Q = D(p) ii) p = c(Q. (2b) Eq.p. w) iii) 5 =0 .n s. or as the solution to the following constrained maximization problem max B (Q ) Qp + Q K (w ) Q.t. The no-toll equilibrium may be characterized as the solution to Q = D(p) and p = c(Q. The indirect social surplus maximization problem.(2a) states that optimal traffic volume is such that the marginal social bene t of a trip equals marginal social cost.

When modes are not perfect substitutes in demand. the maximization problem is extended to multiple modes. the same principle applies. All that is required is to measure the link congestion externality costs. where M is the number of modes. except that there is congestion interaction between buses and cars. Individuals from different groups have different bene t and average cost functions. very little information is needed to decentralize the rst-best optimum. Furthermore. With respect to modal choice. and w M -dimensional vectors. which requires knowledge of only the link congestion functions and traffic levels. The model was also extended early on to treat user heterogeneity (Strotz (1965 )). traffic xed. the marginal social bene t of capacity on each link can be computed straightforwardly as the travel cost savings on that link. With respect to route choice. On the assumption 6 .The basic model has been enriched to account for other margins of choice. in terms of the model. And while not analytically necessary. No information is needed on the identity of travellers or on their demand functions. and user heterogeneity. without consideration of how travellers switch modes and routes in response to the incremental capacity expansion. so as to minimize trip price the generalized Wardrop principle. It is therefore easy to understand why economists have pushed so hard for perfect congestion pricing. Early on the model was extended to treat route choice and modal choice. in the model extended to treat route choice. but no congestion interaction between cars on different links. it is almost always assumed that individuals from different groups enter the congestion cost functions symmetrically. when modes are perfect substitutes in demand. with D. an individual chooses his route on a network. the full optimum can be decentralized by applying an anonymous toll on each link in the network equal to that link s congestion externality cost. These are very important results since they indicate that. modal choice. from a given origin to a given destination. The above models have been extended to treat freight traffic. Under this assumption. p.

With two major exclusions trip timing and land use such broadly is the cur- rent state of the economic theory of urban travel (Lindsey and Verhoef (2000)). The development of the theory has been admirable in many respects. Wheaton (1978) and Wilson (1983) considered how optimal road capacity is altered when again auto congestion is unpriced or underpriced. Tsui. Nijkamp. however. Marchard (1968).that a truck contributes to congestion in the same way as a xed number of cars. and Liu and McDonald (1998)) have been written on the proportion of efficiency gains than can be achieved when only a subset of roads can be tolled. Verhoef. Nijkamp.g. argue that this conclusion is too pessimistic. All 7 . The above models have also been applied to treat a range of second-best problems. Braid (1996). transit pricing. When heterogeneity and hence self-selection across tolled and untolled links according to the value of time is considered. Lévy-Lambert (1968). Several papers (e. Verhoef. Sherman (1971) and Bertrand (1977) examined how other modes should be priced when auto congestion is unpriced or underpriced. rstbest congestion tolling with trucks paying the car-equivalents toll continues to decentralize the rst-best optimum. the theory has moved from a very simple model to models that are increasingly descriptively realistic and incorporate more and more margins of choice. Through elaboration of a canonical model. and road capacity when auto congestion is underpriced. the efficiency gains from tolling only freeways are considerably magni ed. Arnott and Yan (2000) have analyzed simultaneously second-best transit capacity. and Rietveld (1996) have studied how the value of information to car drivers is modi ed by not congestion pricing car travel. Chia. and Rietveld (1996). Most have come to the pessimistic conclusion that only a small fraction of the gains can be achieved. Emmerink. and Whalley (2001) have investigated how much of the efficiency loss from not applying congestion tolls to automobiles can be recovered through a gasoline tax. and derives from ignoring user heterogeneity. Small and Yan (1999) and Verhoef and Small (1999).

dePalma s METROPOLIS) treat the traveller as choosing trip frequency.g. ii) with insurance. I continually decide oor area. are conceptually consistent. As I am driving along. They are thoroughly based on individual maximizing behavior. density and quality. The most sophisticated models of urban travel (e. iii) with housing.the model variants meet the standard criteria for good microeconomic modelling. most of the action is captured by modeling rms choices concerning the level of output and the technology. most of the action can be captured by considering only a small number of margins of choice. Anderson and Mohring (1996 )).g. mode. But urban car drivers make many more decisions than this. as characterized by the level of emissions of a few pollutants per unit output. There are now large literatures on estimating travel demand functions and on developing efficient algorithms to solve variants of the static network equilibrium problem. route. I have six major criticisms of the current state of the theory. and there is a growing number of city-speci c travel simulation models based on the above theory (e. and are parsimonious. Many relevant margins of choice are ignored Applied microeconomic has thrived largely by abstracting from the inessential. and quality. Verhoef (2001) and Hearn ( )). and departure time. and the producer as choosing structural 8 . most of the action can be captured by viewing the consumer as choosing location. These admirable qualities notwithstanding. 1. including the computation of second-best optimal tolls (e. For example: i) with industrial pollution.g. most of the action on the consumer side can be captured by viewing the consumer as choosing how much insurance to purchase and how much unobservable (observable margins of choice can be written into the contract) effort (which affects the probability of accident or more generally the probability distribution of accident damage) to expend. And considerable effort has gone into practical application. In many policy contexts.

As I am approaching my destination. aggressive and timid drivers impose a larger congestion cost externality than socially responsible drivers but pay the same toll. Thus. link congestion pricing. Periodically. But traffic engineers decide on traffic regulations with no explicit economic behavioral analysis. and if on-street what parking strategy to adopt. Drivers would face the right prices on every margin. and would therefore make socially efficient decisions on every margin. I have to choose whether to accept an opportunity to overtake. Each choice by itself may be trivial but cumulatively they are very important. Under anonymous congestion pricing. Another possible reaction is that economists have little useful to say concerning these decisions. even though they are economic. One possible reaction to this enumeration of choices is that they are trivial. which determines my speed and the distance between my car and the car in front. If indeed there were rst-best congestion pricing. which includes how far from my destination to start cruising for parking and under what circumstances to double park. provides no incentive to drive in a socially responsible way. whether to enter an intersection after the light has turned yellow (or in Boston red) or when it is blocked. as computed. and often on the basis of insufficient data and awed statistical analysis. I have to decide whether to park on-street or off. if I do not have employer-provided parking. even with homogenous drivers. A rational driver who faces the same link toll independently of how he drives will drive sel shingly. Think how much better traffic would ow and how less stressful urban driving would be if all drivers were to make socially efficient decisions. what we compute as rst-best link congestion pricing takes as given inefficient driver 9 .how rapidly to accelerate or decelerate. and whether to shift to an apparently faster lane. we would not have to worry that our analysis overlooks some microscopic margins of choice. whether to honk my horn. and that regulation of driver behavior should be left to traffic engineers. But in practice congestion pricing cannot be differentiated according to driver behavior. Less obviously but as importantly.

This point has been demonstrated formally with respect to users trip-timing decisions. reduced-form social surplus or social welfare maximization problem. Since these characteristics are at least partially observable. Equilibrium aggregate travel costs (which include travel time costs and schedule delay costs the costs of being early or late relative to desired arrival time) may then be computed as a function of the number of drivers of various types and of the tolling régime . I have focused on individual margins of choice that the conventional analysis ignores. The congestion function captures not only technology but also behavior The congestion function is treated as being a technological datum.behavior. Since. most notably traffic regulations. in the world of the second best. This has been done by combining a particularly simple state variable characterization of nonstationary traffic ow bottleneck queuing behind a with a behavioral equilibrium condition analogous to the Wardrop condi- tion. Also ignored are car manufacturers margins of choice. 10 . but in fact incorporates many margins of choice. but applies to many other margins of choice. all margins of choice should be explicitly accounted for. that each driver chooses his departure time to minimize his trip price. This aggregate travel cost function may then be treated as the total congestion cost function in a static. 2. A major development in urban transport economic theory. congestion pricing can be based on them. In the above discussion. Thus. are correspondingly ignored. in even our so-called rst-best theory ignoring driver behavior may result in seriously misleading analysis. but if it is not consumers have no incentive to purchase congestion-efficient cars or car makers to manufacture them. Policies associated with these margins of choice. which was pioneered by Vickrey (1969). Automobile characteristics affect the congestion caused by a car. which renders computation of optimal tolling that is not based on individual driver behavior an exercise in the theory of the second best. has been the extension of the conventional static theory to treat the dynamics or evolution of congestion over the rush hour.

which to- 11 . policy instruments that affect only the implicit margins of choice such as traffic regulations may be overlooked. when a policy instrument affects both explicit and implicit margins. and accidents. The same point applies with respect to other margins of choice not explicitly treated in the conventional static model. 3. speed limits. but carries with it the danger that the policy analyst will neglect policy instruments affecting corresponding margins of choice. Capacity is too aggregated a policy variable Transport planners do not choose capacity per se. Hence. reduces congestions costs. the conventional model is more generally applicable than one might have suspected. increasing the level of congestion. pavement quality. Instead. it is easy to forget the impact of the policy instrument on the implicit margins for example.the conventional static model may be interpreted as incorporating traffic dynamics and users trip-timing decisions implicitly via the total congestion cost function. Traffic regulations which discourage aggressive driving make driving more pleasant. gradient. holding traffic volume xed. and so on. pollution. The total congestion function can be interpreted as treating implicitly these other margins of choice. but also causes demand to rise. According to this interpretation. banking. This reduces the value of time and. That is conceptually acceptable. The congestion cost function is sometimes interpreted as incorporating the costs of traffic noise. they choose road width. ramp metering. the total congestion cost function is not completely technological but incorporates users trip timing decisions and is in uenced by the tolling régime in effect. treating margins of choice implicitly carries with it twin dangers: rst. A related point is that treating the value of time function which enters the congestion cost as a datum makes it easy to overlook policy instruments that affect the value of time. At the same time. raising the toll may not only lower demand but also cause motorists to drive faster thereby altering the form of the congestion cost function . and second.

Further work along these lines should incorporate recent developments in microscopic traffic ow theory (Transportation Research Board). and he also con- 12 . he developed several models of congestion other than the link ow and bottleneck models. And Mohring ( ) derives utility-maximizing speed as a tradeoff between travel time and gasoline consumption. Rouwendal. without reference to economic variables. Small. Economists have a role in advising transportation planners how to provide a given level of capacity efficiently in different economic environments. their choice of how to provide a given level of capacity may differ signi cantly from the design which minimize social costs.Winston and Evans looks at the economics of road damage. He regularly attended the Transportation Research Board annual meetings. Since transportation planners tend to use engineering rules of thumb.gether determine capacity. Newbery examines the economics of pavement resurfacing. While Vickrey is best known among transport economists as a crusader for congestion pricing and as the developer of the bottleneck model. including car following theory which derives aggregate traffic ow from a difference-differential equation describing the acceleration of individual vehicles. a set of transport economic models is needed that provide a richer treatment of traffic engineering. Winston and Evans (1989). spacing. or acceleration so as to maximize utility. considering not only how vehicles should be charged for the road damage they cause but also how pavement durability should be chosen. Rotemberg (1985) and Verhoef. and Rietveld (1999) provide models in which drivers decide on speed. he also pioneered in the engineering economics of congestion. Economists can contribute to the traffic engineering/transportation science literature as well by developing microscopic models of traffic ow with behavioral foundations. as well as international conferences in traffic engineering. To do this. Two examples of excellent work along these lines are Newbery (1988) and Small. Some work has already been done along these lines. trading off travel time against the probability of accident.

Link ow congestion excludes transient. but not on city streets. One encounters the assertion in the literature that in congested downtown areas. Remarkably little is known empirically about parking. This seems too high. and phenomena deriving from the physical length of cars such as gridlock. the conventional static link congestion cost function can also be interpreted as providing a reduced form representation of bottleneck congestion. link ow congestion whereby a driver s costs on a link are positively related to traffic volume or ow on the link. It also ignores congestion at nodes. Other forms of congestion include pedestrian-car interaction. non-steady-state ow phenomena such as shock waves and hypercongestion. Link ow congestion is not the only form of congestion The conventional model treats only one form of congestion. congestion at nodes (in switching circuits) is more important than congestion on links (Syski (1986)). As has already been noted. Of these. But there are many traffic congestion phenomena that are not consistent with link ow congestion.sidered efficient subway scheduling. Link congestion dominates nodal congestion in freeway travel. a northbound car which makes a right-hand turn is viewed as traveling along a link joining the N-S road to the E-W road. entry into and exit from parking. In telephone traffic. On-street parking also reduces capacity. intersections are treated as virtual links. which incorporate only link ow congestion. seating. perhaps the most important is parking congestion. merging. fare collection. and platform length. Cars cruising for parking also contribute signi cantly to traffic congestion. and parking congestion. but there is some evidence that in car travel with a downtown destination. 4. the average time devoted to nding a parking space may be as large as the average time lost to congestion en route. and double parking and entry into and exit from on-street In applications of traffic network equilibrium theory. braking. half the cars driving are cruising for parking. and devised schemes to mitigate bus bunching. Examples of nodal congestion in the context of urban travel are intersection1 . but cruising for parking no doubt contributes signi cantly to congestion. freeway entrance and exit. 1 13 . This treatment is better than nothing. but not entirely satisfactory since it ignores the interaction between traffic travelling in different directions. For example.

the associated positive interaction externality must be of the same order of magnitude as the externality associated with unpriced urban auto congestion. In recent years. We know little empirically about the interaction between the labor-leisure distortion and that arising from unpriced auto congestion. One possibility. and Parry and Bento (2000)). however. The second. If this is correct. force encouraging rms to cluster. What implications these interactions between distortions have for urban travel policy is not at all clear.parking can seriously impede traffic ow. To derive efficient urban transport systems we shall need to develop richer and more microscopic models of congestion. Interaction between urban travel distortions and other distortions in the economy may be important The conventional modeling of urban travel ignores the interactions between urban travel distortions and other distortions in the economy. 1998). and even less empirically about the interaction between the interaction externality and the urban auto congestion externality. Two of these are probably particularly important. concerns the connection between interaction externalities and urban travel. interaction should be encouraged. Most experts consider that the non-market exchange of information through face-to-face interaction is a primary. The rst. To mitigate the deadweight loss associated the interaction externality. is that congestion pricing will become less attractive and microscopic policies aimed at improving the efficiency with which a given 14 . 5. the analog of which has been discussed at length in the environmental economics literature (Bovenberg and Goulder (1996. concerns the interaction between the labor-leisure distortion caused by the taxation of wage income and urban travel. and perhaps the primary. there has been considerable research on the economics of agglomeration (see Fujita and Thisse (2000 ) for an excellent survey). which to my knowledge has not been mentioned before. and perhaps subsidizing urban travel is a way to deal with this.

it seems that we are stuck with treating travel as a nal good. The associated scheduling problems are very difficult. The demand for travel is predominantly a derived demand This is a familiar criticism. The conventional model treats individuals as deriving utility from travel per se. While almost everyone acknowledges the correctness of this criticism. By drawing so heavily on this single model as the conceptual basis for our policy analysis. Thus. Regulation of freight deliveries time of day and truck size 15 . little progress has been made in developing activity-based models of derived travel demand. the demand for travel is predominantly a derived demand. the process has been admirable and fruitful. 1. 3. A promising avenue towards redressing this imbalance is to develop more microscopic economic models of urban travel. however. we have probably placed excessive emphasis on congestion pricing while ignoring other potentially valuable policy tools.pattern of traffic is accommodated more attractive. we are apt to forget that it is only one of many possible simpli ed representations of very complex and varied traffic congestion phenomena. we are so familiar with the model. I shall discuss several research topics which will I hope serve to illustrate the points I have made. These activities require transportation. In conclusion: Urban transport economic theory has developed primarily through the elaboration and re nement of a single. This is unfortunate since it muddies application of production efficiency arguments from optimal tax theory in the context of urban travel. Since. In many respects. as well as other goods and services as inputs. A Selection of Research Topics In the remainder of the talk. But individuals derive utility from activities arrayed over time and space. and no progress has been made in the solution of scheduling problems which require schedule coordination between individuals. as well as from goods and services. canonical model. Thus. 6.

as well as determining more precisely the technology of the congestion interaction between cars and trucks. Let A denote the number of cars. I know of no modern economic study of urban freight delivery in the context of urban travel congestion. which assumes that cars and trucks enter the congestion function additively. they completely block traffic on streets when entering and exiting from loading docks. Let e denote the car equivalents of a truck in terms that is. Thus. One such policy is imposing restrictions on the times of day at which downtown freight deliveries can be made. We seem simply to treat trucks as so many car-equivalents. w) = so that ∂e Ak 2 = > 0. argues that trucks enter the congestion function as reductions in capacity. T the number of trucks. As has been noted. Such a policy is. By employing such a crude treatment of urban freight transport. T. Marvin Kraus. we overlook many policies that might signi cantly reduce the unpriced congestion externality imposed by trucks. freight delivery contributes considerably to traffic congestion. it is standard to treat trucks as car-equivalents. Evaluating such a policy requires estimating the costs. I understand. which accords better with my intuition. and assume that whatever policy is best for cars is best for trucks. in 16 . Large interstate (designed for freeway travel) trucks have trouble manoeuvering round corners or narrow streets. and their double-parking for deliveries where loading docks are absent severely reduces capacity. which include inconvenience costs of restricted delivery hours to shippers and receivers.In downtown Boston at least. the car equivalents of a truck is increasing in the number of trucks. and k the amount by which a truck reduces capacity. Then the link travel-time function may be written as t of congestion e(A. ∂T (w kT )2 (5b) ∂t( )/∂T Ak . however. though I have no idea how strictly it is enforced. on the books in Paris. = ∂t( )/∂A w kT (5a) A w kT .

that current practice is for all goods to be transported from supplier to receiver doorto-door. More research along these lines is badly needed.Winston. Their work corresponds to sound cost-bene t practice. Evaluating this policy requires knowledge of warehousing technology and practice. If sound cost-bene t analysis were practiced in all aspects of urban road engineering. but will be hampered by de ciencies in data. and Evans (1989) related to the economics of freeway/highway design. very considerable cost sav- 17 . that current practice is for all goods transported by inter-city truck to be unloaded at suburban warehouses and reloaded onto smaller trucks for delivery within the metropolitan area. and Sivitanidou before her untimely death was studying the spatial economics of urban warehousing. Regulating truck size for urban delivery would then be considerably costlier. Otherwise. and probably too for restricting delivery hours. Suppose. 2. at the other extreme. at one extreme. and maintenance. correcectness of the Kraus conjecture would strengthen the argument for smoothing truck traffic over the course of the day. the smaller truck size would substantially increase inter-city shipping costs. If door-to-door deliveries were to continue. study of the contribution to congestion caused by urban freight delivery should be high on the urban transport economic research agenda. relatively little additional cost would be imposed by restricting truck size for urban deliveries. In this case. More generally. Another freight-delivery-related policy is regulating truck size in areas of severe congestion. Geho (2000) is currently writing her thesis on urban freight consolidation. Suppose.the absence of congestion pricing. construction. new warehousing districts would have to be constructed where goods shipped inter-city in large trucks would be unloaded and loaded onto smaller trucks for urban delivery. The engineering economics of urban auto congestion I have already mentioned the exemplary work by Newbery (1988) and Small.

and should accordingly be more adaptive and exible. We can also contribute by applying economics to the nuts and bolts of road design. And at least a subset of urban transport economists keep abreast of the relevant engineering literature (e. At present there is little professional communication between traffic engineers/transportation scientists and transport economists. particularly with regards to pricing solutions.g. and the transportation science literature is drawing increasingly on economics. Rules for road resurfacing should take into account that future traffic volumes are generated by a stochastic process. etc. Though the work might be rather unglamorous and conceptually prosaic. however. To do transport engineering economics well requires some expertise in transport engineering. taking no account of the literature on irreversible investment and real options. and often faulty statistical analysis. and that technological advances in pavement design will occur. But we can do better. The situation is improving. we can make valuable contributions by demonstrating in practical applications how cost-bene t analysis should be done and by pressing hard for the adoption of sound costbene t procedures by state and local governments. that our understanding of pavement damage will improve. For example. Transportation Sci- ence and Transportation Research). We should not only encourage transportation engineering programs to teach more and better economics. To this I would add that engineering standards tend to be applied without reference to economic variables the discount rate. Ezra Hauer ( ) has argued that the bulk of transport engineering standards and rules are based on scant empirical testing and on-site data collection. perhaps volun- 18 . Eco- nomic studies which devise guidelines for such standards would be valuable. Economics is now taught in many graduate transportation engineering programs. even the most sophisticated cost-bene t procedures currently employed provide very crude treatments of uncertainty. the value of time. especially if second-best considerations were properly accounted for (Kanemoto (1999)).ings could be achieved.

streets wider. accidents. and policy in this area has been rather successful.S. Small and Kazimi (1995). an area of policy where effective policy remedies would be popular and relatively easy to implement. which contribute disproportionately to traffic noise. and pollution. and urban auto pollution The automobile congestion cost function is often interpreted as incorporating the costs associated with the noise. accident. traffic accidents. Horn-honking. We need a comparable body of work on the noise pollution generated by cars and on the economics of traffic accidents.g. participating in thesis supervision of transportation engineering students. It would not be difficult to design quieter cars. and attending transportation engineering seminars. and building setbacks higher) and can be very irritating. Road work and garbage 19 . however. Treating the congestion cost function as technologically determined therefore ignores all the behavioral and engineering decisions that in uence how traffic congestion affects the levels of noise. Automobile noise. could be dealt with by making it a traffic violation except when done to avoid an accident. but we should also include more transportation engineering in our teaching and in our research.. While I doubt this. It is. A German transport economist recently informed me that Germans are more concerned by the noise generated by traffic than by the time delays due to traffic congestion. a curse of living in Boston.teering to teach a course ourselves. What needs study is the cost-bene t calculus of alternative policies. and pollution caused by cars. Urban transport economists have paid little attention to traffic noise. Brownstone ( )). the same point applies to trucks and buses. it is nonetheless true that traffic noise in European cities tends to be signi cantly higher than in U. There has been considerable research into automobile pollution and alternative policies for reducing automobile emissions (e. as I gather is done in many European cities. 3. but at present car makers have little incentive to do so because each driver incurs only a fraction of the noise cost generated by her car. cities (where population is less dense.

but not the equilibrium stress level associated with driving. but it isn t. the design of roads for safety.. Boyer and Dionne (1987)) but have left other aspects such as the regulation of unsafe driving.g. I conjecture that driving stress is strongly related to the incidence of uncivil and dangerous driving: tailgating. It depends on the scheduling constraints she confronts and on how pleasant or unpleasant she nds driving. honking at the slightest provocation. but the topic is worthy Only half in jest.. especially to the effect of insurance on the incentive to drive safely (e. The primary tradeoff determining the many small decisions drivers make is between reducing travel time and increasing accident risk. running yellow and red lights. and accident follow-up procedures. Calfree and Winston (1998)). 2 20 . 4. in this context too. It is unclear how effective public policy can be in discouraging anti-social driving2 . for traffic engineers who. policies which affect the private costs of accidents may have a signi cant effect on how traffic ows. making dangerous and excessive lane changes. I would like to see bad drivers ostracized from the road. Economists have paid some attention to traffic accidents (e. Vickrey (1968)). Uncivil driving behavior and the value of time The standard model of auto congestion treats an individual s value of time as exogenous (e. The difficulty lies in designing a technology which allows a driver to report another without interfering with his driving concentration. A simple way to reduce the cost of traffic congestion and the congestion externality cost is to make driving more pleasant.g. Traffic accidents are costly not only for the direct damage they cause but also for the non-recurrent congestion (including curiosity congestion ) they induce. thereby reducing the value of time (though doing so also lowers trip price which stimulates demand).g.collection could be made quieter. Virtually no attention has been paid by either economists or engineers to the link between traffic accidents and the technology of congestion. The market takes care of automobile comfort. tend to choose policy with little or no explicit attention to economics. Thus.

The major exception is Vickrey. Pedestrian-car congestion interaction is potentially important.g.an enquiry by economists. to expedite passenger entry and exit.g. Relatively little has been done on the economics of mass transit. Sample topics include second-best train/bus size and service frequency and density. but merits serious analysis3. 5. As noted earlier. 21 . and there is a substantial literature which estimates mass transit cost functions (e. How can this be achieved? The current wisdom (Hall (). he did considerable work in the area but most remains unpublished and much is particular to New York City. Berechman (1993)) and a smaller one which examines capital-intensity bias (e. The proposal to impose a minimum walking speed on one side of Oxford Street was met with considerable amusement. in particular economies of service frequency and service density. Economics of mass transit and pedestrian traffic The bulk of the work done on the economics of urban traffic congestion has concerned cars. procedures to mitigate bus bunching (Vickrey (1979)). The economics of pedestrian congestion is unexplored. Hypercongestion has puzzled and intrigued transport engineers and economists. Reducing its incidence would provide substantial efficiency gains. and to reduce the congestion imposed by buses. Verhoef ()) is that 3 Apparently pedestrian traffic too is characterized by hypercongestion. presumably because until recently most of the innovative research in the eld was done in the United States where mass transit is relatively unimportant. Should jaywalking be discouraged? How wide should sidewalks be? 6. Hypercongestion Hypercongestion is the phenomenon whereby a given ow occurs at a signi cantly lower speed than is possible. Mohring (1972) explored some of the basic economic principles of mass transit. But urban transport economists have devoted little attention to the microscopics of urban mass transit. Small and Chu (). We would do well to follow his lead. Frankena (1987)).

that it is re ning models of commuting traffic at a time when 22 . each employer decides when to have his employees start work. signi cantly improve efficiency by modifying the work start times of its employees (Bonsall (1978))? 8. trading off the bene t from having his employees interact with more employees from other rms against the higher wage that he must pay his workers for commuting in congested conditions.hypercongestion is a transient phenomenon generated by events (such as a slow car in the passing lane or a double-parked car) which trigger backward shockwaves. The bottleneck literature takes the distribution of work start times as exogenous. There is then an interaction between interaction and traffic congestion externalities. as the dominant employer in many jurisdictions. Non-commuting trips It has been argued that late twentieth-century urban economics was preoccupied with re ning a model of the nineteenth-century city. holding xed the time pattern of arrivals (Arnott. and Lindsey (1993)). The endogeneity of the distribution of work start times has been studied by Henderson (1981). Accordingly. can the government. such events should be penalized. How speci cally this should be achieved is an exercise which blends traffic ow theory and economics at the microscopic level. In his model. 7. Urban transport economics can be subjected to a similar criticism. dePalma. Is it worthwhile for the government to attempt to modify private rms work start times? Whether it is or not. That literature has strengthened the case for congestion pricing by demonstrating that time-varying congestion pricing can substantially reduce congestion. That oversight has now largely been recti ed in the now substantial literature on bottleneck congestion. Because of excessive focus on the conventional models. initiated by Vickrey (1969). they overlooked the trip timing decision. Flextime and staggered working hours A generation ago urban transport economists (with the exception of Vickrey) essentially ignored the dynamics of rush hour traffic congestion.

since they are sensitive to the purpose assigned to chained trips. Fifteen years ago. it has proved very difficult to operationalize this conceptualization. the demand side of dynamic urban travel models will remain disconcertingly weak. One must take such gures with a grain of salt.an increasing proportion of rush-hour travel has a non-commuting purpose. Calthrop. the average time lost in searching for a parking spot may be as large as the average time lost due to congested traffic. 23 . and Rietveld (1995). However. Rather. 1987). To date. Verhoef. Gillen (1977. The difficulty with treating non-commuting trips within a dynamic model of congestion is that desired arrival time and correspondingly schedule delay costs become fuzzy. and cruising for on-street parking probably contributes signi cantly to downtown traffic congestion. payable at the destination for a trip by car. Shoup (1982. But. Until we do. Increased congestion will cause him to cancel some activities and reschedule others. a steadily increasing proportion of urban travel has a non-commuting purpose. and recently the gure for Chicago has fallen to thirty percent. Roth (1965). either xed or per unit time. the gure was bandied about that less than fty percent of rush-hour trips are for commuting. most of the work on parking by urban transport economists (Vickrey (1954). however measured. Glazer and Niskanen (1992). and the effects of cashing out employer-provided parking. and van Dender(2000)) regard parking as a price. Unfortunately. Particularly important is parking-related congestion. this is a convenient simpli cation.Shoup and Willson (1992). 9. it also misses a lot. Proost.1978). It ignores the contribution of parking to traffic congestion cruising for parking. Parking It was remarked earlier that the focus of our theory on link ow traffic congestion has distracted us from other forms of congestion and associated policy tools. At least for auto travel with a downtown destination. an individual schedules his activities taking traffic congestion into account. In modeling modal choice. Nijkamp.

in terms of the time to nd a parking space and to walk from a parking spot to the destination. one might model a driver s parking time costs as increasing in the ratio of the number of auto commuters already parked to capacity. and Arnott and Rowse (1999) provides rst steps in this direction. and so on. as well as the attendant uncertainty. the stochasticity inherent in the parking search process.and off-street parking fee structures. Axhausen (1990). and queues at entries to off-street parking. The dearth of stylized facts will impede modeling. and congestion due to entry into and exit from on-street parking. 4 24 . the allocation of on-street parking between metered. modelling drivers spatial search for parking. But this approach ignores the microscopics of parking-related traffic congestion. resident. A more satisfactory approach would be both spatial and structural. the spatial aspects of searching for parking. for example. and the allocation of land to parking. These phenomena can be crudely captured by reduced-form models. Axhausen and Polak (1995). private vs public ownership of parking garages. and unrestricted parking. and may require that we undertake the data collection and analysis ourselves.and off-street parking. the spacing between parking garages. Traffic engineers have devoted surprisingly little attention to parking-related congestion.4 This would permit the analysis of parking policies at an appropriately micro level: the allocation of land to on. secondbest on. time limits for on-street parking. It also ignores the congestion cost parkers impose on one another. double parking.capacity reduction from on-street parking.

By treating congestion as purely technological. the research agenda is microscopic in nature. Be that as it may. But looking at urban transport congestion through a single lens has distorted our perception. it has caused us to look at policy at too aggregate a level. Elaboration of this model has given our eld an impressive. Most of my criticisms concerned the link congestion cost function. but think that we have considerably underestimated the costs of implementation. but for the moment we should be pragmatic and devote more of our efforts to considering more immediate. I am certainly not opposed to congestion pricing. consistent. By treating congestion at such an aggregate level. by focusing excessively on congestion pricing. The topics chosen were not intended to be exhaustive. And by treating only link ow congestion. and coherent body of the theory. and may have overestimated the bene ts as well. By and large. there seems to be almost overwhelming opposition to congestion pricing. and mundane policy issues. it has caused us to overlook other forms of traffic congestion and the associated policy variables. I also put forward a research agenda. it has caused us to overlook many individual margins of choice than can be in uenced by policy. Its time may yet come. In my talk today. and complements rather than competes with the macroscopic theory which dominates our 25 . causing us to neglect many real-world aspects of urban traffic congestion and consequently to overlook many promising avenues of policy-related research. Conclusion We the community of urban transport economists have been less effective than we could have been in advising policy makers. leaving microscopic policy to engineers whose decisions are typically ill-informed by economics.4. but rather to illustrate the type of research that would address my criticisms of our existing body of theory. practical. I have argued that our preoccupation with congestion pricing has stemmed from excessive reliance on a single model framework which I have termed the canonical macroscopic link ow model of urban traffic congestion.

The topic headings give a avor of the type of research I have in mind: regulation of freight delivery time of day and truck size the engineering economics of urban auto congestion automobile noise. and urban auto pollution uncivil driving behavior and the value of time economics of mass transit and pedestrian traffic hypercongestion extime and staggered work hours non-commuting trips parking The research agenda I put forward was frustratingly even annoyingly non-speci c. 26 . and I hope than in the years ahead that my own research will go some way to meeting the challenges I have posed. But I am normally not someone who criticizes without offering constructive suggestion. traffic accidents.eld.

The two-mode problem: second-best pricing and capacity . and H. A disaggreagte model of the effects of parking guidance systems . New Approaches in Dynamic and Activity-based Approaches to Travel Analysis. Winsten (1956). McGuire and C. R. and J. Congestion Costs and Congestion Pricing for the Twin Cities. 206-225. [6] Arnott. Minneapolis/St. A structural model of peak-period congestion: a traffic bottleneck with elastic demand .References [1] Ahlstrand. D.Polak (1995). M.M. Journal of Urban Economics 45: 97-124. Rowse (1999). and J.Jones (Ed). Report No. A simultaneous simulation of activity chains . Alleviating traffic congestion: alternatives to road pricing . Boston College [4] Arnott.W.W. R. American Economic Review 83: 161-79 [5] Arnott R. MN/RC-96/32 to Minnesota Department of Transportation (Department of Economics. (1998). R. de Palma and R. Transport Policy 5:205-211 [2] Anderson. Aldershot.. A. University of Minnesota. Yan (2000). 27 . (1990).Mohring (1996).Lindsey (1993). Modeling Parking .B. Paul) [3] Arnott. Avebury. Review of Urban and Regional Development Studies 12:170-199 [7] Axhausen. Yale University Press. Department of Economics.. 7th WCTR Proceedings [9] Beckmann.W.. and A. working paper No. K. The rise and fall of the heroic transport plan for Stockholm . Studies in the Economics of Transportation. New Haven. I.B.(1994). in P. [8] Axhausen.282. C. K.

Winston (1993).. Flexible work hours and public transit in Ottawa . Optimal environmental taxation in the presence of other taxes: general equilibrium analysis . Second-best congestion taxes in transportation systems .(1977). 1367-78 [14] Bovenberg.[10] Berechman. Proceedings of the World Conference on Transport Research. Vol.(1986).(1996).L. T. 28 .(1996).(1978). Vancouver. M. A. R.H. The economics of road safety . S. eds. and C. L.C. in A. J. Journal of Public Economics 69: 83-102.F.2.L.Auerbach and M.Feldstein. B. (1993). 2nd edition. Environmental Taxation . Econometrica 45: 1703-15 [12] Bonsall. Peak load pricing of a transportation route with an unpriced substitute . Handbook of Public Economic. presented at the annual conference of the Roads and Transportation Association of Canada. and Goulder.(1998). Amsterdam) [11] Bertrand. Toronto. American Economic Review 86: 985-1000 [15] Bovenberg. September 23-4 [13] Borins.H. Transportation Research 21B:413-431 [17] Braid. J. North-Holland. and G. and Goulder. Ontario. L. The political economy of road pricing: the case of Hong Kong . A.E. New York [16] Boyer. T.Dionne (1987). Public Transit Economics and Deregulation Policy (NorthHolland. The value of automobile travel time: implications for congestion policy . Journal of Urban Economics 40: 179-97 [18] Brownstone [19] Calfree.

Urban Studies 37: 63-76.-F. D.Thisse. Parking fees and congestion . in J.W. thesis proposal. Economics of Cities:Theoretical Perspectives. The political economy of highway tolls and congestion pricing . Irvine.. The formation of economic agglomerations:old problems and new perspectives . Ph.W. [25] Gillen. J.-F. Synthesis of Highway Practice 210 (National Academy Press.Cambridge UK [24] Geho (2000).[20] Calthrop.. Journal of Urban Economics 21: 180-93 [23] Fujita. Parking policies and road pricing . (1987).-C. A.(1977).(1992). and J.Thisse (2000). Transportation 7: 69-85 [27] Glazer. N. Road Pricing for Congestion Management: A Survey of International Practice.Tsui. S. Regional Science and Urban Economics 22: 123-32 [28] Gómez-Ibáñez. Cambridge. Cambridge University Press.-M.A. and Niskanen.(1992). Transportation Quarterly 46:343-360 [29] Gómez-Ibáñez. D. NBER working paper 8278. MA [22] Frankena. [30] Hall. M.D. University of California. and bus scrapping .(1978). National Cooperative Highway Research Program.Whalley (2001). DC). [21] Chia. eds. E. Capital-biased subsidies. bureaucratic monitoring. Journal of Urban Economics 4: 186-99 [26] Gillen. and K. Washington. E. M. and J. J. Proost.Huriot and J.Small (1994). Estimation and speci cation of the effect of parking costs on urban transport mode choice . and K. Ownership and use taxes as congestion correcting instruments . van Dender (2000). 29 . F. Parking location decisions and the distribution of congestion .

Edward Elgar: Cheltenham. The economics of staggered work hours .[31] Hauer.A. in K. The consumer s time-of-use decision and optimal pricing and service in urban mass transit . A Theory of Incentives in Procurement and Regulation.J.V. R. Traffic Congestion and the Environment: Issues of Efficiency and Social Feasibility .T.Yoshida (2000). H. Vo l. Urban road pricing: public acceptability and barriers to implementation . and E.(1998). in Handbook of Transport Modelling. UK [35] Kanemoto.(1981). Hensher and K. eds. MIT Press. Journal of Urban Economics 9: 349-64 [33] Institute of Transportation Engineers (1982). P. University of Tokyo.(1993). 263-84. London [34] Jones. E. forthcoming. and J. R. Econometrica 36: 564-74 [39] Lindsey.Tirole. Oxford [40] Lindsey. Verhoef (2000).J. mimeo. Button.(1968).1. forthcoming in Journal of Urban Economics [37] Laffront.T. mimeo.-J. Y. Bene t -cost evaluation of transportation investment with underpriced congestion . [32] Henderson. Verhoef. Traffic congestion and congestion pricing . M. MA [38] Lévy-Lambert. D. Tari cation des services à qualité variable application aux péages de circulation . Prentice-Hall. mimeo 30 . Transportation and Traffic Engineering Handbook. [36] Kraus. J. Verhoef (2001). Cambridge. J. Road Pricing. Boston College. and E. eds.(1999). Congestion modelling . Elsevier Sciences. and Y. Button and E.

I.(1998). and J. M. J.Proost (1997). [46] Mohring. (1972).F. Cashing out free parking . and A.(1965). forthcoming. (1982). D. Journal of Urban Economics 44:352-66 [42] Marchand. Cambridge MA [47] Newbery. Transportation Economics. (1988). The efficiency of equilibrium traffic ows . Transportation Quarterly 35: 351-364 31 . Efficient congestion tolls in the presence of unpriced congestion: a peak and off-peak simulation model . [48] Parry. Tax deductions. and the double dividend hypothesis . Econometrica 56: 295-316. (1985).J. London [51] Sherman. I. Optimization and scale economies in urban bus transportation .D. Scandinavian Journal of Economics 99: 261-279 [44] Mohring. (1968). Econometrica 39: 565-76 [52] Shoup. H. Congestion interdependence and urban transit fares . Scandinavian Journal of Economics [49] Rotemberg. (1971). Journal of Public Economics 26:191-205 [50] Roth. American Economic Review 62: 591-604 [45] Mohring. Econometrica 36: 575-81 [43] Mayeres.Paying For Parking.McDonald. Optimal tax and public investment rules for congestion type of externalities . and S. L.Bento (2000). A note on optimal tolls in an imperfect environment . H.[41] Liu. Road damage externalities and road user charges . H. Ballinger. environmental policy. G. R. The Institute of Economic Affairs. (1976).

Road pricing for congestion management: the transition from theory to policy.Gómez-Ibáñez (1998). R. 29A: 141-56 32 ..(1992). Brookings Institution.Congestion Theory. forthcoming [58] Small.R.Verhoef. R. Traffic Flow Theory [62] Verhoef. P. eds. Transportation Research. (1997). Transportation Quarterly 46: 169-92 [55] Small.Kazimi (1995). K. [57] Small. E.C. Amsterdam.Margolis. P. (1965). and C. Employer-paid parking : the problem and proposed solutions . Nijkamp.(1986).. D. Urban transportation parables . Journal of Urban Economics.[53] Shoup. Road Pricing. Evaluating the effects of cashing out employer-paid parking: eight case studies . Transport Policy 4: 201-16 [54] Shoup.Winston. and Rietveld. K.. The economics of regulatory parking policies: the (im)possibilities of parking policies in traffic regulation . and J.Yan (1999) The value of value pricing of roads: second-best pricing and production differentiation .. On the costs of air pollution from motor vehicles . and the Environment: Issues of Efficiency and Social Feasibility. [59] Strotz.Evans (1989) Road Work:A New Highway Pricing and Investment Policy.A. C.(1995). D. North-Holland. Journal of Transport Economics and Policy 29: 7-32.. K. in J. Traffic Congestion. E. ed. The Public Economy of Urban Communities (Resources for the Future:Washington. D. Edward Elgar: Cheltenham. and J. DC) [60] Syski.Button and E. UK [56] Small. [63] Verhoef. and Willson. in K. Washington. 263-84. [61] Transportation Research Board (). K. and C.

Reprinted in Journal of Urban Economics 36: 56-65(1994) [68] Vickrey.(1969) Congestion theory and transportation investment .(1959).S. E. W. tort law. (1961). A. E.S.(1978) Price-induced distortions in urban highway investment . The theory and measurement of private and social cost of highway congestion .Rouwendal and Rietveld.(1968). American Economic Review 59: 251-61 [71] Vickrey.(1954). and insurance: an economist s critique . W.(1996). W. P. Nijkamp. Spacing out those gregarious buses . New York Times. Nov.A. Journal of Urban Economics 40: 279-302 [65] Verhoef.S. Statement on the pricing of urban street use . P. Automobile accidents. J. Tinbergen Institute. W. Hearings.C.Congress Joint Committee on Metropolitan Washington Problems.S.(1983) Optimal road capacity in the presence of unpriced congestion . and Rietveld. externalities.(1999). Discussion Paper TI99-066/3. Traffic Engineering Magazine.S. June [72] Walters. Econometrica 19:679-699 [73] Wheaton.D.[64] Verhoef.[Reprinted in Journal of Urban Economics 36: 42-65 (1994). November 11. P. E and K. Law and Contemportary Problems 33:464-487 [70] Vickrey. U. Product differentiation on roads: second-best congestion pricing with heterogeneity under public and private ownership . Bell Journal of Economics 9: 622-32 [74] Wilson. Congestion caused by speed differences . The economizing of curb parking space . W. Journal of Urban Economics 13: 337-57 33 . Amsterdam-Rotterdam [67] Vickrey. W. J. Second-best congestion pricing: the case of an unpriced alternative .(1979). Journal of Urban Economics 45: 533-56 [66] Verhoef. Small (1999).S.] [69] Vickrey.

METROPOLIS 34 .Hearn dePalma.

Traffic volume Figure 1.Fig1micrscopic. Diagramatic representation the basic of model of trafic congestion .nb 1 $ trip MSC ACH= MC = t* D Q .