Transportation Center

4 downloads 0 Views 1MB Size Report
life-cycle factors on current housing decisions, these studies inadequately explain both overall .... Uncertain Job Location, Urban Form, and the Journey to Work.
The Influence of Expected Suburbanization on Urban Form and the Journey to Work

Randall Crane

Working Paper UCTCNo. 240

].’he University of Caldfomia TransportationCenter Universityof California Berkeley, CA94720

Tile University Transportation

Df California Center

"/’he University of California Transportation Center (UCTC) is one of ten regional units mandated by Congress and established in Fall 1988 to support research, education, and training in surface transportation, The UCCenter serves federal Region IX and is supported by matching grants from the U.S. Departmeat of Transportation, the California Department of Transportation (Caltrans), and the University. Based on the Berkeley Campus, UCTCdraws upon existing capabilities and resources of the Institutes of Transportation Studies at Berkeley, Davis, Irvine, and Los Angeles; the Institute of Urban and Regional Development at Berkeley; and several academic departments at the Berkeley, Davis, Irvine, and Los Angeles campuses. Faculty and students on other University of California campusesmay participate in

Center activities. Researchers at other universities within the region also have opportunities to collaborate with UCfaculty on selected studies. UCTC’seducational and research programs are focused on strategic planning for improving metropolitan accessibility, with emphasis on the special conditions in RegionIX. Particular attention is directed to strategies for using transportation as an instrument of economic development, while also accommodatingto the region’s persistent expansion and while maintaining and enhancing the quality of life there. The Center distributes reports on its research in working papers, monographs, and in reprints of published articles. It also publishes Access, a magazine presenting summaries of selected studies. For a list of publications in print, write to the address below.

Universityof California Trampor~fion Center 108 NavalArchitectureBuilding Berkeley,California 94720 TeI: 510/643-7378 FAX:510/643-5456

Thecontentsof this repor~reflect the viewsof the authorwhois responsible for the facts andaccuracyof the data presentedherein. Thecontentsdo not necessarilyreflect the officio3viewsor policies of the State of Californiaor the U.S. Department of Transportation.This report doesnot constitute a standard, specification,or regulation.

The Influence

of Expected Suburbanization and the Journey to Work

on Urban Form

Randall Crane Departmentof Urban& Regional Planning Universityof California at Irvine lrvine, CA92717-5150 and Center for U.S.-Mexican Studies Universityof California at San Diego La JolIa, CA92093-0510

Working Paper November 1994

UCTCNo. 240 TheUniversity of California Transportation Center Universityof California at Berkeley

Abstract Standard urban models assume residents never think about their next job. More likely, the individual value of a given home and the choice of commute length are based not only on the current job site, but also on the expectation of where future jobs will be and the likelihood of both job separations and residential moves. The first factor lessens the value of access to the present job, while the second determines the opportunity cost of moving. Both sets of factors lead to flatter rent gradients and more sprawl than predicted by standard theories. The analysis further suggests that relatively stable jobs are likely associated with relatively shorter commutes. Past studies of the regional balance of jobs and housing, of ’wasteful ~ commuting, of differences in the length of commute by gender, and of spatial tests for discrimination in housing and local labor markets have neglected these considerations, and may yield biased results as a consequence.

*I am gratefulfor the thoughtfulcommentsof M. Boarnet,G. Crampton,G. Fielding,C. Lave, Ko Small, two referees, and especiallyJ0 Brueckneron earlierversions,and fundingfrom the Universityof CaliforniaTransportationCenter and a UC Irvine FacultyFellowship.

I. Introduction The conventional :~espects, pivotal

theory of urban

as any theory

than

others,

must be. Of course

depending

rsay have direct bearing validity

of subsequent

urban

areas.

life-cycle

both overall

commuting

live.

a.ccount

approach

generalized

for a fuller

site for most

the static

future

these

in commuting

are more

from a model

as well as the that such may be the the structure

job locations

studies

with

patterns

of

and

inadequately

explain

among

various

To

most analysts

for many such rent and density

and the population.

of the resulting

relidential

location

may change

sites.

(Examples

include

reasons°

choice

the household

the value

of access

the opportunity

in this simple

We consider

to move.

of a place

job, but also on the expectation

other

determines

pattern.

but allows

on the characteristics

The ultimate

the next job will be, and (2) how often lessens

centerse

and where it is costly

to depend

where

the usual framework

to

generates

locations,

employment

factor

where

of the commute.

of multiple

based not only on the current

E~:ending

when choosing

a setting

and the job and home

off the value

[381.

the cost of a move is per~dtted

while the second

will trade

this story

the length

to allow

of employment

where job location

The first

employment,

major employment

model

[37] and Yinger

explanation

an environment

location,

argues

the cost of housing

with multiple

This paper also considers

therefore

results

that each household

rents and densities

for labor markets

White [351, Wieland )

individual

of potential

and differences

job against

-- i.e., suburbanization

Moreover,

absence

and by extension

decisions,

is to argue

with a single

equilibrium

have simply nodes

behavior

to their current

In an area

declining

housing

of the real world

Their

This paper

in many

groups.

The standard of access

on current

at hand.

to work,

the influence

to be limited

of theoretical

strategies.

of the journey

By neglecting

factors

population

on the usefulness

is known

some elements

on the problem

empirical

case with past studies

structure

will choose

of the

to live

is

of (1) to move for

to the current

job

cost of moving.

but important

manner

sheds

light

2 on questions policy

raised

issues,

’wasteful

evidence

population

Cervero

that the average

exclusively Gordon,

expect

(The

is pretty

on minimizing

commuting

their

hypothesis,

trip.

Other papers

have

local amenities

and multiple

employment

controlled

for possible

influential

in both

et al. [13],

Madden

this paper suggests small job change mile,

and less

workers

to have

importance

minorities

the amount

[36].)

of

is discussed

in

Los Angeles

the area supports rather

is

home or who

from

the

than focusing

and Small

centers,

[i0,Ii];

the equilibrium

to account

include

for the role of

assume

Giuliano

that

is that

that households and Small

studies

looking

[II],

at

of men and women have not effectively

location

characteristics choice.

and White

a shorter

commute

(See,

potentially for example,

[24].)

The model

for persons

high home move costs, higher

Compared

with

characteristics of gender

one view of the "spatial

face more discrimination

[161,

to minimize

all of which

(These

and career-cycle

costs, relatively

by Hamilton

It is also clear

behavior

we would predict

all these

locations

the segments

question

(Giuliano

the analysis

and Lic [23] and Madden

job turnover.

among

Evidence

open,

necessary

extended

job and residence

of these dimensions

Finally,

introduced

job changes.

and White

for the life-cycle

length

mismatch

[15]).

far exceeds

in the con~nuting

balance

options

commute

work-home

differences

particularly

to soon move into a larger

future

commuting

Small and Song [31],

here,

and the spatial job and housing

et al. [12,151.)

aggregate

do not plan ahead

current

of transportation

of ~obs and housing,

much the same throughout

the current

Kumar and Richardson

balance

jobs/housing

[9] and Gordon,

a number

lengths,

between

introduced

are keeping

The wasteful

gap in commute

who either

commute

that people

on the regional

to change.

[3], Giuliano

work regarding

the connection

by the considerations

job locations

notion

empirical

the gender

In general,

the working expect

including

commuting’,

hypothesis. weakened

by recent

men as a group, except

possibly

differences mismatch’

in housing

Gordon,

presented

in

with relatively

cormm/ting

we would

costs

expect

the latter.

per

women

However,

the

have yet to be measured.

hypothesis

markets

than

is that racial in labor

markets,

such

3 that

they are spatially

commutes

for blacks

length

of longer

than it does

in this paper calls

the first.

Short

relatively

stable

consistent

with a housing

provides

there

is reason

little

market

to believe

the current

The main idea location

a worker

evaluating

alternative

and work

well account

home sites,

By extending demonstrates ~mpirical

commute

the local

Rather

research

commute

more

by

costs.

This

of minorities services.

have either

is

yet

The

is diminished,

however,

if

relatively that future

jobs

market

are unknown

over their

entire

time horizon

than

a standard

urban

importance

questions.

location

number model

think

only about

work later. reasons.

of working

where

somewhere

the expected

in straightforward

begins

factors

live

they

a model

else

by all the factors: costs

of

changes.

fashion,

for a variety

by specifying

when

may

Any one household~s

plus three additional

locations,

with

they

Moreover,

of both job and residence

of these

The paper

that if the

and as such is explained

employment

and the expected

We show

labor

they might

job location,

of future

model°

to move for other

has used to explain

the potential

commute

set of results

or the expectation

urban

costs

where

they plan

as it is to the present

distribution

to social

commutes

is thus as tied to the probability

tlhe literature

and high

and

of a lack

in part be explained

individuals

in a basic

think about

locally

changing

to longer

to live.

location

The expected

or access

sites within

current

things

discrimination,

home location.

places

for how often

of longer

to understanding

the choices

job site is secure

will consider

now, they also

costs,

that the observed

employment

certainty,

may at least

sources

is demonstrated

of future

market

the second

that constrains

discrimination

faith that the distant

will be nearer

into question

high moving

employment

of housing

of housing

Evidence

has in turn been used as evidence

commutes

job location,

centralized

contribution

commutes

for jobs.

in those markets. 1 The approach

discrimination

outlined

when competing

has been used as evidence

the lack of evidence of housing

constrained

this paper

of topical of urban

IseeZllwood [61, Gabriel and Rosenthal [8J, Gordon, et al., [14], Hughes and Madden [17~, lhlanfel~: and Sjoquist [18,19,203, Ihlanfeldt and Young [21], Kain [22], Price and Mills [28], Taylor and Ong [32], Zax [39,40], and Zax and Kain [41].

4 structure.

It then presents

employment

location.

approach

The market

to valuing

employment

lead to cities

which

the discussion

to include

moving

costs

2.

pattern

analysis

associated

access,

to life

Urban starts

one-way

to their

commute

numeraire

length

composite

in miles°

land and capital

production

function

Household location,

h = f(L,K),

behaviors

3 where

factors

Section

to work,

Suggestions

to

4 extends

arguing

that

for empirical

to Work

benchmark

of a single

side,

of the residential

employment

inelastically

also

consumes

to the concave

site

supplies

cost of r(z),

housing

where

housing

is produced

a unit of z is the

h, and a

by competitive

constant-returns-to-scale

L is land and K is capital.

and thus the willingness

may be summarized r(z,U)

model

Each household

according

future

section.

at a commuting

good x. On the supply

for these

assumed.

circumstances.

each household

site,

uncertain

with the conventional

of the journey

and the Journey

urban

that

employment

firms from

usually

with the familiar

with a standard

"Downtown’. 2 The model assumes labor supply

Form,

with

the tendency

that

choice

cycle

example

is compared

suggesting

in the concluding

Job Location,

The formal

rent gradient

the individual

are offered

Uncertain

two period

are more decentralized

vary according

implementation

a two center,

of to pay for housing

at each

by the bid rent function

-maxh, x {[y - x - r(z)]/h:

U(h,x) (i)

This represents

the maximum

maintain

utility

applying

the envelope

price

the household

level U. Differentiating theorem,

can pay per unit of housing

(I) with respect

to distance,

and

and

gives:

2This paper does not explain job location, i.e., labor demand and the suburbanizationof em91oyment. It is entirely concerned with the suburbanizationof labor given the suburbanizationof employment. For related approacheswith somewhat different purposes, see simpson [ 30] and Nijkamp, van Wissen and Rima [27]. 3This is the so-called Mills-Muthmodel of urban residentiallocation; see Mills [25 } and Muth [261 for foundations, and Brueckner [2} and Fujita [7] for recent discussions and extensions. Note that while the story below extends the basic model to include some dynamic elem~_nts, we maintain the classic assumption that capital is perfectly malleableL, distinguishingthis approach from the dynamic urban models of Brueckner [I} and yon Rabenau [33], among others

dr(z,U °) ar( dz

-

dz

This is the standard value

of access

the central

result their

employment

are multiple

employment

different

to reflect

[371

discounted

value

consequences extent

density

m~tlticentric most models occupational

the equilibrium

to the nearest

area

around

as a whole

of either

static

mobility

story,

story

which

are important,

that conventional

Second,

urban

reflect

rents

theory flatter

a choice

regarding

future

process

this

will

and be

(e.g., White

rents

Thus,

will tend

the

on a smaller by

this approach

fails

to

how bids are affected

of residence

employment

First,

areas

would

site.

There

the

are two

we expect

that

to the

will tend to be more

suggest.

than typically journey

if

is chosen.

will then tend to reflect

of each commute

characteristics

center

decentralized,

are important.

the individual

where the length

and life-cycle

expectations

for,

workers

gradient

though

to consider

metropolitan

will be everywhere

models.

demand

the

per se.

when the place

likely

site.

site,

however,

or change,

to the most

perspective

employment

will be more

the multicentric

of access

suggest,

rent

equilibrium

each employment

we show that equilibrium

These results reflect

[38]).

is repeated

of this

gradients

of residents,

In that instances

these factors

decentralized

and the mobility

analysts,

job site is uncertain

speaking,

reflects

that question

and ¥inger

This paper extends

Roughly

areas

is more than one employment

As a fundamentally

an individual’s

of housing

in the city has inelastic

of metropolitan

centers,

by several

The metropolitan

the effects

everyone

in the price

than (2) if there

pattern

construction.

variation

features

the value of access

monocentric

capture

Spatial

site. common

Wieland

scale.

result:

T~ increasingly

s~omewhat

(2)

h{)

to the only location

jobs. As shown

[35],

~ ’(z)

In particular, predicted

to work will depends

rent

and

by

be longer

critically

than

on the

of the commuter. where

residence

suburbanization

bids for any site

of employment,

and the spatial

6 variation below,

in moving

and commute

each household

to move. If home

costs

considers

location

where

bids will simply

con~nuting

costs to every expected

a job at that

each feasible stream

of commuting Before

largely

depend

professor

worker

A young couple within their

planning

city

decisions.

locations

span two time periods,

in the next period,

is subjectively

are introduced

expect

to lose that they

where

of their

That is, the likelihood as the share

of being

house

to stay analysis

distribution

of the

in is

in the literature. site valuation,

section.

Downtown

Though

a

for this

and tomorrow.

job tcmorrc~.

they

at a time.

into a larger

into two parts:

at 2 sites:

note that

while

a few months

unexamined

in the next

are unsure

on the proportion

calculated

largely

costs.

a university

may be planning

the aggregate

today

elements,

the area,

The motivation

this process

employment

employed

at each site.

years.

can affect

to decompose

but they

based

twenty

moving

idiosyncrasies,

to move-up

at

both the expected

these

within

of

of access

On the one hand,

site

and

of future

and with total

with children

area in a manner

working,

expectations

a family

differences

The latter

with (potential)

(Zs) o Expectations

status.

employment

value

minimizing

individual

may be expecting

while

in a metropolitan

It is simplest moving

a family

house for another

that these observable population

from

they plan

the initial

of the value

may not work at the same site more than

the next few years, current

Aside

then

described

by the probability

conceptualizing

age and family

may have only one feasible

construction

weighted

with each location

observable°

and how often

of the present

in time involves

scheme

In the model

variable,

That is, the calculation

a bid-rent

on occupation,

choice

job location,

costs associated

are generally

work later

a comparison

at any point

introducing

such factors

involve

location.

location

they might

is a nonstochastic

subsequent

having

over the time horizon°

Consider

and

a linear

(D) and the Suburbs Today

they

everyone

expect

to be

that job will be. They form

employment employed

of employment

base expected downtown

in that

to be located

in the next period

occupation

expected

to

be found there. Everyone

in the city forms

rational

expectations

in this regard,

based

perhaps

on the !published the city’s

reports

employment

the suburbs

tomorrow

Our story begins relaxed

later°

the present either

of the local

in a particular is denoted

by assuming

Expected

value

occupation

forecasting

commuting

costs

second

depending

share

on whether

to be located

left downtown

costs.

Current

the worker

in

this will be

are current

commuting

now works

of

then ! - ~ 4

the same ~, though

over the two periods

period

model. The share

that is expected

by ~ , with the share

all residents

of expected

"~(z) or T,(z),

university’s

costs

costs

downtown

plus

are

or in the

suburbs. The location downtown.

As one considers

increases, depending %~ile

of the suburban

residential

note that commuting on the location.

T~(z)

next

expenses

center

locations

for suburban

T~(z)

is at distance

further

from

workers

can either

~ from

downtown,

< 0 for z < z S, and T~(z)

i.e.,

as z

rise or fall

> 0 for z > z s,

> ev erywhere. Fu rthermore, th e re levant me asure of com muting cos ts,

are either period

costs

Thus

and hence the value of access costs

employment

to employment,

TD(Z ) or Ts(z),

as mentioned.

are (i -~)TD(Z ) + ~(Z).

is therefore

is in part Expected

The present

value

forward

looking.

commuting

costs

of all expected

Today’s in the commuting

either

or

for current

downtown

workers

and current

suburban

workers,

respectively,

where

6 - 1 / (i + i) and is the inte rest rate . If we focus the analysis

by requiring

that x, h and r be constant

over the two

4~.~his model is describedin terms of a single-workerhousehold,but the general approach can incorporate multiple worker households as well by expanding the interpretationof ~ to capture other elements of the importance of each job site. For example, if Gj measured the stability of the job of w~rker j and yj captured other aspects of the importance of worker j’s job to the household, then we would define the household ~ . I (~J + )’J )~ sun~uing over all household workers° 2

5 periods,

the present

value bid rent function

may be written

+ 6Xy (I + 6)r(z,U~

The rent gradient

for a downtown

(i + 6) #rD

The gradient

=

has the same

commute

world with

cost

Compared rapidly

those workers

The second

happens

commute

¯ ~ = t. Commute

of commuting

6~(r~-r~)

stable

jobs,

in equilibrium,

costs

less rapidly

bid rents

there

with distance

to the standard

in a

model

of job location

sign for z > z will thus decline

at least

less

up to ~ whenever

is also the suggestion

than in standard

for a worker

The first

is the effect

workers

-

currently

that

models. employed

in the suburbs

with distance

with T~, though

The effect

costs are a constant costs for current

for z < z s if ~ < I; i.e., beyond

both are positive

and explore

these

the suburban

expressions

workers

if the

center

depends

there.

$t per unit distance,

suburban

is

(5)

h

is not secure.

To get at that question

+ 6~(T~-r~)]o

it is the analog

downtown

rent gradient

job location

on how T~ compares

to assume

value

term

model with

(I + 6)

suburban

+ 6)T~

will be more decentralized

This also increases

(3)

(4)

for z < z s and s . of indeterminate

for current

on what

The comparable

~ °}

U(h,x)

is then

aa -[(1

in the present

to a two period

~ 0. Depending

worker

: s.t.

11+ 6(i

sign

and is negative

from downtown

h

fixed job locations;

gradient.

uncertainty,

_

E[~]

h

(i + 6)T~ is the change two period

x)

i) " , maxh,x-

as,

are then

further,

it is convenient

such that T~(z) T,(z)

. tz and

= t(~ - and

5A more general treatment would extend the time horizon and allow for more than two possible job sites. For an infinite number of possible job sites along a linear city in a two-period setting, for example~ E[ T ] = ~o(Z) ÷ 6f, f( z)T(Z) dz, where ~o( z ) is the initial commute expense f(z) is the probability density function of Job locations. This could be extended to a multi-period setting by summing expected cc~mute costs in each period over the time horizon.

!

f

ihence T, = -t for z < zs, and rs(z) = t(z - zs) so that T s s= t for z > z In

that

case,

the

sign

of

the

downtown

workers’

bid

rent

gradient

(4)

is

the

same aN

-[I+6(I-2~)1 for z < z,, and

Note the gradient negative

between

is negative

downtown

outside

zs, but it can be either

and the suburbs°

It is definitely

(} = 0 or ~ s 1 / 2. That is, if the prospect sJufficiently

likely,

downtown

for workers

even

externality current

argument

downtown

suburban

center

the current

the rent profile

to derive

job is only temporary, the remainder

slope upward

(Richardson

result.)

and the worker

of their

only where

working

life,

expects

is

with distance

[291 used

In the extreme

or

a spatial

case where

the

to be employed

the present

value

from

in the

of access

to

job can be quite minimal.

In this example, same

downtown.

a similar

negative

of a job move to the suburbs

will actually

who now work

positive

the sign of the suburban

workers’

rent gradient

(5) is the

as

-[6(I

- 2~) - i] for z < z,,

and

Again,

the gradient

definitely

between

positive

only

for both types of workers This example different period.

values

Figure

one downtown

downtown

and the suburb

if 6 = 0 or ~ ~ 1 / 2. Note outside

is illustrated

of z s are the same, in Figure

of ~, the likelihood

la gives the market

and a second

Ib and ic illustrate

may be positive

suburban

that the slopes

employed

rent in an economy

bid rent gradients

over

in the suburbs

in the next job sites,

z s from downtown.

a stochastic

of ~.

rate and three

with two secure

job site at distance

and is

of bid gradients

and are independent

1 for a zero discount

of being

or negative,

Figures

job site economy.

In

i0

Figure

ib, the bids reflect

percent Figure

probability

of being employed

ic increases

that probability

increasing

rent gradient

In a stable adjust

In the special

rent depends

turn will source

be indifferent

households

depend

will locate

in expected

Figure

la shows both how workers

long as the expectation

of having

half; that current

is, so long

suburban

In particular, with the lowest

as z increases),

and will thus locate largest

there.

~’s (the most stable nearest

6 prediction

the suburban

that workers

values

within jobs)

They

downtown

either

center.

with the most stable

Thus,

their

is greater

in

one be

employment to do so so than one-

and ~> 1/2 for

By (4),

are possible,

suburban

sites.

downtown

workers

bid rent gradients

bids closest

to downtowns

workers

bid rents

the model

across

which

might

will continue

will have the steepest

from (5), those

the and thus

one or both employment

will have the steepest

employment

Another

alternatives

each group.

of the

if bids differ

around

workers

will have the winning

Similarly, jobs)

to cluster

of ~, other

around

~ differs

~’s (the most stable

(they slope upward

here by ~.

and stable.

respect,

In this model,

a job in the same location

clustering

assume

profile.

will

population.

in equilibrium,

represented

certain

With other

some with and some without

However,

of employment.

will tend

and rents

in the city,

location

as ~< 1/2 for current

workers.

important

an

site.

of the market

locations

cost

ic gives

The relationship

bid rent everywhere

commute

job stability,

are both

in every

at different

is the initial

site when job locations

to moving

another.

home locations°

on their individual

of bid variation

differences

among

Figure

up to the suburban

on the characteristics are equal

in the next period.

Note that

from downtown

of one site over

that they have a 20

center

can be no advantage

case where households

would

individuals,

locate

to 80 percent.

rent will equal each individual:s

commuters

workers

in the suburban

there

the benefits

bid rent to the market

by downtown

with distance

equilibrium,

to eliminate

market

the belief

with

the

and will thus

supports

jobs will have the shortest

the con~nuteso

61 am grateful to the editor for suggesting this resulto Note also that in the suburban case, there is no sorting of workers by job stability outside of zsbecause the slopes there are independent of ~.

Ii

Figure all workers medium

and unstable.

job location.

and current groups

suburban

of workers,

iFigure

with respect Figure location

3 initially

commutes.

specific

for 6 worker

to job stability:

of stability workers

downtown

Moreover,

Again,

market types.

centers

these parameter

workers

jobs.

zones III and IV), and thus

rent,

values

site,

a pattern

also demonstrate

jobs

might will

Those

jobs --

bids near that site. bids generating

a

of residential jobs;

workers

value

there.

2b

i.e.,

shorter that

workers

with very

access

to the

This does not mean they

in the next period,

only that they have

of doing so.

A naive analysis degree of ’wasteful

of the locational

co~7~6~ting,’

irrational conclude

may also

suburbs

(zone IV) is evidence

it, if these patterns

patterns

that

in Figure

2c would

as that term is traditionally

about the behavior

~alysts

conclusions

us 6

workers

than some suburban

as well.

employed

work in the other location

a high expectation

is nothing

currently

gives

the possibility

jobs look so far ahead in this case as to actually workers

jobs,

with the most stable

with the more stable

vice-versa

to their

and so on. Figure

the winning

from downtown

stable,

stable

the most stable

with

Say that

in the suburbs.

for downtown

those

It reveals

by those

may live farther

job site more than

This

job site -- will have the highest

for the 6 worker

near employment

with

employment

workers.

highly

with ~ = 0 have

stable

types.

is with respect

and 3 working

of ~. Those

the current

2c shows the resulting

will necessarily

those

downtown

values

to their current

s:ome downtown

there

measure

working

bids nearest

pattern

clustering

other

respect

with ~ = 1 also have

the bid rents for suburban

unstable

result

2a shows what the bid rents of the three groups

have the highest

(i.e.,

of this

with

Each

Thus current workers

like, for three

draws

an illustration

are split into 3 groups

stability,

current

look

2 provides

the presence

of housing

are correlated

with

constructed

either

market race

might in fact be true the example

defined.

a high

However,

in this example.

of s~e downtown

or labor

suggest

workers

living

discrimination,

or gender,

suggests

in the

or lack

or both.

While

that job stability,

!2

which itself explaining

may well have racial

or gender

components,

could

also play a role in

these patterns.

3. Life Cycle Factors The analysis This could relocate

thus far has abstracted

be relaxed

in several

to the sites of their

employment

ways.

winning

base for each employment

can be derived

by redefining

in the previous

section.

For example, bids,

site,

individual

To explain

the choice

of site

movement

explicitly

is another

strategy.

approach

might attempt

costs associated

For example, generate

the same site

residence

individuals

is relatively living

of employment

might be expected

located

elsewhere,

they may rationally different

commutes

by the number location,

small.

choose from

of workers

rents

a longer

her home,

Modeling

costs

costs

as

also

the costs

of

for example, is, based

that

on the

even where

current

in the location expect

for the stability to move°

of later

them by relocated

site. 7 These

of

to see such

it is more costly

to the likelihood

commute

one’s

decisions

home°

are naturally

of their

That

jobs being

to in part anticipate

the stability

they

the cost of changing

and we might

her home,

to follow

in each household,

and their propensity

commutes

each household

to a change

sensitive

the same home

gradients,

of expected

they work -- controlling

owns

since it is costly

the

density

in Zax [40],

the same relocation

near where

to be more

of individual

As pursued

Adjusting

If a person

in terms

the hedonico

easy in that instance,

relatively

location.

faces

can costlessly

characteristics.

bids. If a person

may be relatively

employment

person

not everyone

along

costs

how ’footloose’

with their individual

of home location.

if households

and the corresponding

the pattern

explaining

choice

then the size of the city,

commute

requires

to identify

from the actual

later

Thus, and

affected

employment

to move for any and all reasons.

7These assertions are supported by the Wachs, et al. [34] study of commute behavior over time in Southern California. Workers who moved into rental housing tended to move closer to their job site, while workers who moved into a home they owned tended to increase their commute length.

13 Life cycle factors

are particularly

the last of these factors, reasons

than following

household apart

from the location

of their

the observed

between

with

or smaller, planned

commute location

change.

ten years,

section,

associated

modeling

increases,

has expectations as their family

will be changing markets).

to future,

attractiveness

Most models

sense

process

life is

of commute

complex

costs

that the twoof job

if the job location

is complicated

and that any choice

commute

lengths.

by the of home

Put another

they will want to move (as their

spatial

section,

of urban form assume

is valid is an empirical location

working

labor

these

etc.)

and how often

market,

and between

expectations

affect

way, income they labor

the

value of each location.

research

orL both urban form and the choice

residential

may

move due to each expected

when they buy a home, a given

in the previous

for empirical

finds itself

a month.

this decision

and hence the household

Implications

quite

the opportunity

can look at the stream

move may make

about how often

jobs (both within

As argued

for reasons

stream

are more

as the worker

as yet unknown

expands,

over one’s

in the discounted

expectations

but not if it only lasts

is relevant

each worker

approach

tradeoff

that this may not be the last job change,

location

or a

to move in any event.

the household

with each possible

That is, a particular

The task of actually knowledge

spatial

the move. Multiperiod

cost savings

homeowners,

reduce

was going

in

to move for other

residence

moves

the variation

length.

and the savings

story in the previous

make plans

if the household

ways the critical

the cost of moving

associated period

commute

in explaining

may plan to become

job. Such

in job location,

Put another

4.,

Renters

the point in the life cycle or life course

influence

lasts

employment.

occasionally

may hope to move into a larger,

cost of a change Thus,

as people

important

that the influence

of commute question

may well be based

length

of uncertain

is minimal.

to be sure,

Whether

but this paper

not only on where

job mobility

the current

that

argues

that

job is located,

14 but also on the expectation area,

and how often

of where

the household

plans

this view suggests

that controlling

relatively

commutes

or jobs

shorter

within

have rarely, length

life-cycle empirical

They depend

factors, studies

market,

and dynamic

A lower

has implied distribution

and higher

therefore

that workers of future

of expectations°

subjectively

the risks of committing

the site

may prove

their practical certainly

less

be more important

metropolitan

areas.

moving,

than ideal

significance

in workers,

of changing costs°

These

specifications

argues

should

those

either

with

homes

variables

of commute

characteristics,

must

in addition

be given

have rational

employment

not the rationality

when they consider

In particular,

to

more weight

in

to work.

is made for convenience

apartment

moving

profile

but that assumption

assess

probability

the metropolitan

reasons°

differences

in past empirical

which this paper

be within

to move for other

on occupational

of the journey

The analysis spatial

also have:

the local labor

jobs will

for all other

if ever, been included

models.

future

only.

It is whether

for these

The empirical we believe

themselves

in the sense for later

as employment

job or other

continues

results

that

at hand is

people home or

the possibility

opportunities.

these

over the

to matter,

question

to a particular

they consider

has yet to be demonstrated,

expectations

Though

considerations

to decentralize

that

in major

will

15 References

J. K. Brueckner, A dynamic model Economics, I0, 173-182 (1981).

of housing

production,

2.

J. K. Brueckner, The structure of urban equilibria, of Regional and Urban Economics, Vol. 2," Elsevier,

3.

R. Cervero, "America’s Suburban Unwin Hyman, Boston (1989).

Centers:

Journal

of Urban

(E. Mills, ed.) "Handbook Amsterdam, 821-845 (1987).

The Land Use-Transportation

Link."

G. R. Crampton, The trade-off model, ’wasteful commuting’ and the labour market search process, Discussion Paper No. 40, Department of Economics, Reading University (1988). 5.

G. R. Crampton, Labour market search and urban residential structure, Discussion Paper, Working Paper, Reading University (1992). D. Ellwood, The spatial mismatch hypothesis: Are there teenage jobs missing in the ghetto? in "The Black Youth Employment Crisis," (R. Freeman and H. Holzer, ads.) university of Chicago Press, Chicago (1986). M. Fujita, (1989).

7.

"Urban

Economic

Theory."

Cambridge

University

Press,

New York

S. Gabriel and S. Rosenthal, Conunute times, compensating differentials, and racial discrimination: An analysis of black and white workers, unpublished, University of Southern California and University of British Columbia (1992). .

G. Giuliano, Is jobs-housing balance a transportation Research Record, 1305, 305-312 (1991).

issue?,

Transportation

10.

G. Giuliano and K. A. Smalls Subcenters in the Los Angeles Science and Urban Economics, 21, 163-182 (1991).

11.

G. Giuliano and K. A. Small, Is the journey to work explained structure? Urban Studies, 30, 1485-1500 (1993).

12.

Po Gordont st:cucture

13o

P. Gordon, A. Kumar and Ho Richardson, Gender differences travel behavior, Regional Studies, 23, 499-510 (1989b).

].4 ¯

P. Gordon, Ao Kumar and S. Richardson, The spatial new evidence, Urban Studies, 26, 315-326 (1989c).

1!5¯

P. Gordon, A. Kumar and H. Richardson, The commuting paradox -- Evidence from the top 20, Journal of the American Planning Association, 57, 416-420 (1991).

1 (5.

B. Hamilton, (1982).

1’7 ®

M. A. Hughes and Jo Madden, Residential segregation and the economic status black workers -- New evidence for an old debate, Journal of Urban Economics, 29, 28-49 (1991).

18.

K. lhlanfeldt and D. Sjoquist, Job accessibility and racial differences youth employment rates, American Economic Review, 80, 267-276 (1990).

region,

Regional

by urban

A. Kumar and H. Richardson, The influence of metropolitan on conlnuting time, Journal of Urban Economics, 26, 138-151

Wasteful

commuting,

Journal

mismatch

of Political

spatial (1989a).

in metropolitan

hypothesis

Economy,

-- Some

90, 1035-1053

in

of

16 19.

K. Ihlanfeldt and D. Sjo~dist, occupations of black and white 21, 295-315 (199!a)

The role of space in determining the workers, Regional Science and Urban Economics,

20.

K. lhlanfeldt and D. Sjoquist, employment -- A cross-sectional

The effect of job access on black and white analysis, Urban Studies, 28, 255-265 (1991b).

21.

Ko lhlanfeldt and M Young, The spatial distribution of black employment between the central city and the suburbs, Working Paper, Georgia State University and Converse College (1993).

22°

J. F. Kain, Housing segregation, Negro employmentt and metropolitan decentralization, Quarterly Journal of Economics, 82, 175-198 (1968).

23.

J. Madden and C. Lic, The wage effects constraints on employed married women,

24.

J. Madden and M. White, Spatial implications force, Land Economics, 56, 432-446 (1980).

25.

Eo Mills, An aggregative model of resource allocation American Economic Review, 57, 197-210 (1967).

26.

R. F. Muth, "Cities

27.

P. Nijkamp, residential (1993).

28.

R. Price and E. Mills, Race and residence of Urban Economics, 17, 1-18 (1985).

29.

H. Richardson, On the possibility Economics, 4, 60-68 (1977).

30°

W. Simpson, ~Urban Structure and the Labour Market: Worker Mobility, Commuting, and Underemployment in Cities, ~ Clarendon Press, Oxford (1992).

31.

K. A. Small and S. Song, ’Wasteful’ commuting: Political Economy, 100, 888-898 (1992).

32.

B. D. Taylor and P. M. Ong, Racial and ethnic variations in employment access: An examination of residential location and c~,-~uting in metropolitan areas, University of California Transportation Center working Paper No. 175 (1993)o

33.

B. von Rabenau, (1976).

34°

M. Wachs, B. Taylor, N. Levine and Po Ong, The changing cor~nute: A case-study the jobs-housing balance over time, Urban Studies, 30, 1711-1729 (1993).

35°

M. J. White, Firm suburbanization Economics, 3, 323-343 (1976)o

36.

M. J. White, Urban commuting journeys Economy, 96, 1097-1110 (1988).

37.

K. F. Wieland, An extension of the monocentric urban to a multicenter setting: The case of the two-center Economics, 21, 259-271 (1987)o

38.

J. Yinger, City and suburb: urban models with more than one employment center, Journal of Urban Economics, 31, 181-205 (1992).

and Housing,"

of residential Urban Studies,

location and commuting 27, 353-369 (1990).

of increases

University

in the female

labor

in a metropolitan

of Chicago

Press,

Chicago

area,

(1969).

P., L. van Wissen and A. Rima, A household life cycle model for relocation behavior, Socio-Economic Planning Sciences, 27, 35-53

Optimal

growth

in earnings

of positive

of a factory

and urban

determination,

rent gradients,

A resolution,

town,

Journal

subcenters,

are not ’wasteful’,

Journal

Journal

Journal

of Urban

Journal

of Urban

of

Economics,

of Urban

Journal

of Political

spatial equilibrium model city, Journal of Urban

3

oi

17

39.

J. S. Zax, Race and commutes,

Journal

40.

J. S. Zax, When is a move a migration? 24, 341-360 (1994).

41.

J. S. Zax and J. F. Kain, Commutes, Economics, 29, 153-165 (1991).

of Urban Regional

quits,

Economics, Science

and moves~

28, 336-348

and Urban

Journal

(1990).

Economics,

of Urban

18

Figure I: Transportation Cost and Bid Rent Gradients for Downtown Workers in a Linear City with Employment Centers at Downtown (D) and the Suburbs (za)

Market r(z) with nonstochastic employment sites Downtown (D) and

~

D uburbs~/~~~ ~s

D

zs Figure la

Bid rent gradient for current Downtown workers with a 20 percent expectation of Suburban employment in the next period. (~ = 0.2)

D Figure ib

zs

Bid rent gradient for current Downtown workers with an 80 percent expectation of Suburban employment in the next period. (~ = 0.8)

D s

z Figure Ic

19 Figure 2: An Example of when Stable Jobs Result in Shorter Commutes Bid rent of Downtown worker with high job stability (6 = 0)

~/

Bid rent of Downtown worker with medium job stability (6 = 0.2)

Bid rent of Downtown

~/

k~.P ~~

/ worker with i~ job ~ stability (6 ~ 0 8)

D

zs

Figure 2a: Bid Rents of 3 types of Current

Downtown

Workers

Bid rent of Suburban worker with high job stability (~ = I)

$

stability(~-0S)

~ ~/Bia rent of Suburban

z z s

Figure 2b: Bid Rents of 3 types of Current Suburban Workers

$

Market |

Rent

KEY to Location Codes

,

I: Downtown workers Downtown workers with

~ If:

I/I

~

with stable jobs. medium job stability.

: Downtown v~x: Suburban workers workers with with unstable unstable jobs. jobs. Suburban workers with medium job stability. VII: Unsorted Suburban workers

vlwrw \ Z

z

s

Figure 2c" Resulting Market Rents and Locations Corresponding to Parameters in Figures 2a and 2b.